Slashdot Mirror

← Back to Stories (view on slashdot.org)

UPDATED: Transmeta's Crusoe Unveiled

Posted by ryuzaki0 on from the rolling-it-out dept.

I've gotten the first round of details about Transmeta's *two* new chips (Thanks Chris!). It's very cool - x86 compatible, Linus has written "Mobile Linux" to run on the chip, and totally insane power consumption. Click below for details - and we'll be updating this story throughout the day so check back again for more. Update: 01/20 02:33 by E : David Cassel, who was at the unveiling, sent in his notes and some great quotes from the unveiling. His take is appended to the end of this article.

There's two chips:
TM3120

  • Scales to 400 mhz
  • .22 micron process
  • 73 die-type
  • Released: Now
  • $65-89

PM5400

  • Scales up to 700 mhz
  • .18 micron process
  • 73 die type
  • Released: Mid-2000
  • Projected Pricing: $119-329

The chips themselves are 128-bit chips, and are aimed at the mobile market, as TM has said before. One of the incredible parts is their power consumption: 20 milliwatts of power in deep sleep, and 1 watt of power in regular usage. They've written their own BIOS, with Power Management on the chip called "LongRun". The chip actually gauges how much of the processing power that is needed and adjusts the power accordingly, meaning a much longer battery life.

The thermal difference is cool, too - the Pentium III is 113 Celsius, while Crusoe runs at 48 Celsius. That means no fans needed, another power saving move. And my lap won't be as warm. They're aiming this at everything from cell phones to laptops. At this time, they've said that samples have been shipped to "leading notebook vendors" but have declined to name them. As they've said before, IBM is making the chips for them.

What Linus has been doing: He's been writing a version of Linux called "Mobile Linux". It's written into the ROM and you can use the machine through a touchpad screen. The IDE and everything will be released to the Community. Yes, Linux has gone even more mobile. Oh - and Dave Taylor & Linus played Quake to demo it. Linus lost.

The x86 emulation is done at the hardware level - although emulation is the wrong word. We'll have more information on that as well.

We'll be updating this story - the press conference is still going on, but I figured people would want to know. This looks amazing. Check out ZDNet's tech coverage.

UPDATE by David Cassel

What I Saw At the Revolution

Transmeta rented an auditorium on an estate 20 minutes from their headquarters -- and everyone was excited. Walking through the rain -- past the huge lawn, the PacSat satellite uplink, and guys in suits talking into cellphones -- was Transmeta manager Rob Bedichek, who worked on Crusoe's dynamic translator. I asked him how he liked working at Transmeta, and he told me "The first couple of years," I'd wake up and I'd go, 'I have the most fun job in Silicon Valley." On the way into the auditorium I asked him about about the company ("The people I work with are amazing: people whose work I'd read about as a grad student.") and Linus. ("Great guy. Very capable.")

Transmeta had packed the press into an auditorium known as the "carriage house" -- I saw a dozen TV cameras, and I'd guess 150 reporters. A big screen filled part of the wall by the stage, flashing a fast montage of pictures (circuit boards, people's faces) over cheesy jazz music. But when Transmeta CEO David Ditzel took the stage at 9:05, there was a dead silence. "I know some of you have been waiting a while to hear about what we've been doing," he said to play up the tension, prompting a few laughs. "Some of you have been waiting four and a half years..."

Ditzel ran through his Power Point Presentation. (1995. "Something was fundamentally wrong with processors...") and pointed out that the people looking for solutions had been the entrenched semi-conductor companies. Then he announced, of course, Transmeta's "combined hardware/software solution.... The first microprocessor re-thought explicitly for the problems of mobile computing." By now everyone knows that it retains x86 compatibility while allowing a a completely new chip architecture. Ditzel remembered that when he was recruiting for Transmeta, after sharing his plans he'd hear, "If you start this company, I'll quit my job and come join you to do this.

Ditzel ticked off the specs, using phrases like "dynamic translation" and "software-optimized execution," and pointing out that only one-quarter of the functionality would be on the Crusoe chip itself. And there were frequent mentions of the mobile Linux operating system. (More about that later.) Wednesday's announcement was just the first two chips in the Crusoe family: the TM3120 (with 400 Mhz, 108 KB of cache, using 1 watt of power) and the TM5400 (700 Mhz, 400 Kb cache, and 1 watt of power.) Towards the end, Ditzel demonstrated a WebPad -- running Linux -- and pointed out that today's notebooks still use chips designed for servers and desktops. Then he staked his claim. "If it has a battery and a Web browser, it's going to be built with Crusoe."

Ditzel had to stress details for business reporters -- "significant staff" in Taiwan and Japan and "a very strong partnership with IBM -- and later Doug Laird, Transmeta's VP of Product Development, described IBM as "Great guys" and added, "We are in production right now." But I liked one of Ditzel's last comments: "Our goal is to fundamentally change the rules."

Doug Laird was more intense, arguing with current benchmarks ("Today's benchmarks address performance and battery life separately") and promising to show "what is fundamentally different here... Where's the beef." Using a red laser pointer, he ran through taped footage of a system running MS Word 2000 and Excel 2000 on a system with a Crusoe chip, "translating on the fly, as you're running the programs." Then he displayed thermal images comparing processors. (Sensing a photo-op, the cameras started flashing when he held up two "thermal solutions" and started talking about fans...) Laird made his point by showing a Crusoe system using less than 2 watts of power while playing a DVD and pointing out that it can adjust from frame to frame. (The audience laughed at the PowerPoint movie that showed two laptops playing a DVD. Two thermometers showed the temperature rising; then the laptop on the right started smoking...)

Then to break things up, there was the historic Quake showdown between Quake co-creator David Taylor and Linus. "I can't think of anybody better on the face of the planet to demonstrate Crusoe on Linux than Linus Torvalds," Laird joked. The photographers rushed towards the stage again for the even-more-obvious photo-op as Linus came out in his denim shirt, jeans, and sandals. ("I'd like to point out that if I lose, it's not the operating system," Linus joked.) It all ended when Linus fired all his bullets in a spray, then got nailed when he ran out of ammo. (Later in the press conference, after a bunch of questions about his role and Transmeta, Linus referred back to the Quake game, saying it was "meant to show that I'm here, but I'm not supposed to be the main point of it all.") One of Transmeta's technical staffers told me at lunch that "We all knew Linus was gonna get his ass kicked," and sure enough, when I asked Dave Taylor what he thought of Linus's Quake-playing, he said "I thought he sucked." But then he added modestly "I suck at code compared to him. So that felt good."

After the Quake match, the scripted presentation ended and the open press conference began. Linus had worked on the code morphing, but he wasn't one of the execs in this first round of questions. Still, he was clearly on people's minds. Almost immediately a reporter asked what Linus's role was at Transmeta, and then Boardwatch's Thom Stark drew a laugh when he asked when Transmeta would open source the code morphing software. (Since it's considered part of the chip's intellectual property, they probably won't.) And Mark from Linux Journal asked why everything had been so tightly guarded, arguing "There's no demand for secrecy."

Ditzel's answer was that he'd learned the difference between hype and buzz. ("Buzz is when you're quiet and someone else talks about you.")

Rob Bedichek told me later they were proud to have not made promises until they had something to show -- and David agreed. "You've heard what we have here. Today." Right before lunch, Rob remembered that it had been like working on the Manhattan Project. "You don't talk."

The audience wasn't easy. Two back-to-back questions raised the issue of benchmarks (which are answered extensively on Transmeta's Web site) and PC Week asked where their OEM's were. But Ditzel did a good job fielding the questions. He stressed that this announcement had intentionally left out OEM's, to focus attention on the chip itself -- and VP of Marketing Jim Chapman joked that anyways, "I don't think 'contract' is a germane word in the PC industry."

In fact, Ditzel was really building up momentum. I asked him what had happened in early 1998 -- when he was quoted as saying "We had a major change in direction a few months ago, and that has slowed us down a bit." His immediate answer drew applause, and probably the biggest laugh of the morning. "That was just something to throw off reporters.

I'm not sure if he was referring to the same period, but when Linus came on later he mentioned that the first chip didn't perform as well as they'd hoped. But thanks to the code morphing software, "one of the advantages is being able to change the way the chip works..." After some early bugs, "We were able to tell our translation software: Don't do that." He pointed out the chip could easily handle something like the Pentium's famous long-division bug. "Maybe we will have a bug -- but at least we can fix it."

Anyway, at this point, Ditzel was building up so much momentum that the next question was just, "Ask the President to say something." (Mark Allen had been introduced as the new president and CEO for Transmeta, hired just two weeks earlier.) There was a laugh when Ditzel aced the question about expected chip volume. (Was it hundreds of thousands or millions? "Yes.") Chris DiBona asked about the size of the marketing and sales organization (25 people) and as things were wrapping up, someone asked the obvious question about running Windows: does Crusoe *improve* the stability? Ditzel's answer? "If you get a blue screen on another chip, we'll reproduce that faithfully."

Later they brought out Linus, Bill Roses from the code morphing division, Doug Laird again, and three other technicians for the "Engineering Press Conference" -- but during the break, I talked to Rob Bedichek some more. "I'm totally pumped, totally pumped," he said. "This is a big mountain to climb." So how did they do it? "With an unbelieveable team. And an unbelieveable amount of money." (I said I'd heard $100 million, and he said "Well north of that.") Reporters were everywhere -- mulling in clusters out of the rain.

"What do you think of this stuff?" I heard one ask another.

"I think they fixated on a market that's not being well-served."

One of the first questions in the Engineering Press Conference was for Linus, about the mobile Linux operating system that kept coming up in the presentation. It's a "small distribution to give to OEM's so they could have something to run with....not a Transmeta Linux, but more of a vehicle for supporting OEM's." (Rob told me later, "We recompiled Linux for our machine. There's no advantage!") Later Linus added that "It looks a lot better this week than it did last week," and that it "needs some work..." ("Like the chip, we're not releasing anything until it's ready.") Naturally, he specified that it will be Open Source. Someone asked him if his Transmeta job would affect kernel development. "My interests have always affected kernel development," he pointed out. "That's not gonna change."

Linus also talked about how much he liked mobile computing, saying he loves his Gateway but that it takes forever to boot. When asked about how he'd decided to come to work for Transmeta, he described the presentation Transmeta had given him. "I went back to the hotel room and I thought, 'These people are crazy.' And that was a positive reaction. Despite the simulations they showed him "at glacial speed," Linus wanted to work for "a company that does something for and something interesting."

So what were the other job offers that he'd had? Linux companies, of course, Linus answered, but "I didn't want to polarize the Linux market." And Transmeta is a good solution. "We were a chip company where Linux is part of a much larger strategy."

Then he asked the reporters, "Do you have questions for someone else?" (No real surprises; except Bill Roses conceding that Mac compatibility was "theoretically possible.")

When it was over, reporters milled around for the free lunch or crowded into the next building to play with the demo equipment. Basically it was boxes showing the Crusoe chip's ability to run existing software. (There was a Windows 2000 box running Office 2000, next to a Linux box running Quake) and some blue laptops in front of cards that said things like "Ultralight Mobile". But towards the end Transmeta VP of Software Engineering Colin Hunter did show me a neat WebPad using Transmeta boards and software and IDEO mechanicals which let you plug-in attachments for games and cameras.

And with that, as the press release said, "Transmeta breaks the silence."

768 comments

  1. This is unbelievably cool. by Anonymous Coward · · Score: 1

    Actually, cool doesn't quite describe it. Viva Linux!

    1. Re:This is unbelievably cool. by Anonymous Coward · · Score: 0

      actually, cool quite describe it. 48 celcius.. whoaa.. :)

    2. Re:This is unbelievably cool. by Ventilator · · Score: 1

      In that case, the crusoe is definitely no option for me. You can't boil coffee at 48C... =:-)

      --
      --- If OS were buildings, then the first woodpecker to come around would erase 95 % of civilization.
    3. Re:This is unbelievably cool. by Anonymous Coward · · Score: 0

      Are you nuts? 48 degres celsius is killing hot!

  2. Emulators by Hoonis · · Score: 4

    I'm wondering about emulator programming.. Linus said something about "emulators on steroids". From the various comments, can anyone tell if the processor instruction can be dynamic, done in user space? ie can I pop open a MacOS/ppc vm and have it get the cpu instructions while I run another host os a-la VMWare?

    1. Re:Emulators by taniwha · · Score: 1
      I bet there are two flavors of emulators - one that does an instruction by instruction emulation of an ISA and one that recompiles into the native ISA on the fly (one of the patents last year pointed to how they would handle the fall-back from the 2nd to the first in the case of exceptions)

      Writing an emulator won't be easy - writing a well performing emulator (suite) will be very hard.

    2. Re:Emulators by Gurlia · · Score: 3

      Hmm, this raises an interesting thought: Would Crusoe eventually replace other architectures? Since it can simply run different Code Morphing software to emulate every existing architecture out there, why can't we code directly for Crusoe's native instruction set?? We don't need other architectures at all...

      I understand that one advantage of *not* coding directly in the native instruction set is that Transmeta can totally revamp the instruction set and simply release a new Code Morphing software... but still, while a particular release lasts, why not take advantage of it?

      Taking this further, Transmeta *could* release a "static" external instruction set that the Code Morphing software translates into whatever the current native set is. Then we don't need other architectures *at all*. All we need is to use this static external instruction set. We don't even have to worry about being compatible with future releases of Transmeta, since the Code Morphing software takes care of that.

      To conclude... WoW! I think Transmeta could be hitting something real big here... congrats!

      --
      mikre he sophia he tou Mikrosophou.
    3. Re:Emulators by Anonymous Coward · · Score: 1

      Is there any relation between Crusoe emulation techniques and E2K proposed chip ?

    4. Re:Emulators by pnagel · · Score: 5
      Crazy as it may sound, even if you do code straight to the native Crusoe instruction set, you would still need a Crusoe-to-Crusoe "code morphing" layer to get full performance.

      Remember, their chips' have no out of order execution units; they do this all in software. Instead of having lots of silicon to do agressive instruction scheduling optimizations on each instruction every time round a loop you re-execute the same old instruction again, their "code morphing" layer gets to lazily decide when to put in more effort into instruction recoding as it becomes obvious that a section of code need it.

      And the beauty of it all is that these instruction translations are saved in memory - you get to preserve a lot more state, you get to save instruction sheduling decisions, whereas silicon has to always repeatedly make those decisions over and over again as it reexecutes the same instruction.

    5. Re:Emulators by David+Greene · · Score: 1
      You could get exactly what you want if Transmeta released firmware for running its native instruction set on the CPU. The "code morphing" would simply be a direct issue to the processor core. However, as pnagel explained so well in another response to your message (Moderators, up the score on that one!), the code morphing does more than translate. It optimizes. A lot. It's very important for performance reasons that this happens. If Transmeta ever releases an O-O-O CPU, then this effect is somewhat less important and a direct issue would perform almost as well.

      --

      --

    6. Re:Emulators by David+Greene · · Score: 1
      You could get exactly what you want if Transmeta released firmware for running its native instruction set on the CPU. The "code morphing" would simply be a direct issue to the processor core.

      However, as pnagel explained so well in another response to your message (Moderators, up the score on that one!), the code morphing does more than translate. It optimizes. A lot. It's very important for performance reasons that this happens. If Transmeta ever releases an O-O-O CPU, then this effect is somewhat less important and a direct issue would perform almost as well.

      --

      --

    7. Re:Emulators by pod · · Score: 1
      As the tech brief section of the unveiling revealed, the cpu can receive updates of the opcodes it uses along with a code morpher engine update to generate them. So essentially, at any given point in time a given crusoe cpu can be running an instruction set that is different from the one sitting next to it.

      In effect there are no 'native' crusoe instructions. Indeed, tm is reluctant to even release the current instructions for fear that major developers will code to them and then force backward compatibility on tm!

      The idea behind crusoe is apprently to dissociate the software from the hardware by having developers write to a known target (x86, sparc, java bytecode, alpha, whatever) and have the code morphing unit of the cpu (or 'bios') convert that to whatever happens to be the native instruction set at the time.

      --
      "Hot lesbian witches! It's fucking genius!"
    8. Re:Emulators by Anonymous Coward · · Score: 1
      If you read their website and their PDF files that describe more details, they purposely don't want to have people write to their VLIW instruction set.

      They plan to change the instruction set from time to time, and they already have. Their faster processor they are working on has an incompatible instruction set from their original one. They keep the abstraction on purpose.

      Also, as for x86 binaries going away anytime soon, I don't see it happening. I think we're as likely to see HTML disappear from the internet about as fast, despite all IA-64 and so on stuff.

      - Tom

    9. Re:Emulators by Yakk · · Score: 1

      > [snip: why not code for crusoe directly]
      >
      > I understand that one advantage of *not* coding
      > directly in the native instruction set is that
      > Transmeta can totally revamp the instruction set
      > and simply release a new Code Morphing
      > software... but still, while a particular
      > release lasts, why not take advantage of
      > it?

      Of course in the Open Source arena, binary compatibility isn't such an important issue. We just need to make sure gcc's backend is ported to the new Transmeta architecture and we can simply recompile all our code :)

      Ian

    10. Re:Emulators by Anonymous Coward · · Score: 0

      pnagel's post may very well get moderated up, but only by clueless moderators. VLIW chips preset the OOO in software, true. but doing it in a morphing layer would be amazingly stupid. Intel's VLIW design depends on the compiler to do the ordering. Its the fast way to do it, since you compile once, whereas you translate every time the code gets pushed out of the "translation cache."

      Don't believe the hype. Wait till we get samples, then draw conclusions. Basing anything on a marketing extravaganza is just plain stupid.

    11. Re:Emulators by Anonymous Coward · · Score: 0
      Of course in the Open Source arena, binary compatibility isn't such an important issue.

      I think you mean "Open Source playpen".

    12. Re:Emulators by Anonymous Coward · · Score: 0

      I heard the question answered in the presentation. The hardware instruction set is different for each model of chip, therefore, no it is not program visible nor is it intended to be.

    13. Re:Emulators by David+Greene · · Score: 1
      If you read any of the recent papers about FX!32 and Dynamo, you'll see some interesting results. The code caches can be made to work quite effectively.

      The problem with static compiler scheduling is that it doesn't have any run-time context. The translation layer provides this context, in effect allowing a compiler to optimize your code as it runs, which gives the optimizer more exact information. It's like an O-O-O machine, only more powerful since software can generally "see" further ahead than hardware. Because the generated code is re-used, the software translator can take a bit more time to do a good job.

      --

      --

    14. Re:Emulators by NMSpaz · · Score: 1

      This sounds very much like Sun's argument that HotSpot could potentially be faster than native code because the code isn't constrained to being kept static, but the common cases could be optimized by profiling the execution paths. Does anybody more knowledgable than myself know if the similarities extend beyond the superficial? I thought Sun's line had always been that such dynamic optimizations weren't really possible outside of Java because the pointer arithemetic screws everything up. Perhaps such cases might simply not be optimized.

      --Jared in WA
      PS. pnagel: Any relation to the Platinum Puma?

    15. Re:Emulators by Sangui5 · · Score: 1

      >>In effect there are no 'native' crusoe instructions. Indeed, tm is reluctant to even release the current instructions for fear that major developers will code to them and then force backward compatibility on tm!

      Not really. They can just have the code morpher for their later chips translate the old crusoe intructions into the new, updated one.

      Really, they way they have it set up, they never have to worry about a compatability problem, backwards or otherwise. They can run any instruction set, including their own old one. They just have to be able to detect which instruction set it ought to be translating from.

  3. Well the wait is over ..... by joel_archer · · Score: 0

    Is that all there is?

  4. Interesting. This should advance mobile computing by bago · · Score: 1

    I have always wanted a personal secretary that would be powerful enough to run a full suite of financial and office apps on the go, with voice/pen/keyboard inputs and wireless voice/net/ir connectivity. This should help speed that up.

    --
    .
  5. x86 compatible? by crush · · Score: 1

    It's the code-morphing that makes it compatible right? Or is there some deeper compatibility at the hardware level? Is there any reason to believe that applications written natively for this would be able to avoid the code-morphing layer and run even faster?

    1. Re:x86 compatible? by tve · · Score: 2

      The cool thing about it is that it uses software only to translate x86-instructions to it's native instructionset. It then stores this translations in a translationcache to avoid retranslating everything over and over. This is the code-morphing you are talking about.

      But the translationprocess doesn't simply convert x86 to native instructions. It also optimizes the instructions. This can sometimes reduce de number of instructions by 50%!

      As I understand it this means two things:

      1) It should be possible for very fast software to use it's native instructionset.

      2.) It isn't as easy to addept for other architectures then the x86 as writing a new translationpart into the code-morphing layer, because the whole optimizationroutine has to be rewritten too.

      --

      If there is hope, it lies in the trolls.
    2. Re:x86 compatible? by Wraithlyn · · Score: 1

      The code morphing software is the only software that is actually compiled specifically for the hardware processor. The code morphing software, however, was designed side by side with the hardware, so in effect the code translation is "hardware accelerated" for what they described as the more frequent x86 instructions.

      --
      "Mind, as manifested by the capacity to make choices, is to some extent present in every electron." -Freeman Dyson
    3. Re:x86 compatible? by Foogle · · Score: 1
      Yes, it's the code morphing that makes it compatible. While, in theory, native-instructions would run faster, the statements they made on that issue suggested that the performance increase would be marginal at best. Apparently the Crusoe is smart enough to do a real bang-up job of the "emulation".

      On top of that, I think they said they would hold off on releasing the specs to write native code. One of the reporters asked about "Open Sourcing" the instructions and they said it wasn't likely any time soon.

      -----------

      "You can't shake the Devil's hand and say you're only kidding."

    4. Re:x86 compatible? by coredog · · Score: 1

      Been there. Done that. Got the CD.

      FX!32 and it's Linux equivalent have been available for quite some time.

      The optimization is nice, but how many times can the x86 emulation wheel be re-invented (FX!32, K7, Crusoe)?

      --
      Do anal-retentive people hyphenate 'anal retentive'?
    5. Re:x86 compatible? by Anonymous Coward · · Score: 0
      Just a hunch...

      Suppose you decide to run two fairly complex applications on a Crusoe processor. By complex, I mean different portions of the code are very dissimilar and thus get translated differently by Transmeta's morphing software. In this case, the translation cache may not be big enough to contain all the translated instructions. In effect, you would have lots of cache misses and thus the x86 code would run very slow.

    6. Re:x86 compatible? by Inoshiro · · Score: 5

      "It's the code-morphing that makes it compatible right? Or is there some deeper compatibility at the hardware level?"

      Unlike the PIII or K7, which have special microhardware to translate 32-bit x86 ISA to the 32-bit RISC internal cores, the Crusoe uses a hybrid hardware/software module (patenteded, et all) that translates the 32-bit x86 ISA into 128-bit VLIW core. The fact it's in software allows them to apply RAD techniques to their translation code, which is a big win, allows them to reconfigure the hardware based on the programs running for optimal power consumption, and allows software overclocking :-) This is the reprogrammible CPU we've all secretly lusted after... And because its core is VLIW, I hope we can leave behind those "[R|C]ISC is better" flamewars (which were really pathetic). The VLIW core is also very much simpler in design and cleaner (asthetically speaking) than either CISC or RISC. This is why the chip can be so cheap for the speeds they get, why it's so cool (pun), and why it uses so little power (more so inconjuntion with their translator).

      "Is there any reason to believe that applications written natively for this would be able to avoid the code-morphing layer and run even faster?"

      Nope. Unlike the K7 and PIII, the translation is important to the process. For performance, they likely have the translator units running at a fast enough speed that the VLIW core is kept as full as if it didn't have a translator unit. And without the translator unit, you'd have to spend a few more man years designing a new BIOS, chipset, etc, that understand VLIW, as well as a new instruction fetching unit. It's easier to support the x86 ISA (which everyone supports), and stick with the design they have now. Besides, as they have pointed out, the purpose is to have a reliable low power CPU, not an "oh my god, I came it was so fast!" processor. This is best accomplished with a smart translator that is software reconfigurable around a simple VLIW core.

      This doesn't stop the idea of a very high performance VLIW core desktop machine, which is what Intel is lusting after with its Merced. Luckily, the Cursoe seems a lower-level version of the Merced, which should stop any Intel strangehold on the VLIW market. And when AMD extends the x86 ISA with 64-bit instructions, your Cursusoe from the ol' year 2000 will be able to handle it. Flexbile; extensible; cheap -- I like it.
      ---

      --
      --
      Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
    7. Re:x86 compatible? by rent · · Score: 2

      A translation cache is still better then nothing. If you think of a cache like a window on a portion of the code, and then you can imagine that only a small portion of the code (less than 1K) would be kept in the cache. The TM3120 has 108KB of cache, but the TM5400 has 400KB cache, which sounds like a good amount.

      Cache works because it exploits the locality principle. The locality principle says that the 2nd memory reference that comes in sequence after the 1st memory reference will most likely be close to the 1st reference. This means that addresses are grouped in clusters, instead of being in stored in random locations.

      Cache memory is many times faster then main memory, therefore the CPU can access the cache memory much faster. If we know that memory references are grouped in clusters, then it makes sense to copy a whole cluster of memory into the cache all at one go. The CPU accesses the cache at one address at a time, and because of the locality principle, the CPU will always get more hits then misses.

      If the cache ever runs out of space, the least used cluster is evicted from the cache - this guarantees that the most used clusters are always kept in cache memory the longest.

      I also believe that the locality principle can be applied even better to code, since code tends to occur more naturally in groups of clusters (eg functions) and better predictions can be made on how frequently the clusters will be needed (eg inner loops)

      I do not think the Crusoe chips will have any problems with cache..

    8. Re:x86 compatible? by Paolo · · Score: 1

      OK, here's two pessemistic questions I have--

      What about Merced? Although Transmeta has a 128bit chip, they still emulate the x86 instructions, and are completely reliant on Intel setting standards. Suppose Intel doesn't want to let Transmeta in on a prerelease stepping of the Merced chip? Are they dooomed?

      Secondly, I realize this is an interesting turn in chip development, as it is a step away from bragging about megahertz. Was 4 1/2 years too long to wait?
      Fud for thot

      --
      "In individuals, insanity is rare, but in groups, parties, nations, and epochs it is the rule." -Nietzsche
    9. Re:x86 compatible? by Inoshiro · · Score: 2

      "Although Transmeta has a 128bit chip, they still emulate the x86 instructions, and are completely reliant on Intel setting standards."

      I addressed this in another thread. So what if Intel decides night is day and white is black? It will only affect their instruction sets. If Intel goes and reverses the order of the argument bytes to the mov instruction, all they hurt are themselves. AMD and Transmeta both carry the x86 ISA torch along side Intel. They'd have to resort to another method to beat them, which leads ...

      "Suppose Intel doesn't want to let Transmeta in on a prerelease stepping of the Merced chip?"

      Intel has said before that the Merced is a 64-bit VLIW chip with a "powerful" core. In fact, they set the pricing point to be around 5 to 6 figures. I really don't think that'll clash with a low-end designed for embeded marlet processor, which uses a different simplified 128-bit VLIW core. The ol' x86 ISA doesn't get a look in anymore :-) Besides, this has never posed a problem for AMD.. Being first to market doesn't always mean being best to market (K7 vs PIII).

      ---

      --
      --
      Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
    10. Re:x86 compatible? by infodragon · · Score: 1

      To quote the article...

      "Then he asked the reporters, "Do you have questions for someone else?" (No real surprises; except Bill Roses conceding that Mac compatibility was "theoretically possible.") "

      If the Mac compatibility is "theoretically possible." a "Java Chip" could be part of the morphing code giving you awsome Java performance and cutting out the cumbersom Java Virtual Machine. Basically Java enabling EVERYTING.



      --
      If at first you don't succeed, skydiving is not for you.
    11. Re:x86 compatible? by rew · · Score: 1

      If you think of a cache like a window on a portion of the code, and then you can imagine that only a small portion of the code (less than 1K) would be kept in the cache. The TM3120 has 108KB of cache, but the TM5400 has 400KB cache, which sounds like a good amount.

      Crusoe uses about seven to fifteen Megabytes of code-cache. The 1M codemorpher is copied from ROM to sdram, and then the system uses 8 to 16Mb of
      system ram which is not accessible to the x86 code.

      The word cache refers to several different things. Just like a screw can be in the dashboard of your car, on the rear wheel, or in the motor.

      Roger.

    12. Re:x86 compatible? by happydaze · · Score: 1

      Thanks for the very good explanation. It seems that Crusoe's, main non-technical focas is just that, attempting to solve a strong customer need, more functional portable computing device(s). New software innovation is exciting and because its easy to change more responsive to customer needs. I believe transmeta's work will touch off a wide range of activity (the tip of the iceberg) which will add lots of new computing and business activity to an already exciting world.

    13. Re:x86 compatible? by Anonymous Coward · · Score: 0
      The word cache refers to several different things. Just like a screw can be in the dashboard of your car, on the rear wheel, [...] in the motor
      ...or in the the back seat.
  6. Sweet. by kwsNI · · Score: 1

    Cool. I like the idea of smaller, more portable Linux systems. Just think - the idea of a Mobile Linux PDA. Say goodbye to Windows CE :)

    kwsNI

  7. IBM & Transmeta by Anonymous Coward · · Score: 0

    I have this feeling that IBM figured that there is going to be a HUGE need for mobile computers in the future and thought that if they sortof made it look like this new company was coming out and had a great chip that they would make some more money. If you understand where i am coming from atleast. Don't get me wrong, I love the chip, i think it's a great idea, but it reminds me of the Cyrix chip a few years back. Well that's my 2 cents. TacoKing

    1. Re:IBM & Transmeta by Anonymous Coward · · Score: 0
      Yeah, that's probably why IBM has created the chips for the new Nintendo (Dolphin); monetary gain. And it probably has something to do with bendable LCD screens, blue laser CD burners, and GMR heads. I sure am glad that they have
      • some
      reason to become the single largest patent registrar for the last five years. Why should you care, you reap the rewards of the technology in the end.

      Save that kind of bashing for a company that is diametrically opposed to the Linux/open source community (i.e. Microsoft).
  8. Crusoe News Article Links by BeIshmael · · Score: 4
    CNET: http://news.cnet.com/news/0-1003-200-1526340.html? tag=st.ne.1002.tgif?st.ne.fd.gif.e

    MSNBC: http://www.msnbc.com/news/358517.asp

    ZDNet: http ://www.zdnet.com/zdnn/stories/news/0,4586,2423914, 00.html?chkpt=zdhpnews01

    CNN: http://www.cnn. com/2000/TECH/computing/01/19/crusoe1.idg/index.ht ml



    BTW, Arstechnica is promising an in-depth look at the CPU later today from their CPU expert.

    1. Re:Crusoe News Article Links by dewet · · Score: 1

      Thorough articles.

      Just one thing: ZDNet makes the following claim: "...deep sleep mode that uses only 20 milowatts of power". Now, I dunno whether I've been sleeping, but since when is "milo" an ISO prefix? And what does it signify, anyway?

      --
      -------------------------------------------------- -
      This sig could have been put to good use.
    2. Re:Crusoe News Article Links by Roofus · · Score: 1


      Perhaps they meant "milli"
      You never know with those marketing types.

    3. Re:Crusoe News Article Links by banasw01 · · Score: 1

      I just wanted to add another link: http://www.tomshardware.com/technews/technews-2000 0118.html#0101 and oh yeah... Viva Open Source/Free Software/Linux/Other OS than Windows!!!!!!!!!! Neo: I'm gonna show you all a world without you, Bill, a world without control, without boundaries, where everyone is free to REALLY innovate! Windows = The Matrix, what a prove? Go to: http://condor.stcloudstate.edu/~banasw01/matriks.h tml (low bandwith version)

      --
      This space intentionally left blank.
    4. Re:Crusoe News Article Links by Sloppy · · Score: 2

      And what does it signify, anyway?

      It signifies ZD's committment to journalistic excellence, of course.


      ---
      --
      As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
    5. Re:Crusoe News Article Links by Jonavin · · Score: 1

      From reading all these CNET/ZDNET/MSNBC/CNN articles, you'd think that it's just another x86 clone. I think "Mobile Internet" is just a marketing grab-bag of buzz word. Oh and "Mobile Linux Web Pad"; wow bonus points for getting 'web', 'internet', and 'linux' in there.

      I seriously hope that once they get some recognition they'll get beyond this markdroid stuff and do something really interesting with Crusoe. Releasing Java, DragonBall Z (PalmPilot), and other code-morphing layers would be a start. As just-anther-intel-clone, they're gonna go nowhere.

      ObStockMarket: IBM, and AMD were up a few of bucks, and Intel was down a couple today. Coincident?

      Crusoe: "Where do you want to be stranded today?"

    6. Re:Crusoe News Article Links by Tomahawk · · Score: 1

      Crusoe: What do you want to be standard today?

  9. obligatory AC post by toast0 · · Score: 0

    lets make a beowulf cluster out of a bunch of these overclocked and smp'ed into a wearable computer running linux



    --
    Need a Catering Connection
    1. Re:obligatory AC post by Wah · · Score: 0

      ...now _that's_ an obligatory AC post.

      --
      +&x
    2. Re:obligatory AC post by Anonymous Coward · · Score: 0
      Heh. Suppose I were handed 1000 Pentium III systems for a Beowulf cluster, but were told that I had to pay for upkeep. The electric bill alone would bankrupt me! Actually, the bill for a new power hookup that could handle these would bankrupt me. Don't forget the giant air conditioner.

      On the other hand, 1000 of these chips is just a hairdryer. Of course, you got power consumption from support electronics, but it's looking a whole lot better. Maybe no worse than leaving your oven on all the time.

  10. bout time by gh0ul · · Score: 0

    bout time they released the specs, damn these sound cool.

  11. I want one for the desktop by Anonymous Coward · · Score: 0

    I want one of these babies on my desk...very cool presentation. Intel should be scared - Athlon, Crusoe, problems with PIII's Kewl!

    1. Re:I want one for the desktop by Anonymous Coward · · Score: 0

      Huh? Why on earth would you want it on the desktop? It will run slower than a native x86 processor. The advantage is the power consumption.

  12. Hrmm... by c.r.o.c.o · · Score: 3

    I just wonder how successful this chip is going to be. I mean nobody is denying the fact that Athlon is far superior to the P3, but in all the adds from Toronto computer magazines, you barely see any systems running them.

    Will this chip have the same hard time to enter the market?

    1. Re:Hrmm... by Anonymous Coward · · Score: 0
      I just wonder how successful this chip is going to be. I mean nobody is denying the fact that Athlon is far superior to the P3, but in all the adds from Toronto computer magazines, you barely see any systems running them.

      If the Athlon turns out to have a compatibility problem, you can't fix it without replacing the CPU. Corporates would rather pay more and get lesss than take this risk.

      If the Transmeta turns out to have a compatibility problem, it's a firmware upgrade to fix. There's actually *less* risk with Transmeta on this score than with Intel. Think about what that means to Corporates.

    2. Re:Hrmm... by sjames · · Score: 5

      Will this chip have the same hard time to enter the market?

      Probably not. The Athlon is a fine CPU, but it's advantages over PIII are not huge (reletivly speaking).

      Crusoe on the other hand isn't that fast (but really, 500Mhz equiv. isn't all that slow either), but it's power consumption is 1/35 that of PIII. That translates DIRECTLY into battery life which has been the bane of laptops from the beginning. Add in to that that a fan isn't needed and you really have something. The lack of a fan is more significant than it seems. Lack of fan means lack of vent holes (with good heatsink technique) which means a sealed case that can tolerate wet conditions MUCH better than a laptop with a fan.

      It opens the door to a new class of handheld device. The PalmVII is great (I use one myself), but compared to the Crusoe, it's CPU is absolutely anemic. So in it's niche (tiny laptops and handhelds), it really is tremendously better than the competition.

    3. Re:Hrmm... by Mad+Browser · · Score: 1

      Seriously. I'd love to be able to get PGP mail on my Palm without waiting 5 minutes to decrypt the dang message!

      --hunter

      --
      RateVegas.com - Vegas Reviews
    4. Re:Hrmm... by Score+Whore · · Score: 2

      The Pentium II and above series processors have the ability to upgrade their microcode at boot time. It requires a signed hunk of code and is only possible during a short period before the rest of the chip is initialized, but it is there and available.

      I think the PPro might also have it but I'm not sure.

    5. Re:Hrmm... by Ekapshi · · Score: 1

      "I mean nobody is denying the fact that Athlon is far superior to the P3"

      Intel says that the P3 is better than the Athlon, because "the Athlon doesn't enhance your internet experience" :-)

      -Ekapshi

    6. Re:Hrmm... by SpinyNorman · · Score: 2

      The Athlon is a fine CPU, but it's advantages over PIII are not huge (reletivly speaking).

      I guess that depends on what you mean by "relatively speaking"...

      The Crusoe certainly would appear to have a huge edge over the Pentium in terms of cost and power consumption, which will particularly give it an edge in mobile and low cost applications. It's real competition in the non-x86 arena is probably the ARM.

      However, in it's own domain - high end systems - the Athlon *does* have quite an edge over the PIII. It has 3 integer execution units compared to the PIII's 2, plus 3 floating point execution units again compared to 2 for the PIII. Just as importantly, unlike the PIII the Athlon actually has the unrestricted ability to simultaneously issue the instructions to keep those execution units busy.

      There's a full PIII vs Athlon comparison over on Ars.

    7. Re:Hrmm... by Anonymous Coward · · Score: 0

      I was impressed with their presentation, but it seemed like a lot of marketing hype. Since most /. folks probably don't have a lot of IBM background, this concept is old hat. The AS/400 has had a technology similar to this since the early 80's. This is how IBM migrated the AS/400 from a 32-bit platform to 64-bit without recompiling (moving to RISC). Granted, the target markets are different, but -- well -- until I see the thing and actually have a chance to play with it, I'm not impressed. I think it's interesting that they had IBM peer review a lot of their material through the R&D cycle. Who else to have look at something like this: They've already done it!

    8. Re:Hrmm... by sjames · · Score: 2

      However, in it's own domain - high end systems - the Athlon *does* have quite an edge over the PIII.

      Vigorously agreed. I'd like to have a few myself. The comparison at Ars cemented my opinion that AMD is to be taken seriously in the high end PC market. It also made me consider them for a Beowulf cluster (I suppose that HAD to come up eventually :-).

      The difference is that Athlon's advantages do not make the impractical become practical or create a new class of machine (though it does make PCs much better). No combination of Athlon's advantages add up to a 35 to 1 improvement.

      ARM is close to the mark. It does solve the power problems and run cool. The one thing that relegates it to a niche market (for now anyway) is that it won't run most consumer software out of the box. That's not a big deal to me, I have source for all of the software I use and it runs Linux! But it does limit mass appeal which limits my chances of finding a nice ARM based laptop I can afford.

    9. Re:Hrmm... by Zurk · · Score: 1

      nope. IBM CISC AS/400s cant run IBM RISC compiled AS/400 software. 32->64 is a helluva lot easier than converting instructions on the fly to a VLIW CPU.

    10. Re:Hrmm... by charlesnadeau · · Score: 1

      Maybe that could open the door to a dual-CPU notebook... Just dreaming...

      Charles

    11. Re:Hrmm... by Anonymous Coward · · Score: 0

      Yes.. My distructive viruses can disable both Families of CPU with trojen opcode/firmware updates!! Life is too easy.

    12. Re:Hrmm... by mindstrm · · Score: 2

      Like they said.. their market is not servers and high end workstations. Their market is PORTABLES.
      The main feature of the Crusoe process is the extremely low heat and extremely low power consumption.

      So.. if it turns out that a Crusoe can perform about as well as that Celeron 450 you wanted to use, but uses a tenth of the power and doesn't need any extra cooling.... gee.. what are you going to use (Oh.. and it's cheaper)
      This is not a hard sell for vendors.... and this is their *first* couple chips.. wait to see what they do next.

    13. Re:Hrmm... by GargoyleMT · · Score: 1

      Update? You make it sound like flashing your BIOS. AFAIK (not very far, mind you), the microcode patches are just loaded in at every boot.

    14. Re:Hrmm... by Score+Whore · · Score: 1

      That's what I understand as well. Funny you should mention flashing your BIOS, since that is what you have to do to get the updated microcode. But the lack of persistence in the CPU doesn't really change the fact that it is possible to change the microcode that the CPU uses.

      Of course it's not like upgrading your "code morpher" (did they just make that up so they could patent it?) is going to change the TMxxxx at all, just change the firmware that is used to translate foreign code.

    15. Re:Hrmm... by The+Variable+Man · · Score: 1

      How about the Psion series 7. This would seem to meet your requirements.

    16. Re:Hrmm... by BinxBolling · · Score: 1
      However, in it's own domain - high end systems - the Athlon *does* have quite an edge over the PIII. It has 3 integer execution units compared to the PIII's 2, plus 3 floating point execution units again compared to 2 for the PIII. Just as importantly, unlike the PIII the Athlon actually has the unrestricted ability to simultaneously issue the instructions to keep those execution units busy.

      The question is: Do the technical improvements you mention result in observable improvements in areas that are important to the customer? The user cares how fast applications run, not how many integer execution units his CPU has. If the Athlon's technical advantages don't translate into significant speed improvements (say, because the CPU spends so much time waiting on other system components anyways), then they aren't significant in the marketplace.

      On the other hand, Crusoe's technical advantages do have the potential to improve system performance in areas visible and important to the user: lower cost, and either greater battery life or lower battery weight. And the low temperature has a potential to lead to user-visible improvements, as well.

    17. Re:Hrmm... by mystik · · Score: 1

      If I recall coreectly, The IBM AS/400 uses a layered, object-oriented Hardware approach. This lets them put Java VMs, Intel VMs, as well as Native As/400 apps all on the same CPU, running simultanously. All VMs go throught the abstraction layer, which in turn access the the Core Hardware.

      --
      Why aren't you encrypting your e-mail?
  13. Lots of great information here by Anonymous Coward · · Score: 1
    I also notice that Linus HAS been working on Linux in his real job. So much for the myth that it was solely a labor of love for him.

    Don't flame me--I have no problem whatsoever with Linus or anyone else finding such a great match between occupation and interest. But I think it's interesting that yet another of those things we all thought we knew was true, wasn't.

    1. Re:Lots of great information here by Don+Negro · · Score: 4

      Linus really didn't have a choice when you think about it. If he'd ever said 'Yes, what I do for Transmeta is Linux-based.' It would not have been very hard to conjecture what he really was doing.

      Think about it. If it's x86 compatible, what does he need to do? Combine the knowledge that he's doing 'something' with the knowledge that it's portable/embedded/low-power, and right there you've got a pretty good picture of the market Transmeta is going for; other's could have moved to cut them off at the pass.

      So he *had* to say that his job wasn't Linux-related. To do otherwise would have been to tip Transmeta's hand.

      He did give us enough clues, though. In every interview I've read in the last 9 months, he's mentioned how interested he was in the embedded market, and how cool it would be to see Linux going in that direction.

      Don Negro

      --

      Don Negro
      Perl 6 will give you the big knob. -- Larry Wall

    2. Re:Lots of great information here by Anonymous Coward · · Score: 0

      Yea, but the problem is that everyone has been saying "Linus doesn't make any money off of Linux. Yea, he has those speaking engagements but he doesn't sell it as a product." Well, now he probably has a fat stock option package AND a salary in a company that does sell Linux as part of a product. I'm not saying that it is wrong, he's got a right to make a buck off of his creation, just that some of you guys can stop being so sanctimonious now.

    3. Re:Lots of great information here by Cramer · · Score: 2

      Actually, he did say that (sort of) in an interview some years ago -- something to the effect of "linux is in my contract."

  14. Celcius or Farenheit? by sheckard · · Score: 1

    48 C is still awful hot! I sure hope that was a typo and really meant Farenheit.

    My chip (Celeron) is running right now at 34 F, and that's plenty hot...

    I think a P3 would melt at 113 C anyways...

    1. Re:Celcius or Farenheit? by John+Fulmer · · Score: 2

      >My chip (Celeron) is running right now at 34 F

      Wow! An exothermic chip?
      :)

      Room temperature is about 70 F, and I BELIEVE that 48 C -> ~118 F. That qualifies as 'slighly warm' in silicon temperatures. A P3, on the other hand, would give you a pretty nasty burn, if you touched it with your bare hand.

      jf

    2. Re:Celcius or Farenheit? by Bartmoss · · Score: 1

      34F? That's below freezing, isn't it?

    3. Re:Celcius or Farenheit? by Ron+Harwood · · Score: 1

      48C is about the temperature of hot water straight from the water heater...

      I'm Canadian, so my knowledge of the F scale is limited... but isn't 34F close to freezing???

      I'm sure that the temperatures quoted are without heat-sinks...

      --
      BlackNova Traders
    4. Re:Celcius or Farenheit? by Anonymous Coward · · Score: 0

      I highly doubt that your chip is running at 34F, unless you're using a cryo style system, that's just over freezing. to give an example of how cool the Transmeta chip is running ambient room temperature runs between 20 - 25 Celsius depending on taste... 48C is not all that hot.

      Me

    5. Re:Celcius or Farenheit? by Kvort · · Score: 1

      I think that's the temperature of the actual processor core. Once the heat dissapates into the surrounding packaging, the whole packaging will be cooler that a celeron. And of course, the temperature of the chip will depend SOLELY on the dissapation. If there's nowhere for the heat to go, the chip will just rise in temp until it melts; doesn't matter how low power consumption, etc.

      Probably the reason why they used core temp was the numbers looked better. :) I don't know how special their chip is, but from what little I saw of the presentation, Transmeta's marketing people are EXTREMELY good.

      >>>>>>>> Kvort the Duck

      --
      -Don't mind me, I'm personality-deficient and mentally-impaired.
    6. Re:Celcius or Farenheit? by specht · · Score: 1

      > Probably the reason why they used core temp was the numbers looked better. :)

      Actually the core temperature is the only one that makes sense. Everything else depends on packaging of the chip, the cooling mechanism and the environment. So in order to come up with the
      necessary cooling strategy you have to know the
      max core temp and then go from there.

    7. Re:Celcius or Farenheit? by specht · · Score: 1

      > Probably the reason why they used core temp was the numbers looked better. :)

      Actually the core temperature is the only one that makes sense. Everything else depends on packaging of the chip, the cooling mechanism and the environment. So in order to come up with the necessary cooling strategy you have to know the max core temp and then go from there.

    8. Re:Celcius or Farenheit? by Anonymous Coward · · Score: 2
      Er, 48 degrees is around 120F. You have a celeron running at close to (h2o) freezing temps? Sorta doubtful.

      113 C is fairly trivial. Yes, water will boil, but copper and plastic melt at higher temps.

      Yeah, intel has ALWAYS run hot. The whole damn architecture does. That's why PCs usually die of poor ventilation (next likely cause of death is cheapo power supplies).

      Will this take off? Is it Better(tm)?

      Well, the StrongARM chip ROCKS. You can power a nice little home computer with a wall wart ala TRS-80's and C64s. But at 250MIPS.

      Will they win?
       No. Not unless there are things running on them that are compelling to the customer. Users don't care about Excel, or Netscape or whatnot. They want to be able to do things with their computers. If they are taught that Excel is what they need to do these things, they will use that. My Mom wants to write up a report. Doesn't matter what it is as long as it does what she needs and does it without showing itself too much.

      Star Office and Applix are close but not there yet, but they show themselves to much. People want to type things in and get them out looking pretty.

      The Palm beats Wince and Psion because (except for learning graffiti) it gets out of the way and lets people do things. Psion goes a bit further by punishing developers by needing to profit from their developments kits (hint: make it easy to write software for your hardware and more people will use it and you'll sell more hardware).

      If the Crusoe processor is underneath any of these, fine. If my phone runs (RT)Linux, fine. But if I can't make a call or get stuff done, then it's getting tossed into the river with my Wince device.

      Summary
      Heat? Not your issue. Battery life it your issue.
      You have an Intel chip with snow forming on it? Give me some of what you're smoking.

    9. Re:Celcius or Farenheit? by Anonymous Coward · · Score: 0

      >I think a P3 would melt at 113 C anyways... :) Of course if wouldn't melt. It's all metal and sillicon. 113x10 C then it could begin to melt.

    10. Re:Celcius or Farenheit? by Anonymous Coward · · Score: 0

      Your case temp is probably 34 degrees celcius, but if your not full of shit, I better get one of those celerons you have, 2 degrees ferenheit above freeseing in a 60 or 70 degree room meens that has got to be one cool chip.

    11. Re:Celcius or Farenheit? by Anonymous Coward · · Score: 0

      Formulas to use:
      C(f)=(f-32)*5/9
      f(c)=c*9/5+32

      Hope you find it useful

    12. Re:Celcius or Farenheit? by Anonymous Coward · · Score: 0

      The packaging of the chip would be standard, wouldn't it? Celerons always come in same Intel-made packaging. I don't know if it would be any different for OEM chips.

      The core temp doesn't really make sense. The amount of heat generated in a given slice of time makes more sense. (I can't remember proper units for that) Temperature is really a measure of its current electron state, if I remember correctly. Its more useful to indicate how much it changes the electron state in a given length of time. (amount of heat generated) This would eliminate differences in the test. They may have tested celeron with no fan attached, which would make the test foobar, since the celeron was designed to have fan attached (or maybe more like: "This thing keeps melting. Lets attach a fan.")

      But that's just the scientific part of me. Of course they had to dumb it down for the marketing people. :)

      >>>>>>>>> Kvort the Duck

    13. Re:Celcius or Farenheit? by the+red+pen · · Score: 2

      Please ignore the complete assholes who are flaming you for an obvious typo. Anyone with a three-digit IQ knows you meant 34C.

    14. Re:Celcius or Farenheit? by QuMa · · Score: 2

      48C? Get a new water heater. If it's the kind that stores the warm water, not only is it not really hot enough for washing up etc, but also, it's a breeding place for all kinds of germs...

      average water heater is 60-80C.

    15. Re:Celcius or Farenheit? by Anonymous Coward · · Score: 0

      The scientific side of me says that if this thing has three times the battery life then who gives a shit whether they measured core temperature or some other temperature, obviously it's cooler where it counts.

    16. Re:Celcius or Farenheit? by Anonymous Coward · · Score: 0

      Wow! An exothermic chip? Exothermic means it gives off hot, and I'm pretty sure most CPUs give off heat. At least mine does.

    17. Re:Celcius or Farenheit? by Anonymous Coward · · Score: 0

      >The amount of heat generated in a given slice of time makes more sense. (I can't remember proper units for that) Heat is a form of energy, and energy is measured in joules. The ammount of joules in a slice of time (Joules per second) is a watt.

    18. Re:Celcius or Farenheit? by mindstrm · · Score: 2

      No.. they mean 48C. Your celeron is running at 34 C (The F is a typo)... because it has a BIG HEATSINK AND FAN. The crusoe will run at 48C *without* any cooling.

      And yes.. the PIII would.. well.. not melt, but possibly do some damage to itself... but it only runs at 113 C *without* a fan & heatsink.

    19. Re:Celcius or Farenheit? by Anonymous Coward · · Score: 0

      The temperature > Rise on other electrical aparatus is the temperature of the device plus ambient. So if an electrical motor for example is rated at 40C Rise and is running in an ambient temperature of say 70F, then the motor will get to a temperature 40C + 70F. I would think yhat the temperature rating on a chip would be the same way.

    20. Re:Celcius or Farenheit? by Anonymous Coward · · Score: 0

      Bud, In some countries that hot is illegal - something about kids getting scalded in the bathtub, and compulsory to fit them in oldies and disabled homes - so the germs can kill them quicker

    21. Re:Celcius or Farenheit? by QuMa · · Score: 1

      Illegal? And what about stoves? They can heat to 100c, are they illegal too? Sorry, but I have a hard time believing this...

    22. Re:Celcius or Farenheit? by druthers · · Score: 1

      To run at a temperature below amibient, it would need to be an ENDOTHERMIC chip and thus register as cold because it was absorbing heat from your hand, the environment, the calibrated sensing instrument, or whatever. If it could do that you could drink your glass of tea "iced" by a wireless, multi-cpu design!

      --
      *** "It's only trivia until you need it." JMR ***
    23. Re:Celcius or Farenheit? by John+Fulmer · · Score: 1

      My bad.. Exo- , Endo-.. Not terms I use every day... :)

    24. Re:Celcius or Farenheit? by vanadium · · Score: 1

      heh - some ppl with a two-digit IQ would probobally be able to figure that one out too.

  15. The Webpad by Nerds · · Score: 1

    Can someone tell me more about this thing? Specifically, how big (small) is it compared to the existing handhelds? Also, what kind of GUI were they using? Thanks.

    --
    My other .sig is 'The Art of Computer Programming'
    1. Re:The Webpad by Smack · · Score: 1

      Big. Like a 15.4" LCD w/ a thin border around it, about 3/4" thick. It was hard to tell what it was running, you could just see windows floating around.

    2. Re:The Webpad by mircea · · Score: 2

      About the size of a 12" LCD panel (thickness, too). I couldn't figure out which window manager it had on, the image was quite bad.

    3. Re:The Webpad by MTO · · Score: 3

      The web pad was about 8.5 by 11, and had no keyboard. it had a one inch frame, and the rest was a touchscreen. it looked about one inch thick. The model was using some sort of stylus on the screen, and it had a drop down virtual keyboard. Otherwise, it looked like a busy KDE or Gnome environment. They were really showing off the web-browsing more than the environment. I never heard any details about how the device connected to the net either.

    4. Re:The Webpad by Chris+Brewer · · Score: 1
      If you look here you will see a couple of concepts:
      • a really thin laptop running Windows 98 with Explorer and Solitaire.
      • a flat panel with a screen - similar to ST:TNG+ (with Klingon style buttons!) running Netscape which happens to be pointing to Slashdot!
      --
      Consultancy: If you're not part of the solution, there's money to be made in prolonging the problem
    5. Re:The Webpad by cave76 · · Score: 1

      I was waiting to hear about net connectivity also. The only thing that was relavent to connectivity in any of the presentations that I saw was when the marketing guy showed a slide of a wireless connection to a hub or desktop PC in your home that had a physical net connection. He explained how this would be ideal for carrying a small internet device, such as the webpad or one of those smaller devices, around the house. I don't see how this would translate into longer range wireless net connectivity, nor did I hear them explain any other way they plan to achieve this.

      Mike

  16. Mobile Linux? by Ravagin · · Score: 1

    Somebody wanna clarify for a non-Linux-user? Is Mobile Linux a functional equivalent to PalmOS or WinCE?

    ===
    -Ravagin

    --

    Karma: T-rexcellent.

    1. Re:Mobile Linux? by PureFiction · · Score: 2

      Id say it means small kernel and system overhead (efficient use of memory, and small executables) as well as efficient use of other system resources.

      So in that sense it would be similar to the design of the PalmOS, i wont even mention WinCE as its such a pile of crap...

      Cant wait to see the code released so we can know exactly wot was changed / optimized / added.

    2. Re:Mobile Linux? by DGolden · · Score: 1

      Hmm... I wonder how "mobile linux" will stack up against EPOC32, which is a pretty good OS for palmtops?

      Also, this processor could be pretty cool for running the tao virtual machine that amiga are using for their network environment - remember, amiga have already said they're going to run stuff on top of linux in palmtops.

      --
      Choice of masters is not freedom.
    3. Re:Mobile Linux? by Troed · · Score: 1
      You really should link Epoc to Symbian you know ;)

      But I agree, Epoc is an incredibly good operating system for handheld computers and smartphones. It will be interesting to see Linux competing in this arena.

      --
      it's in my head
    4. Re:Mobile Linux? by Anonymous Coward · · Score: 0

      >Also, this processor could be pretty cool for >running the tao virtual machine that amiga are >using for their network environment - remember, >amiga have already said they're going to run >stuff on top of linux in palmtops. And note that even if it(the taoAmiga -thing) sounds like the Gateway AmigaOE/AmigaObjects, it is a totally different beast... This time they deliver .... hopefully... 8|

  17. Linus needs to work on his Quake skills! by Anonymous Coward · · Score: 1

    I watched the webcast, and linus REALLY needs to bone up on his Gibbing technique :).
    Overall, I am quite excited about the new processor(s), and I can't wait to see who the Transmeta customers are!

    1. Re:Linus needs to work on his Quake skills! by Chas · · Score: 3

      You forget, Linux, by his own admission, doesn't care about games.

      He only cares about compiling his kernels.


      Chas - The one, the only.
      THANK GOD!!!

      --


      Chas - The one, the only.
      THANK GOD!!!
    2. Re:Linus needs to work on his Quake skills! by alphaparticle · · Score: 1

      Correction. David did introduce him as a "closet" quake player.

    3. Re:Linus needs to work on his Quake skills! by tialaramex · · Score: 1

      Linus was partly responsible for one of the Quake (that's Quake 1) ports to Linux IIRC

  18. ObBeowulf by Anomalous+Canard · · Score: 1

    I don't have to spell it out, do I. /me wipes drool from chin.
    Anomalous: inconsistent with or deviating from what is usual, normal, or expected

    --
    Anomalous: deviating from what is usual, normal, or expected
    Canard: a false or unfounded repor
  19. So, Linux programmers can... by Jimhotep · · Score: 1

    Linux programmers can start building software
    to run on hand helds.

    And what about wearables???? Sounds like somebody
    could build a cool computer into a tuxedo! Or would
    that be too obvious?

    1. Re:So, Linux programmers can... by Anonymous Coward · · Score: 0

      The software already exists. Didn't you know we mean Linux when we say Linux? vi will run fine on this hyar thing. the mail command should work well, also, and will leverage the vi command for use in composing email. We call it a Communications Suite, vi and mail.

      Anything more than that is bloatware and doesn't fit in with the Unix philosopy.

  20. Multiple OS use on the Crusoe by didjit · · Score: 1

    Man this is really sweet. Since it will be able to run any x86 OS, it should make the transition from windows to linux easy for windows user, plus it will make it possible to easily switch between several OS's such as linux, bsd, be, java, etc. Much thanks to Chris for the insightful questions at the announcement. Much unlike the typical question by the woman from the AP who just had to ask about possible IPO possibilities. I can't wait to get a Crusoe

    1. Re:Multiple OS use on the Crusoe by GeorgeH · · Score: 1

      It won't really smooth over the transition from Windows to Linux for anyone. Remember, 95% of Windows users have the hardware to run Linux. The main transitional problem is that Linux has is that it has different (some might say more complex) interface then Windows. To change your network settings, you have to run ifconfig instead of Right-clicking on Network, going to properties, blah blah blah. This discussion has been had a thousand times, and expect to hear it again. I don't really feel like going into the whole simplicity vs. power debate, but suffice it to say a different processor won't change much.

      Another of your arguments was that it would be easier to switch between different OSs. There really won't be much difference between the current way and the Crusoe way. Remember, Linux, BSD, Be, Java, etc, all run under the x86 platform. Unless the Cruesoe lets you open up multiple emulated processors, which I don't think is the case, you'll still have to reboot and repartition, just like you have to under x86. The only multiple OS benefit I see is the possibility of running Windows and MacOS on the same computer.

      I agree that the Crusoe looks cool and all, but I don't see it revoltionizing the way we compute.
      --

      --
      Why can't I moderate something "Wrong" or at least "Grossly Misinformed"?
    2. Re:Multiple OS use on the Crusoe by Jamie+Zawinski · · Score: 2

      Since it will be able to run any x86 OS, it should make the transition from windows to linux easy for windows user, plus it will make it possible to easily switch between several OS's such as linux, bsd, be, java, etc.

      There's another chip that can run any x86 OS: it's called the Pentium.

      Just because this chip is good at emulating other instruction sets doesn't mean that you can magically run multiple OSes at the same time without rebooting. That's still an incredibly hard problem.

      They did say that the chip itself can run multiple instruction sets without resetting between them, but that's still a far cry from saying they've solved the problem of scheduling multiple OSes: since every OS has its own scheduler, you'd need a meta-OS with a meta-scheduler. Not to mention the problems of locking hardware access, etc.

    3. Re:Multiple OS use on the Crusoe by um...+Lucas · · Score: 3

      No offense to you or anything, but how in the world was that "interesting"?

      The Crusoe does nothing for making the switch from Windows to Linux easier. Nada. Same for BSD, Be, etc... It does just as much as did the Athlon, except Crusoes effectively a version 1.0 product coming from a brand new company, as opposed to Athlong which is a much more matured product that merges the best of two already mature companies products.

      Java shouldn't even be included in that discussion, unless a JVM is ported to run directly on top of Crusoe.

      And of course any journalist worth their salt HAS to ask about IPO's. With Amazon's, Ebays, Redhats, VA Linuxs, etc, it's the simple truth that Transmeta will in all likely hood go public (they talked about who supplied their first round and second round of financing. The next logical step is to go public). That, plus the hype they've drawn to them selves, and the market they've painted for their processor.

    4. Re:Multiple OS use on the Crusoe by BinxBolling · · Score: 1
      They did say that the chip itself can run multiple instruction sets without resetting between them, but that's still a far cry from saying they've solved the problem of scheduling multiple OSes: since every OS has its own scheduler, you'd need a meta-OS with a meta-scheduler. Not to mention the problems of locking hardware access, etc.

      I don't see too many people wanting to run multiple OSs at once - it's an absurd waste of resources. But I do see people using this technology to build emulators that compare to native hardware. Consider what might happen to SoftWindows, on MacOS, if this chip were able to run PPC code at a good pace. In that case, one could write a SoftWindows that ran x86 apps as fast as if they were on x86 hardware. However, the trick is that you need OS support to really do this right.

      What I'm envisioning is that there's a register somewhere in the chip that indicates what instruction set it's currently using. There's also an instruction to atomically set both this register and the program counter register at the same time. Then, the OS stores a value for this register (along with the program counter and various other info that it already maintains) local to each process. When a process is dispatched, that register gets set simultaneously with the program counter, using the instruction I mentioned. Of course, this means your scheduler needs to be written in the VLIW instruction set (since this instruction won't exist in an emulated instruction set). But basically, any time a context switch is required to get from one chunk of code to another, they can have different instruction sets. (Note also that hardware interrupt vectors will have to contain a value for this register as well as a memory address.)

  21. Here's a thought by ctembreull · · Score: 3
    So, Linus has written "Mobile Linux" for these chips. The chips themselves are low-power and
    low-cost. This is all very, very good.

    Given the mention of the "touchpad screen", is it possible that a Crusoe/MLinux system would
    be able to serve as the basis for kiosk-class systems, like ATM machines, information stands,
    and so on and so forth?

    If the chip is that cheap, and the OS is free, wouldn't it sort of make sense to harness that
    and direct it towards those sort of ubiquitous consumer machines that you are starting to see all over the place?


    Chris Tembreull
    Web Developer, NEC Systems, Inc.

    My opinions are my own, and nobody else's.

    --

    Chris Tembreull
    "My karma just ran over your dogma."
    1. Re:Here's a thought by XNormal · · Score: 2

      Well, isn't that exactly what they are already doing? How can a "ubiquitous consumer machine" get any more ubiquitous than a cordless webpad?

      If you really want a kiosk form factor so badly you can take the webpad and stick it on the front of a big, empty box.


      ----

      --
      Stop worrying about the risks of nuclear power and start worrying about the risks of not using nuclear power.
    2. Re:Here's a thought by TheJohn · · Score: 1

      Given that the big thrust of Crusoe seems to be low power and heat, which is not a problem is kiosks, I don't see that there is a compelling reason for the switch.

    3. Re:Here's a thought by jopasm · · Score: 1

      True enough; however, the chips seem to be
      relatively inexpensive, in-production, and
      fast (compared to other "low power" and "cheap"
      processors). In addition they are x86
      compatible and can run existing apps. This
      could make them excellent choice for kiosks.
      Additionally, their low power comsumption brings
      up the possibility of "alternative" power sources
      (solar, etc) - thus making cheap, easily
      install kiosks a viable product - sorta like
      the solar powered satellite "payphones" that
      have been developed. The next Woodstock, for
      example, could have a bunch of webpads on poles
      w/ solar panels on top. Swipe a credit card
      through and send e-mail to your friends.

      Also brings up the possibility of solar-recharged
      notebooks - very nice. :>

      --

      ObTagLine: The more you run over the 'possum, the flatter it gets.

    4. Re:Here's a thought by dillon_rinker · · Score: 1

      So you're saying that kiosks could be run using vacuum tube technology, with all the heat they produce and power they require? :)

    5. Re:Here's a thought by Mars+Saxman · · Score: 2

      If you'll pardon my pedantry for a moment -

      THERE'S NO SUCH THING AS AN "ATM MACHINE."

      Nor is there an "automatic ATM teller machine", a "PIN number", a "PIN ID", a "CPU processor", "RAM memory", a "DSL line", or any other such redundancy (all but the first of which phrases I have actually heard in conversation).

      Whew, got that out of my system for the day.

      -Mars

    6. Re:Here's a thought by datazone · · Score: 1

      don't forget the ever popular "NIC cards"

      --
      Its spelt "L-I-N-U-X", but pronunced as "Free Beer"
    7. Re:Here's a thought by Anonymous Coward · · Score: 0

      The best one that is actually used (in car commercials):
      "ABS antilock braking system"

    8. Re:Here's a thought by IntlHarvester · · Score: 1

      "Kiosks" that played music were run using vacuum tube technology for many years. They're called jukeboxes.

      (If you can be a weenie, so can I!)
      --

      --
      Business. Numbers. Money. People. Computer World.
    9. Re:Here's a thought by chip+guy · · Score: 1

      Don't forget TTL logic

  22. Like I know what I'm talking about by sheckard · · Score: 1

    that should read 34 C... I wish it ran at 34 F!

  23. History by dreamchaser · · Score: 4

    Intel must be collectively quivering in their proverbial shoes after this conference. After watching and listening, I am wondering, are we seeing the Next Great Innovation(tm) in processors? The paradigm that Transmeta has created with Crusoe is so different, I have the feeling I was watching a new chapter of the history of computing being written before my very eyes.

    What is does under the hood, between it's translation of instructions and its optimization of the actual code (profiling on the fly), is phenomenal.

    1. Re:History by jkovach · · Score: 1

      I don't think this is really a paradigm shift because similar things have been done before. DEC's FX/32, the Java Virtual Machine, and the 68K emulation in the PowerPC MacOS have been doing this for a while. The biggest thing Transmeta has going for them right now is low power consumption, not a software upgradeable processor. Besides, software upgradability can't make the hardware run faster. There's a limit to how much faster you can get a processor by simply upgrading it's microcode.

    2. Re:History by jkovach · · Score: 1
      I don't think this is really a paradigm shift because similar things have been done before. DEC's FX/32, the Java Virtual Machine, and the 68K emulation in the PowerPC MacOS have been doing this for a while. The biggest thing Transmeta has going for them right now is low power consumption, not a software upgradeable processor. Transmeta isn't even positioning this as a software-upgradable processor - they're positioning it as an X86 alternative, and as others have mentioned, the X86 architecture is pretty darn lousy these days.

      Besides, software upgradability can't make the hardware run faster. There's a limit to how much faster you can get a processor by simply upgrading it's microcode.

    3. Re:History by BinxBolling · · Score: 1
      I don't think this is really a paradigm shift because similar things have been done before. DEC's FX/32, the Java Virtual Machine, and the 68K emulation in the PowerPC MacOS have been doing this for a while.

      AFAIK, all the examples you name are software emulation. Sure, FX/32 allows an Alpha to emulate an x86 at speeds comparable to x86 native hardware. But an Alpha capable of doing this is much more expensive than the x86 to which it compares, and the program still executes at only around 70% of the speed of native-compiled Alpha code.

      If Crusoe (or its descendants) allow you to execute an instruction set at 90+% of the speed of a CPU that is hardwired for that set, when the Crusoe and the hardwired CPU in question are the same price, then this could indeed be a paradigm shift.

    4. Re:History by IntlHarvester · · Score: 1

      The key is really the market the CPU is designed for. Alphas are only feasible for high end servers and workstations. MacOS's emulator was designed for Macintoshes only. The AS/400's emulator was designed for AS/400's only. Hardware Java Machines haven't shown up in realworld product yet.

      The "paradigm shift" is that you can now run x86 binaries at a reasonable speed with very little power and heat. If the Crusoe required Alpha or PIII-like watts, it would hardly be a big deal.
      --

      --
      Business. Numbers. Money. People. Computer World.
    5. Re:History by Guy+Harris · · Score: 2
      AFAIK, all the examples you name are software emulation.

      You don't know far enough. FX!32, at least, did binary-to-binary translation, and at least some JVMs translate Java bytecodes to native code. (I don't know whether the 68K emulation in PowerPC MacOS ever did binary-to-binary translation.)

      There are other examples of binary-to-binary translation as well - the IBM System/38 and AS/400 (compilers for which generated a high-level very CISC pseudo-instruction set; that code gets translated into the native instruction set), a 68K Smalltalk done by L. Peter Deutsch that translated Smalltalk bytecodes into 68K code (caching the results of the translation), binary-to-binary translators that turned 16-bit stack-machine HP 3000 code into 32-bit PA-RISC HP 3000 code, and probably others.

      Binary-to-binary translation isn't a Transmeta invention; the hardware assistance they provide for it is, as far as I know, as is the idea that the native instruction set isn't even exposed to the OS or the BIOS/PROM monitor.

    6. Re:History by JackAssPenguin · · Score: 1

      I doubt it.

      Remember when the PowerPC came out?

      3 huge companies threw all their weight behind the chip and it was meant to herald in the new era of Risc processing. The intel chip was doomed.

      That was a few years ago. What happened?

      --
      "DNA is God's contribution to the Open Source movement"
    7. Re:History by infodragon · · Score: 1

      What is does under the hood, between it's translation of instructions and its optimization of the actual code (profiling on the fly), is phenomenal

      Taking that into consideration, Transmeta could put 2 CPU cores on the die and bam you have doubled the speed of your CPU. This would be utilized by the translation layer making it seamless to the applications.

      Why stop at 2 go to 4 or 8 or ... the translation software will utilize the extra cores transparent of the softare. Now take this with multiple units on an SMP machine you have one hell of a powerfull server.


      --
      If at first you don't succeed, skydiving is not for you.
    8. Re:History by BinxBolling · · Score: 1
      You don't know far enough. FX!32, at least, did binary-to-binary translation, and at least some JVMs translate Java bytecodes to native code. (I don't know whether the 68K emulation in PowerPC MacOS ever did binary-to-binary translation.)

      I'm aware of how FX!32 works - I went and looked it up after I wrote that. But I don't retract my point: it's still a strictly software solution, and the hardware provides no support for the process. Presumably, a considerable amount of software infrastructure is needed to make it work - x86 binaries have to be recognized and translated, and those translated binaries stored, for example. OTOH, unless you're trying to handle multiple instruction sets at once, there's no reason why Crusoe should require any more 'help' from the OS than an ordinary x86. And even if it does start handling multiple instruction sets, the solution is considerably simpler and more elegant than FX!32.

      Also, FX!32's performance target is considerably less ambitious than that of Crusoe: The FX!32 white paper says they're shooting for code with 70% of the performance of a native-compiled Alpha application.

  24. This information is not correct by mdtanx · · Score: 4

    This Crusoe information is all very incorrect. I can't believe Slashdot was so badly misled. If you go here (http://www.nitrozac.com), you'll see what tech-savvy readers have known for months: Transmeta is building a multi-story abacus. In fact, I thought it was unveiled a month or two ago, but mysteriously disappeared.

    1. Re:This information is not correct by MWright · · Score: 1

      Yes, but it can run Linux. That's the good part: It's the only abacus that can run it.


      -----

      --
      "But really, I think life is just a game of Mao Nomic." -Purplebob
    2. Re:This information is not correct by Anonymous Coward · · Score: 0

      The project you're talking about is codenamed "International Space Station" It is a very hush hush project that is getting massive funding from various governments. Billions of dollars. Supposedly the same people that produced the Aurora craft that abducted Special Agent Mulder are working on this.

  25. no release of the vliw instruction set? by rogerbo · · Score: 3

    they said that they may not release the native vliw instruction set because they want to keep the freedom to change it in the future and don't want to worry about breaking compatilibility.

    While this is a good thing in one sense it means
    we're limited to only the code morphing software they want to release (since that's native).

    So if they don't release code morphing software for PPC, or MIPS or SPARC or ALPHA then you're SOL, you can't write it. And may also be difficult or impossible to write a native version of linux.

    Anyone have any thouhgts on this?

    1. Re:no release of the vliw instruction set? by AJWM · · Score: 2

      They also said that the VLIW instruction sets are different for the TM3120 and TM5400 chips, the Code Morphing Software has different back-ends for each chip (and is compiled to different targets).

      But yeah, unless/until Transmeta releases the specs or someone reverse-engineers the instruction set, we won't see any do-it-yourself CPU emulators.

      They did mention that one of the demos was a Crusoe running Java bytecode 'natively' (ie CMS translating bytecode directly to VLIW), so perhaps we will see some other CPU emulations in future.

      However, the current architectures (and it wasn't clear whether this was of the chip or of the system the chip is in) don't seem to allow for dynamically switching instruction sets - once the thing has booted up the CMS code the pathways to that low level stuff are closed.

      This makes sense for Transmeta at this point to keep the market from getting confused, but I hope that once this settles out that they start looking at making the thing more flexible.

      --
      -- Alastair
    2. Re:no release of the vliw instruction set? by franl · · Score: 5

      rogerbo wrote: So if they don't release code morphing software for PPC, or MIPS or SPARC or ALPHA then you're SOL, you can't write it. And may also be difficult or impossible to write a native version of linux.

      Linus said that they explicitly decided against doing a native version of Linux for the Crusoe. The whole idea of Crusoe is to keep you from having to recompile while still letting you take advantage of advances in the underlying CPU architecture. Nobody should want a native Crusoe application, because when a new Crusoe comes out with different instructions or whatever, you'll have to recompile. As much as I hate the phrase, this is really a paradigm shift in processor and OS technology.

      The lack of a SPARC or Alpha or PPC morphing layer is probably more a pragmatic decision on Transmeta's part. They can't do it all (right away). They didn't rule out a morphing implementation for PPC, Sparc, etc., but they get the most bang for their buck from doing the x86 first.

    3. Re:no release of the vliw instruction set? by Anonymous Coward · · Score: 1

      It might surprise you to learn that this paradigm shift already happened many years ago in the form of the AS/400. The AS/400 uses a technology independent machine interface, TIMI (or MI for short), to isolate the underlying hardware from the execution environment. Thanks to the MI thousands of programs compiled on CISC machines continued to run perfectly fine without recompilation when the AS/400 migrated to RISC PowerPC chips a few years ago. I'll admit that Crusoe sounds very neat (and I'll be on the lookout for such a portable), but the idea that you can keep running applications without recompilation isn't new.

    4. Re:no release of the vliw instruction set? by atspink · · Score: 1

      >As much as I hate the phrase, this is really a
      >paradigm shift in processor and OS technology.

      Um, no. Look up the IBM AS/400 which has been shipping for quite a long time. Its contains an MI(Machine Interface) layer to which all programs are written and then the programs are translated and executed by the MI layer. This is basically the same thing that "code morphing" does.


    5. Re:no release of the vliw instruction set? by jefflinwood · · Score: 1

      Maybe I'm just being stupid here, but most Intel chips past the original Pentium, the old NexGen 586 and 686, and the newer AMD chips all work this way - they have an internal instruction set (microcode, I guess), and x86 instructions are translated into the native format. I can run old DOS stuff from 1981 on my brand new Pentium III if I wanted to, and nothing in the hardware has broken it.

      --
      Tips and Tricks for Mozilla
    6. Re:no release of the vliw instruction set? by Esperandi · · Score: 1

      While I would really like to start writing assembly code for the VLIW processor and all that (hey, i think it would be fun, don't look at me like that), i do see one advantage to this view...

      Theoretically I could be sitting using my brand year-old 700MHz Transmeta-based notebook, and use it to go to the Transmeta website. Wow, an upgrade that increases my speed by 50%!

      This probably will not happen, I'm assuming they pretty much squeezed as much optimization into their code morphing module as possible before release to have an impressive product, but in the future it could happen when they emulate other platforms.

      For instance, they might release a sort of dual-boot configuration for the code morphing module that lets you run Alpha-based apps. They might release this early, without many optimizations, and then it gets better over time. If someone swoops in and write native VLIW code for it, well, that code is going to get broken. Also, with separate code morphing engines, the VLIW code would become a very strange type of kinda-platform-specific beast. There'd be releases of software "compatible with revision 9.02.3 of the Crusoe TM5400 which emulates the Alpha platform" and such things.

      But in that case, recompilers would come along... it would be a fairly sizable mess, but it will happen. You think no hacker is going to get in there and get curious about the code-morphing module? Its stored in DRAM, theres gotta be a way to get at it ;)

      Esperandi

    7. Re:no release of the vliw instruction set? by Anonymous Coward · · Score: 0

      they said that they may not release the native vliw instruction set ... Anyone have any thouhgts on this? Yes. If they keep the native instruction set proprietary, I won't buy any. Why should I buy hardware I'm not allowed to hack? Are they afraid somebody will come up with a better code-morpher, or that somebody will write an emulator for a machine language of which they don't approve? I don't want to have to wait for Transmeta to make an interpreter to let people run X machine code. I don't want to take a performance hit of 3 or more by writing an interpreter on top of an interpreter, when the native architecture is designed solely to run interpreters efficiently. Why the hell do they have Linus for a figurehead if they're going to pull this kind of shit?

    8. Re:no release of the vliw instruction set? by Anonymous Coward · · Score: 0
      they said that they may not release the native vliw instruction set
      ...
      Anyone have any thouhgts on this?

      Yes. If they keep the native instruction set proprietary, I won't buy any.

      Why should I buy hardware I'm not allowed to hack? Are they afraid somebody will come up with a better code-morpher, or that somebody will write an emulator for a machine language of which they don't approve?

      I don't want to have to wait for Transmeta to make an interpreter to let people run X machine code. I don't want to take a performance hit of 3 or more by writing an interpreter on top of an interpreter, when the native architecture is designed solely to run interpreters efficiently.

      Why the hell do they have Linus for a figurehead if they're going to pull this kind of shit?

    9. Re:no release of the vliw instruction set? by Anonymous Coward · · Score: 0
      Nobody should want a native Crusoe application, because when a new Crusoe comes out with different instructions or whatever, you'll have to recompile. As much as I hate the phrase, this is really a paradigm shift in processor and OS technology.

      Paradigm shift: yes and no, or with other words "partly". I remember, for example, the Siemens 7530 big iron many, many years ago. While it could not do anything as nice as code morphing on the fly, it had already a layered instruction scheme (like many other mainframes too). The simple idea is to have a basic instruction set (something inbetween microcode and normal opcode instructions). Then on this layer the system programmers built other high-level cpu instructions, which then were used by applications. Of course, in this scheme applications also used low-level instructions in combination with the macro instructions. But at least part of the basic idea was already there, to have a high-level instruction and a very basic cpu instruction set (the latter avoided as much as possible in applications). Well, the term "x86 high level cpu instruction set" is an oxymoron, of course...

      Harry Deer

    10. Re:no release of the vliw instruction set? by BrentN · · Score: 1
      I believe the question of morphing implementations for various platforms will be answered *after* Transmeta's penetration into the PC laptop and handheld market. At this point, there is no financial benefit for them to spend valuable man-hours supporting other platforms. In the future, though, it may be worthwhile for them to license the VLIW instruction set to Moto or to Sun so that Transmeta doesn't have to do the scutwork. It is a wise decision not to do that from the get-go, however - it is far better for them to concentrate on penetrating their initial target market so that they have a strong revenue base from which they can finance improvements to the platform. The venture capital won't last forever!

      In some sense, the code morphing software acts like a hardware abstraction layer like a microkernel architecture. It would be silly of them not to capitalize on this in the long-term. So don't worry about it. If you want to see it sooner, tell all your friends to buy WebPads and the whatnot....

  26. I want upgrade!!!! by pecka · · Score: 1

    Hey IBM what about upgrading my one year old ThinkPad???...please....pretty please....damn...

  27. well it wont be anymore but... by dragontails · · Score: 0

    FIRST POST!!!!!!!!!!!!!!!!!!! MUAHAHA! I really hate to say this because even Linux needs some competition but I wonder whether dear Billy is pondering buying out any chip makers now. Embedding a tiny linux OS onto a chip with incredible potential should scare the hell out of him. Anyone have any ideas on implications on the future of hardware/software now?

    1. Re:well it wont be anymore but... by Anonymous Coward · · Score: 0

      why not, billy loves the x86 instruction set. when there's a jvm code morpher, now that's when the fun starts...

  28. Is the microprocessor so important? by ektor · · Score: 2

    It seems hard to believe that they'll get twice as much battery life as existing laptops. I'm no expert, but I'd say that the screen, HDD, DVD drive, etc waste much more energy than the microprocessor. Anybody that knows this stuff cares to give his opinion?

    1. Re:Is the microprocessor so important? by Mark+F.+Komarinski · · Score: 2

      IBM's Travelstar IDE drive uses a max of 5.0W @ +5V at startup. Typical is about 2.5-3.0W @ +5V. I'm not sure what voltage is used for the PIII (5.0 draw or what actually gets to the PIII, which is about 1.x V), but it'll be more than that.

      There was mention of two types of laptops, one without removable media that would be less than 3 lbs. To me, removable media is probably the most power hungry (constantly spinning up CD-ROMs and floppy disks can really chew up battery life).

      --
      -- Ever notice that fast-burning fuse looks exactly the same as slow-burning fuse? I didn't... (Edgar Montrose)
    2. Re:Is the microprocessor so important? by dragontails · · Score: 1

      IMHO, it is rather unfortunate that this chip may not get much of a chance to prove itself (now that I think about it).
      The rate of microprocessor improvements (specifically speed) has increased exponentially, while at the same time, HDs, memory, input devices, and especially softare overall are...
      BLOATED?

      Unless Transmeta manages to spark a revolution to produce smaller, more durable and adaptable devices, all this excitement may be short lived and they will end up plodding along like AMD has been doing for the last while despite superior products.

      dt.

    3. Re:Is the microprocessor so important? by ToLu+the+Happy+Furby · · Score: 4

      It seems hard to believe that they'll get twice as much battery life as existing laptops. I'm no expert, but I'd say that the screen, HDD, DVD drive, etc waste much more energy than the microprocessor. Anybody that knows this stuff cares to give his opinion?

      Actually, if you read their web site, Transmeta gave their "opinion" on this right here. Essentially, the gist of it is that the battery savings are quite significant, even on one of those giant laptops with the 15" screens and the DVD players, and even while playing a DVD in software (which, because it requires a nearly constant (and rather hefty) level of CPU power, can't take much advantage of their technology which dynamically scales down power usage and voltage to meet the current system needs).

      Basically, according to the tables on the above page, the worst-case for a Crusoe processor--running soft DVD (2 watts used in CPU + Northbridge) on a bigass notebook (8 watts)--gives 3.2 hours battery life. IIRC, one of those new G3's (and remember, a G3 consumes *way* less power than any (native) x86 chip) can barely manage an hour and a half.

      Plus, they're not even taking into account the fact that unlike any other notebook on the planet, these suckers don't need a fan; that should be reflected in the 8 watt system overhead, but isn't. (Not sure how much power a fan takes, but it has to be significant.)

      Now...in the normal case, in which the CPU is at full throttle only a little bit of the time, then Crusoe starts to clean up. For one thing, as they point out, traditional notebooks try to conserve power by just shutting off the CPU when it's not being used. The problem with this is it doesn't help the normal case when it's being used only a little bit, and it adds a noticable delay while it gets switched on again, which for most users is a lot more important than its peak speed anyways. The T5400 (the especially badass one that's not coming out until the summer) gets around this by scaling CPU power and voltage to meet current needs--and it shows.

      Witness their mobile benchmark report [note: 116k pdf], based on a new benchmarking methodology they invented (read up on it he re [note: 93k pdf]) which:

      1) mirrors actual use--i.e. doesn't run full throttle all the time, which almost never happens under normal use, especially for a notebook

      2) includes metrics for energy efficiency--that is, it reports not just work/time, but work/WattHour and work/time/WattHour.

      For those who don't want to check it out, the result is that across 6 tests (operating system load, system idle, Office 2000, web browsing, mp3 playback, and soft DVD playback) comparing the T5400 to a P3 500, the Crusoe processor was:

      95.3% as fast (yeah, this includes the "system idle" test, which is a bit of a cheap freebie in this category) [note--this is just my straight average of the 6 categories, which is absolutely unmathematically correct, but oh well]

      409.2% as efficient in terms of work/Watt-hour

      395.3% as efficient in terms of work/time/watt-hour.

      All in all, pretty damn impressive. And it's worth noting that it's over 6 times as efficient in the system idle test--which is what your system probably does most anyways.

      Of course, this only measures the power drained by the CPU+NG, and not the screen, HD, etc. But...I have no trouble believing that a CPU that's 4 times as efficient under normal use will give 2 times the overall battery power.

      I gotta go now, but the point of all this rambling is, this chip is pretty damn neat. I'm impressed.

    4. Re:Is the microprocessor so important? by plsuh · · Score: 1

      >IIRC, one of those new G3's (and remember, a G3
      >consumes *way* less power than any (native) x86
      >chip) can barely manage an hour and a half.

      Actually, Apple claims that the current PowerBook G3 can run up 5 hours on one battery. See Apple's spec page. Independent reviews of the PB G3 give a "real life" work time of about 3 hours while playing a DVD. (Emphasis mine. 3 hours comes from playing Austin Powers twice, which has a run time of 90 minutes.) A table of the G3 processor's power modes shows that its typical power consumption is not terribly high compared to the TransMeta chips, given the older technology of the G3 and the large number of additional functional units in the G4:

      PPC750 (G3) @ 400 MHz
      --Typical 5.8 W
      --Maximum 8.0 W
      PPC7400 (G4) @ ?
      --Typical 5 W
      --Maximum 8 W

      (Quoted from Motorola's PPC 750 info page and PPC 7400 info page. There is no figure given for the MHz of the G4 on the page.)

      Another way to look at this is what percentage of the laptop's power consumption is taken up by the CPU. On a PowerBook G3, we know that one battery holds 50 Watt-hours and the average battery life is 3 hours, so the average power consumption rate is: 50 / 3 = 16.67 Watts. Thus, 5.8 Watts of the power consumed is used by the CPU, and the remaining 10.87 Watts are consumed by the rest of the machine; this is a lot higher than the 4 Watts of other system power on TransMeta's web page and still significantly higher than the assumption of 8 Watts of other system power for the multimedia laptop. Using the TM3120 figure of 2.9 Watts for consumption while playing a DVD, the total power consumption of a hypothetical TransMeta-based laptop is 10.87 + 2.9 = 13.77 Watts, leading to a battery life of 50 / 13.77 = 3.6 hours, only 36 minutes longer than the current PowerBook G3. That's not a revolutionary difference, IMHO.

      FWIW, I think that the TransMeta approach has very strong similarities to the AMD Athlon (in its on-the-fly translation of x86 instructions to the RISC-type internal instructions) and also to the Apple 68K emulator for its PowerPC-based machines and Java JITs (caching of the translated instructions after initial execution to speed execution the second time through). The weakness that I see of the TM approach is that they really are losing a fair chunk of the possible performance all of the time by their code-morphing approach. I personally don't see a lot of difference these days between a post-RISC superscalar architecture with strong branch prediction and speculative execution and a VLIW architecture with a translation unit on the front end that takes x86 instructions and turns them into VLIW instructions whose sub-pieces may be executing out of order. Both keep the functional units going a high percentage of the time.

      Another weak spot may be that the x86 registers are only shadowed once. If an x86 exception occurs, there is a rollback followed by and in-order execution to figure out where to break for the exception. In post-RISC architectures (both PPC and Athlon), there is a plethora of registers, enough to easily support multiple processor states, not just two. Rollbacks due to exceptions and speculative execution are much less costly in this architecture.

      Anyway, enough babbling from me. As a note, I am an Apple WebObjects engineer, but these views are my own. I just couldn't let some Furby ;-) get away with giving a PowerBook G3 only 1.5 hours of battery life.

    5. Re:Is the microprocessor so important? by ToLu+the+Happy+Furby · · Score: 1

      Erg. My fault; I remembered that a G3 Powerbook was barely able to play through the Austin Powers DVD...I just forgot the part about them playing it through twice...

      In any case, the TM chips still consume quite a bit less power than a G3, and even less under normal operating conditions, which will mean more battery life with lighter batteries.

    6. Re:Is the microprocessor so important? by Brasidas · · Score: 1
      Funny, what with this "paradigm shift" and all, everybody's forgotten about the PPC. The main difference is TM can run Windows software, it seems, through it's "code morphing" ability. Never heard of that till now, must be marketing speak for "emulation", like "service pack" instead of "patch".

      The original poster is correct, taking a look at the TM comparisons, the TM chip looks good until you add in the "system power" column across the board, then the difference is pretty minor. This reminds me of when I was working at a company where the common view was our software was a "dog" because of our reliance on an ancient (7 year old, unsupported) compiler. But the real problem was network and disk latency, which really adds up.

      Here, the situation is similar. To put this all in perspective, my ancient IBM L40 (386) lasted roughly the same time as my Thinkpad 600. Battery efficiency has improved, but not enough to offset the increased demands on the laptop from the display and os. Or another example: while processor speed has been vastly improved, why is the computer as slow as ever? It's the software.

      The point is, it is better to focus on other areas to make the most improvement. IMO, a couple hours improvement is not going to cut it, nor is the low price. Ditzel is touting this as a mobile , internet appliance. "The internet changes everything..." blah blah. I think while Linus and Ditzel were hunkered down in their office for the past 4 years, the rest of the world roared by, in the form of the Palm, Psion, Windows Powered. The Palm runs on AA batteries for *2 weeks*.

  29. I am convinced by Bartmoss · · Score: 1

    I want one of these things. No, make that two. If they work anything like advertised, they're the best thing that happened since Linus started to code that hobby OS.

    Again a tip for those not getting a feed from zdtv, try CNN, works great. It's advertised on their front page.

  30. Webcast notes by Signal+11 · · Score: 3
    You can find my (mostly) complete notes at this link. Executive summary follows --
    • Transmeta is focused on software, not hardware.
    • Strategic alliance with IBM for the hardware.
    • Two product families, both focused on mobile computing.
    • Linus got his ass kicked by the CEO in quake. =)
    • code-morphing - 75% of the "processor" will really be in software - translating everything down to the actual processor's internal instructions.
    • VERY low power consumption, somewhere around 1 watt during normal use. Less than a few milliwatts for "standby" mode. You can leave this thing on for weeks without difficulty. On a battery.
    • Transmeta website will be live at 2:00 CST, 12:00 PST, and 3:00 EST (for those who can't convert like me. *g*)
    • Linux will have the code-morphing code added in, as demonstrated on the webcast. However, no kernel patch is forthcoming - yet. Linus will likely make an announcement within 24 hours of this webcast (However, this is my opinion).
    1. Re:Webcast notes by Inoshiro · · Score: 1

      "VERY low power consumption, somewhere around 1 watt during normal use. Less than a few milliwatts for "standby" mode. You can leave this thing on for weeks without difficulty. On a battery."

      Not if you're cracking RC5-64 :-)

      Now I can crack on my N64, too! Bwuhahaha...

      ---

      --
      --
      Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
    2. Re:Webcast notes by puppet10 · · Score: 1

      So do you have something against MDT?

      --
      -------- This space intentionally left blank --------
  31. Mainboard? by NYFreddie · · Score: 1

    Will these chips use a standard mainboard, or will they be developed via proprietary means from the various laptop companies? How about prices on these? What architecture is expected - caching, memory addressing limits, etc, etc.

    --
    Barbie of Borg - She doesn't just Assimilate, She Accessorizes too!
  32. Only PIII-500 Performance?? by god_of_the_machine · · Score: 1

    Here I thought it would be a ground-breaking processor in terms of x86 performance... I had my trigger finger on "SELL" for my AMD stock but happily I dind't have to!

    Imagine a PIII-500 laptop by mid-year... it's already old hat! Sure, it has good batterly life, but it will only be applicable for the low end.

    --

    -rt-
    ** Evil Canadians are taking over the world. Learn about the conspiracy
    1. Re:Only PIII-500 Performance?? by jopasm · · Score: 2

      They're not trying to make the "newest fastest
      most-expensivest" processor - they're going to
      viable portable devices (1.5-2 hours is not
      viable :>). Sure, a PIII 500 isn't the fastest
      thing out there anymore, but it's not bad.
      A PIII 500 equivalent w/ 6-8 hours of battery
      life is a truly viable portable. Admittedly
      the other components may suck down a lot of the
      power, but hopefully having this chip available
      will spur IBM, WD, or some other storage solution
      company to create a large capacity, low power
      comsumption solution that is also affordable.
      The IBM Microdrive is a relative low capacity
      (340 meg?), but it's small with low power
      consumption.

      Basically, it will depend on your needs (as usual), but what I'm looking for in a notebook
      is compatibility w/ my existing apps (not a
      rewrite of them, like the WinCE version of Word),
      the ability to do basic tasks (word processing,
      some game-playing, e-mail, etc) and BATTERY
      LIFE - something I can lug around all day
      and use, not something that becomes dead weight
      after 1-2 hours of use.

      --

      ObTagLine: The more you run over the 'possum, the flatter it gets.

    2. Re:Only PIII-500 Performance?? by Anonymous Coward · · Score: 0

      Pair the new Motorola Fuel Cell with this instead of a battery and you really have something.

  33. Computers vs People by Da+Penguin · · Score: 0

    The way it is going, computers will be at the stage of superiority over humans.
    But it might not be really noticeable because we do not have the proper software to utilize it (like AI).
    Luckily there is the Great AIP (Artificial Intelligence Project), but if it was not for that not much would be happening

  34. Wow... by Anonymous Coward · · Score: 0
    I'm psyched to see a company come out with a product that's trully cool. Now we can actually have wearable pcs that'll actually kick ass. Very cool Linus and Transmeta... keep up the good work!

    Geek: someone who understands that their life as a balance of techology and the sublime and sees the truth trough it all...

    Blessings and Safe Travels

  35. Too cool by lhand · · Score: 1

    I sure wish I had one of these. If I did I might have been able to get in the first post!

    Now I have to decide if I want to wait to get a laptop with one of these on it. Damn.

  36. New Toy! New Toy! Gimmie one!!!! by doublem · · Score: 1
    I keep hitting refresh on the Transmeta homepage, waiting for it to load the new content. This sounds amazing. If this thing is what it sounds like we'll have the ability to run all of the x86 Linux apps from a low cost handheld! The Star Trek data pad melds with the modern PC!!!! All I need to know is this: Will it run Quake?????

    This also reminds me of the olden days when BASIC was built into the ROM, and would load if no other OS was available. Anyone else ever get the BASIC ROM Not Found error from a Phoenix bios? (Ah, the memories!)

    --
    "Live Free or Die." Don't like it? Then keep out of the USA
    1. Re:New Toy! New Toy! Gimmie one!!!! by Inoshiro · · Score: 2

      " Will it run Quake????? "

      Ahh, the penultimate question of the ages -- will it run Quake.

      Itchy reply finger, meet my itchy trigger finger..

      Even with the site and 'cast asside, this is a laptop processor. Nothing says Quake can't run on a laptop, unless the laptop is an iBook or something. But the Cursoe proc emulates x86 ISA... So the answer is a simple 0x44 0x75 0x68..


      Thanks for your time.
      ---

      --
      --
      Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
  37. mac os by DarkClown · · Score: 1

    This doesn't really address what you're talking about, but listening to the engineering briefing, a guy just asked about where the mac os fits/will fit in, and it was stated that code morphing could and possibly would be developed that would allow the chip to handle mac instruction sets. so, i guess this means the same thing for sparc, arm, alpha, etc..
    that's what i want to see. screw emulators, i want to see it apps running side by side that are meant for different architectures.
    vmware for macintosh? haven't seen that yet - does it exist?

    1. Re:mac os by mcrandello · · Score: 1

      If you mean the first question, the guy also asked if the morphing S/W would not be released open source, however that they may change that...meaning even more choice of virtual architectures if/when they do?


      mcrandello@my-deja.com
      rschaar{at}pegasus.cc.ucf.edu if it's important.

    2. Re:mac os by puetzk · · Score: 2

      It's called mac-on-linux and ios under GPL. See www.ibrium.se/linux for details. You'll need a copy of MacOS and a PowerPC linux system. (just like vmware needs windows and an x86). Of course, on this new toy, cpu emulation of the G3 might be practical if someone will write it...

      --
      The Matrix is going down for reboot now! Stopping reality: OK. The system is halted.
  38. Internet Appliances... by Anonymous Coward · · Score: 0

    Amiga announcing that they are going to use Linux and concentrate on "internet appliances".. Be announcing that is gearing toward "internet appliances".. Transmeta designs a chip for "internet appliances".. Is it all connected?

  39. Code morphing - this is not that new by Anonymous Coward · · Score: 0

    I hate to say it, but this is the idea DEC had with FX/32. This is not new, maybe the hardware part?

    DEC has always had great technology, not always good at getting it out there

    1. Re:Code morphing - this is not that new by Anonymous Coward · · Score: 1

      Both are compilers of sorts but Code Morphing takes place at level much lower than FX/32 - it is part of the system (I'm sure the Code Morphing software rests in ROM). FX/32 is a program on your operating system. You can't do a ps (from unix) and see Code Morphing - system software doesn't even know it exists. Code morphing is like the reordering / microcoding stages of the CPU - it exists at that level. So while they have similarities, they really are different beasts. -Slothmonster

    2. Re:Code morphing - this is not that new by Anonymous Coward · · Score: 0

      You can't do a ps (from unix) and see Code Morphing - system software doesn't even know it exists.

      Great--one more hiding place for the NSA to spy on my pr0n (and you thought the PIII ID was bad).

  40. Cool by __shad__ · · Score: 1

    I am glad I didn't buy myself a new labtop... I am going to wait for this. Bill Gates Looks cool as a borg Flames will be sent to /dev/null

  41. Is there a Multi-processor MB in the works? by GMontag · · Score: 1

    Sounds like this could be the beginning of the practical multiprocessor laptop market.

    If the chips live up to their hype, I will not have to wait for Wine to run everything I like AND everything the folks at work want.

    I want 4!!!

    --
    Eve Fairbanks says I drive a hybrid!LOL
  42. Long live Linus! by Anonymous Coward · · Score: 0

    Just think of it. This is the closest hardware development is going to get to open source, closer than the opengl and glide development going on right now. Everything software controlled and running Linux! wonderful, wundervoll....

  43. Screw the emulation by Anonymous Coward · · Score: 0

    I want to code on this sucker native! When intel are just releasing their first 64 bit chip, Transmeta are releasing the 128 bit chip.

    That's an address space of 3.4E38 bytes - enough to hold the complete DNA sequences of everybody on the planet around 3E15 times.

    That will be the largest address space we'll need for a very, very long time indeed.

  44. That was all? by um...+Lucas · · Score: 2

    Yes, it sounds like a kick ass product... I listened to the conference until my connection konked out around 1:05. But I was still hoping for more.

    According to all the rumors I'd heard, this chip would be able to load many different instruction sets (PowerPC, SPARC, etc...) and always pretty much be running in a native mode.

    Of all the complaints I've heard about x86 processors across the years, the one that sticks out the most is that they're a pain to develop for. This new chip does nothing to alleviate that.

    In a way, transmeta's become like Be. Be originally set out to take over the world (or at least, the Mac OS market place), moved onto x86 and realized that can't beat Microsoft out of the market place, so they might as well try to co-habitate as well as they can. Transmeta has a really cool product, but has also realized that they can't really push intel out of the market place, so they might as well just aim for intel compatibility...

    Too bad it sounds like they expect nothing to be written in Crusoe's native language... There must be some speed improvement that could be gained, if say, Crusoe's achieved 40% market share in the notebook market, that would make it worthwhile for developers to create Crusoe ports.

    1. Re:That was all? by GrayArea · · Score: 1

      Too bad it sounds like they expect nothing to be written in Crusoe's native language... There must be some speed improvement that could be gained, if say, Crusoe's achieved 40% market share in the notebook market, that would make it worthwhile for developers to create Crusoe ports.

      I don't think ripping out the software part of Crusoe would result in a performance gain. From what I have seen in their patents and today's presentation, the software component seems essential to higher performance than a hardware-only solution. In fact, it might be considerably harder to natively port an operating system to Crusoe, though I don't have any information to confirm this. Any way, the dynamic compilation and special power management features are all controlled by the software, and these are the main reasons that make Crusoe attractive.

      Just my two cents,
      --
      "The deluded are always filled with absolutes. The rest of us have to live with ambiguity." - Aristoi, Walter Jon Willia
    2. Re:That was all? by um...+Lucas · · Score: 2

      Well, what about a Mixed mode solution, akin the the MacOS? Programs could run in translated mode at regular speed, and portions of them could be translated to the full VLIW instruction set. It semems that Crusoe is already dynamically recompiling x86 to it's own instruction set, so, with smarter compiliers coming (which intel is funding), wouldn't it only add to the performance?

      After all, if intel's tools are indeed open-source, then it shouldn't be hard to add support for other VLIW instruction sets to the compiler. Of course, that would all hinge on Transmeta releasing the instruction set...

    3. Re:That was all? by David+Greene · · Score: 1
      What you're essentially saying is that the ISA doesn't matter.

      Well, it does. I don't know what Crusoe does with memory references, but I can almost guarantee you it can't map all of them to registers. The x86 is severely limited in the register area, and this really, really hurts compiled code. Spill code is a major problem on that architecture, not to mention all of the implicit dependencies between instructions and the general non-orthogonal nature of the instruction set. All of these things contribute to poor code generation on the compiler's part.

      Even if it could map everything to registers (it might actually be able to handle most references to the stack), there's still the problem of instruction fetch. Getting instructions into the core is a big, big bottleneck. Having your program expressed in fewer instructions means you save I-Cache and instruction window/reservation station space. It means the hardware can "see" more of the program at once. Now the processors introduced today are not O-O-O, but the I-Cache still matters.

      More registers usually means fewer instructions, as the "overhead" memory operations are not necessary. None of Crusoe's code morphing can help (static) instruction count.

      What piques my interest in something like Crusoe is that now the compiler can generate its own ISA! This has been done by some researchers before, especially in the area of code compression where a custom ISA is designed to compress well for a particular program.

      Think about this for a minute: the developer profiles a program to find some representative behavior. The compiler takes this information and figures out the best code it can generate without ISA restrictions. Need a vector add? Make one up! The compiler must still be aware of hardware issues (number of function units, etc.), but this removes one layer of restriction.

      So the compiler generates object code as well as the firmware needed to run it. Linux (or your favorite OS) is modified so that the loader loads the object code into main memory, loads the firmware into flash memory (or wherever it goes) and yells, "Go!"

      Now if we could only add registers... :)

      --

      --

    4. Re:That was all? by Inoshiro · · Score: 2

      "Too bad it sounds like they expect nothing to be written in Crusoe's native language... There must be some speed improvement that could be gained, if say, Crusoe's achieved 40% market share in the notebook market, that would make it worthwhile for developers to create Crusoe ports."

      I can't believe the number of navel gazing, performance lovers who keep saying this (or maybe I can, it's the same crowd that crows about K7 native RISC stuff).

      My reply to that line of thought (the tin god of performance over all else):
      "If you want performance, don't use userland apps -- they use bloated things like Java, and C++. Implement it in kernel space. Oh, and don't use C, use ASM. C has too much overhead. If you really want space, just write the specific task you want into the boot sector the floppy or HD. Of course, since it's just one task, you can dump uneeded things like interface libraries or a BIOS...."

      And the reason why:
      You lose flexibility when you remove the translator:
      1) You now must recompile all apps.
      2) You can't fix the any erratum in software any more
      3) You are stuck with that VLIW core, instead of having a core you can extend up the ying-yang.
      4) You can't profile the program running as it runs, and get speed boosts or power consumption benefits anymore.

      The purpose is not performance. You might get hard thinking about a 1Ghz K7, but the fact is that the ram, HD, bus speed, and other parts of the computer are no where near the 1Ghz line. You're getting increasingly less of a return as you go up the Mhz ladder. The purpose of the Cursoe is flexibility. Now you have a 400Mhz proc which can emulate any other proc around a modular core, and it can profile apps, and it can live without a fan.. It is the perfect processor for embeded things like smart houses, and it does make the Palm's Dragon Ball look anemic in terms of performance -- you should like that part. The best thing is, I can also run it on a desktop with the "slow" peripherals, and it's still bitchying fast, and now allows software overclocking (as power consumption is taken care of). So even if they aren't targetting it for it, it also has the potential to be the fast thing running on electrons -- and that's because it is flexible, not performance oriented.
      ---

      --
      --
      Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
    5. Re:That was all? by Anonymous Coward · · Score: 0

      Why can't the software morphing layer morph your old code to the new code base?

    6. Re:That was all? by Inoshiro · · Score: 1

      "Why can't the software morphing layer morph your old code to the new code base?"

      That's what is does do, at runtime. It doesn't morph non-running code because it's profiling to find the best performance/power usage ratio.
      ---

      --
      --
      Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
    7. Re:That was all? by stripes · · Score: 1

      I mostly agree. Here are my few points of contention:

      • A CPU bound process in user space, like POVray isn't at a mesurably diffrent speed from a in-kernel "process", because it makes almost no U<->K transitions.
      • The Cursoe makes the DragonBall look anemic in terms of performance. It is about 20 times faster then the DragonBallVX. However it is also 15 times the price, and fails to include a LCD controler, and many of the other intgrated components on the DragonBall. Then again the ColdFire has been able to exevute (almost exactly) the same instruction set as the DragonBall at 200Mhz, and it costs more like $50.
      • It can execute x86 code. Great, we have tons of x86 code that needs executing. In thery it could execute PowerPC or SPARC code as well (as long as doing the FP internally at 80bits, and having a MMU with x86-style page attributes doesn't get in the way too much, oh, and remember it will probbably have to be a 32bit SPARC or PPC, but a very fast one). In practice we have no idea if we will see anything other then the x86 morphing engine. It would have been nice to have a more sane ISA in addtion to the x86 one. There is no reason it would have to be the natiave VLIW one (and in fact the natiave VLIW one seems like a bad idea, as it is totally not what comercial compilers are good at targeting, and differes between all of Transmetas current products).
      • You say you can get a 1Ghz K7, but it is hamstrung by the RAM, drive, and other parts. Well, for some things, yes, totally. for other things you get good cache use, and your only bottleneck is RAM access speed. Weather prediction, Raytracing, Bruteforce crypto keyspace search, maybe SETTI@Home. And the RAM thing is getting better all the time (in part just by incresing the size and number of levels in the CPU cache system).
  45. So what else can it do? SPARC, Alpha, native? by peter · · Score: 1

    Everyone is making a big deal about the x86 compatibility. What I want to know is how well it runs code for other processors (which is the whole point of doing everything in software, right?) It would be cool if it could run SPARC or Alpha binaries, because those chips (esp. Alpha) have much better floating point units than x86 processors. (yes I know about K7s. they're still limited by the x86 arch.). Will the Crusoe have a PPC mode, so I can run d.net's G4 Altivec cruncher? hehehe. :-)

    Besides the speed factor, x86s use 80bit max floating point numbers. SPARCs can use 128bit floating point. (You get this with gcc's long double type.)

    Even more interesting, what is the Crusoe's "native" instruction format like? Is there a gcc that generates native code for it that runs faster? How many registers does it have, or does it even have registers? (maybe just a bit of normal RAM which is as fast as registers, but even less special than the general purpose regs on most RISC cpus.)

    I guess all of this will be answered in about 1 minute because the transmeta webpage is supposed to go live around now!
    #define X(x,y) x##y

    --
    #define X(x,y) x##y
    Peter Cordes ; e-mail: X(peter@cordes , .ca)
  46. The Real fun is in the possibilites.. by dougman · · Score: 3

    It sounds to me like the real fun is in speculating on what the Crusoe COULD do, or make possible, once various CPU instruction sets are implemented.

    For example, a Motorola G4 instruction set software piece could pave the way for someone to sell a handheld/mobile Mac!

    Amiga faithful could potentially see a handheld Amiga with a 68000 instruction set component!

    Heck, now that I think of it, arcade games these days use various RISC processors, imagine going to an arcade, and renting the use of a handheld arcade game!

    Fascinating stuff, I have to say. Fascinating.

    1. Re:The Real fun is in the possibilites.. by Anonymous Coward · · Score: 0

      Amiga faithful these days would want a 68060 instruction set, at least, if not PowerPC.

    2. Re:The Real fun is in the possibilites.. by Eidolon · · Score: 1

      A mobile Mac... is called a PowerBook, which until something actually ships with a Transmeta CPU in it, is the best-performing, coolest-running portable with longer battery life than pretty much anything else. A 300 MHz G3 consumes only about 2 or 3 watts of current for typical use... not as low as StrongArm or Transmeta, but a lot less than x86 hardware. As for temperature... the G3 in front of me usually stays around 31 C, as high as 35 C if I really push the system.

      Paving the way for a mobile Mac, indeed. ;-)

  47. Transmeta Could Make a Bundle Off Slashdot Alone! by Jack+William+Bell · · Score: 2

    Transmeta could make a bundle by selling cheap Crusoe(tm) development systems built on a PCI card. You know we would all get one if they cost

    Or, I suppose, they could just create a Crusoe(tm) system emulator program that ran on x86 architectures that would then emulate x86 architectures :-) Of course someone would immediately try to run the emulator in the emulator in the emulator in the...

    How come they still haven't updated their web site? I was expecting some change at least...

    Jack

    --
    - -
    Are you an SF Fan? Are you a Tru-Fan?
  48. I'm drooling now by (void*) · · Score: 1

    Well I'm pretty amazed that a chip of that speed can take only 1 W of power to run. Forget the mobile market. If this chip does what it is touted to do, having it on the desktop would save tons of power in itself.

  49. So what does Crusoe mean? by Anonymous Coward · · Score: 0

    What's the significance (if any) of the name Crusoe?

    My favorite bit of speculation before today's anouncement was Crusoe being an anagram for source- i.e out of order source.


    -ec

    1. Re:So what does Crusoe mean? by alphaparticle · · Score: 1

      It's Mr. Robinson....

  50. PC/104 / cardPC form factor board plans? by nosferatu · · Score: 1
    Anyone know of manufacturers announcing PC/104 or cardPC form factor single board computers (SBCs) based on the transmeta released CPU? I am involved in wearable computing and this would be a very nice solution.

  51. no linux on the 700Mhz version? by rogerbo · · Score: 3

    another interesting point...

    it was mentioned near the end that the vliw instructions AND the code morphing software on the two chips are different and NOT binary compatible.

    So possibly no Linux on the 700Mhz version?
    (they said it's optimised for 16 bit x86 instructions)

    1. Re:no linux on the 700Mhz version? by Edward+Kmett · · Score: 1
      From listening to the presentation, the only real changes for Linux are adding support for the LongRun power management software. It looks like Linux will still be running on it in x86 compatibility mode because immediately after bootup it shuts down access to the code morphing section. Basically it leaves the doors open long enough to download software bugpatches to the chip itself, then shuts down to avoid virus writers having a field day.

      So Linux should run just fine. The binary compatibility issues are only because the underlying VLIW architecture could change out to anything, so the only thing that has to be recompiled is their code morphing software.

      For that matter given the structure of the thing they could completely redesign the entire architecture of the system, risc to cisc, add dozens of registers, change every single internal opcode, shorten the bus, and as long as they could recompile their code morphing software to run on the new target they'll get binary compatibility with x86.

      Linux doesn't run in microcode. It would be a cool effort to port to it, but from the comments made they will not be documenting that layer of the chipset at all.

      --
      Sanity is a sandbox. I prefer the swings.
    2. Re:no linux on the 700Mhz version? by Anonymous Coward · · Score: 0

      Isn't Paul Allen mixed up in this somehow?

    3. Re:no linux on the 700Mhz version? by Anonymous Coward · · Score: 1

      The two chips' VLIW instruction sets are not compatible with each other. However, both chips run x86 code. This means any x86 OS including Linux, BSD, BeOS, Windows etc etc will run on these chips. My understanding is that the 5400 contains some extra optimizations to allow it to run Windows faster. The ZDnet article is very misleading in that it suggests that the one chip only runs Windows while the other only runs Linux...this is false! Yeroc

    4. Re:no linux on the 700Mhz version? by iceaxe · · Score: 1

      Actually, all that the "non-compatible" bit means is that you can't run the "code-morphing" software for one chip on the other. The emulated x86 instruction set is the same, so your binaries should work on either.

      So Linux will work quite readily on either chip.

      The 16 bit business means that old worn out windows and DOS binaries will work better on the faster chip than they do on the slower one.


      Cheers-
      Iceaxe

      --
      WALSTIB!
    5. Re:no linux on the 700Mhz version? by MindStalker · · Score: 2

      Well the 400MHZ version comes with linux, the 700MHZ version comes with windows. The 700MHZ supports 16 bit better, they 400 doesn't. 16 Bit support slows down a system. the 700MHZ runs windows perfectly and is exactly x86 compatible. Anything that is 100% x86 compatible can run x86 linux. But I'd be willing to bet that on linux the 700MHZ with 16 bit support runs exactly the same speed as the 400MHZ without 16 bit optimization as linux doesn't need this. (I assume both support 16 bit, just one is optimized for it, ruducing its speed/MHZ.)

    6. Re:no linux on the 700Mhz version? by Anonymous Coward · · Score: 1

      Of course Linux runs on the 700MHz version, the TM54xx chip. During the presentation, Linus Torvalds played a short game of Quake against Dave Taylor (one of Quake's co-authors). They both played on identical laptop machines built with the 700MHz chip (actually only running at 667MHz, but close enuff). Dave's machine was running Win9x and Linus' machine was running Linux, naturally. Linus got beat quite badly, he got zapped 3 times in a few seconds, but the smoothness and fluidity of frame rates of the actual game performance was virtually identical on both machines.

    7. Re:no linux on the 700Mhz version? by dizzydogg · · Score: 1

      I may be wrong, but I doubt that linux running on the 700Mhz chip will run at the same speed as on the 400Mhz. The chip only translates those codes that are used, so if there is no 16 bit operations, it will not do anything in 16 bit. 16bit support itself does not slow down a system, it is 16 bit instructions that slow down a system because they are less flexible. And besides, if switching from 32bit to a 16 bit operating system brought your computer from a 400Mhz to a 700Mhz, we would all be running dos and windows 3.11.

    8. Re:no linux on the 700Mhz version? by dizzydogg · · Score: 1

      I may be wrong, but I doubt that linux running on the 700Mhz chip will run at the same speed as on the 400Mhz. The chip only translates those codes that are used, so if there is no 16 bit operations, it will not do anything in 16 bit. 16bit support itself does not slow down a system, it is 16 bit instructions that slow down a system because they are less flexible. And besides, if optimizing for 16bit brought your computer from a 400Mhz to a 700Mhz, intel would have optimized their chips for 16bit and windows would be almost twice as fast(PIII 1400Mhz anybody?). The 16bit optimizations are just designed so that the parts of windows that are still 16bit, and hence slower, will be less of a bottleneck. It might cause a slight decrease in speed for a fully 32bit operating system, but linux should still far outperform windows.

    9. Re:no linux on the 700Mhz version? by Zer0Rage · · Score: 1

      Someone correct me if I am wrong, but one point was the ability to upgrade the instruction set w/ a software upgrade. This means that all the opcodes and whatnot are not stored on the chip so what would stop anyone from just taking the instruction set from the tm3120 and load them up to the tm5400. So Linux on a 700mhz sounds very possible.

      ZeroRage

    10. Re:no linux on the 700Mhz version? by Zer0Rage · · Score: 1

      Can you tell me what video feed your were watching, the one I saw was about 24 blocks that resembled Quake, granted it looked to run fine, I would say it was fluid, those blocks were pretty harsh.

    11. Re:no linux on the 700Mhz version? by alphaparticle · · Score: 1

      well, I was there on site, the graphic quality is VERY poor to give the full fps game experience. This is due to the limit of graphic unit onchip.

    12. Re:no linux on the 700Mhz version? by Guy+Harris · · Score: 2
      Someone correct me if I am wrong, but one point was the ability to upgrade the instruction set w/ a software upgrade. This means that all the opcodes and whatnot are not stored on the chip so what would stop anyone from just taking the instruction set from the tm3120 and load them up to the tm5400.

      You're wrong about what that ability means.

      The native instruction set of the processor is hard-wired.

      However, the only code in that instruction set is

      1. the binary-to-binary translation code;
      2. code generated by that binary-to-binary translation code

      "Upgrading the instruction set" would change the instruction set for OSes, applications, etc. that the machine would be willing to run - and that instruction set is the same for both chips; it's x86.

      So Linux on a 700mhz sounds very possible.

      Of course it is - Linux does exist for x86 processors. :-) They quite explicitly say that Linux will run on the TM5400 (the faster of the two machines, that being the "700 MHz version" - the Crusoe processor family page says quite explicitly:

      The TM5400 is compatible with the complete range of x86-based operating systems. This includes all versions of Linux, as well as Microsoft's popular Windows 98, Windows NT, and Windows 2000 operating systems.
    13. Re:no linux on the 700Mhz version? by Guy+Harris · · Score: 2
      Isn't Paul Allen mixed up in this somehow?

      Yes, he's one of the investors. The mere fact that one of the founders of Microsoft invested in them doesn't ipso facto mean that they will make one of their processors incapable of running OSes not from Microsoft; given that the chip presents an x86-compatible interface to non-Transmeta code running on it, the TM5400, as Transmeta said on their Web site, "...is compatible with the complete range of x86-based operating systems. This includes all versions of Linux, as well as Microsoft's popular Windows 98, Windows NT, and Windows 2000 operating systems."

  52. Long live Linus! by Anonymous Coward · · Score: 0

    Just think of it. This is the closest hardware development is going to get to open source, closer than the opengl development going on right now. Everything software controlled and running Linux! wonderful, wundervoll....thanks a lot Transmeta, but I really hope the release the code...

  53. I assume the "windows chip" runs Linux by SurfsUp · · Score: 2

    In all the reports I've seen so far the 5400 has been described as a "windows chip", and nothing is said about Linux. Naturally I assume that anything that can run Windows can run Linux too.

    --
    Life's a bitch but somebody's gotta do it.
    1. Re:I assume the "windows chip" runs Linux by Ralph+Wiggam · · Score: 1

      I've seen many people on /. and most of the major media outlets describing the two chips as a "400 MHz Linux version" and a "700 MHz Windows version". I could only get an audio feed of the press conference, but that's not what I understood at all.

      My understanding was that the 400 MHz chip (I forget the number they gave it), was designed for very small handheld devices. Maybe something the size of a PDA, but running x86 code and very integrated with your desktop. I didn't see the device they were using for the demo, but it was running Linux. I'm fairly sure that it COULD have run Windows. Someone correct me if I'm wrong.

      The TM5400 is aimed at a more traditional "notebook" computer. It consumes a little more power and costs quite a bit more (.18 micron is pricey). They mentioned Windows a lot in connection with it, but didn't Linus play Quake on a TM5400 machine running Linux?

      Sidenote: Was anyone else VERY impressed by the credentials of the people they brought out?

      -B

  54. API correspondent by Volatile_Memory · · Score: 2
    Did anyone watch the Webcast? The Associated Press goon didn't seem to be paying any attention whatsoever.

    In the Q&A he asked, "Since Linus helped to develop the chip, it is Linux-based. Will it also run Windows?"

    This, after an hour of presentations, including running several demos of Windows apps (including Quake), references to the fact that they have tested 30 OSs on it, and a myriad of references to Windows specifically.

    The "mainstream" press is missing the point. I'll bet they label this an Intel-clone from Linus and the public will yawn.

    --

    /**
    I have a "Zero Policy" tolerance.
    */

  55. Relative performance? by Outland+Traveller · · Score: 5

    I appologize if this has already been addressed. Alas, my connection to slashdot is poor and it took forever to bring up the reply page.

    I see that this processor comes in two versions with different Mhz, but what does that actually mean in real world performance? What is a 700Mhz or 400Mhz Crusoe chip equivalent to? As we all
    know, the Mhz rating does not mean everything.

    -OT (bogomips for everyone!)

    1. Re:Relative performance? by stienman · · Score: 1

      Well, as a little peep at the capabilities:

      The intel line is a 32 bit max line...
      Crusoe is a 128bit

      This means you could compare (roughly) a PIII 800 with a TM5400 200.

      BONUS! I just got on the transmeta web site!!

      -Adam

      All power corrupts, but we need electricity...

    2. Re:Relative performance? by rcromwell2 · · Score: 1


      Probably not as good as a Cyrix. :) Seriously, they have to be making a tradeoff somewhere. It sounds like nothing more than a CPU with a simple instruction set using something similar to a Java Just-In-Time compiler to translate x86 code to its native format.

      Don't trade in your PentiumIII/Athlon yet.

      The only killer app I see for Crusoe in the short term is if they can get it to emulate a Motorola Dragonball processor, but at great speeds, so that Palm Pilot's have the cpu power to do interesting things like play MP3's.

      -Ray

    3. Re:Relative performance? by Scrybe · · Score: 5

      During the Q&A session someone asked how the performance on the crusoe compared with that of the P!!!. After a bunch of normal disclaimers about different system configs, beta silicon, and beta x86 decode software they finally equated a TM5400 @ 667Mhz to a p!!! @ 500. Not bad for code morphing in a beta state plus all the potential power savings.

      --

      <This .sig left intentionally blank>

    4. Re:Relative performance? by rogerbo · · Score: 1

      HUH? You don't consider a palmtop with a colour screen, that runs x86 linux, netscape, plays mp3's with xmms and has a 24 hour battery life or longer a killer app?

      Why do they need to emulate the dragonball? How long do you think it will take for someone to write some nice date/calender/todo apps and a hotsync function for their linux webpad?

      I'd much rather have a crusoe linux webpad than a beefed up pilot.

    5. Re:Relative performance? by ecampbel · · Score: 3

      You're kidding right? Remember that a Transmeta processor needs to translate EVERY instruction to an equivalent sent of 128bit instructions. This carries a significant performance penalty. Also, the Transmeta chip doesn't have any of the multimedia enhancements that Petium's, Athlon's, and G4's carry with it. Software does not beat hardware when doing extremely optimized vector based instructions that can run in parallel inside the native processor.

      --

      Sig goes here
    6. Re:Relative performance? by Dominic_Mazzoni · · Score: 2

      I could be wrong, but I got the impression that they cache the instructions once they're translated, thus allowing short loops that are executed many times to operate just as fast as if they were native. This is the same trick used by Java Just-In-Time compilers.

      There is also no reason to believe that they couldn't add support later for MMX or other such instructions natively, but keep in mind that very few normal programs take advantage of these anyway.

    7. Re:Relative performance? by Sharki · · Score: 1

      I am pretty sure that relative performance of their 400-700Mhz CPU (which btw does not exist yet, they are working on the prototype... Hmm.. paper CPU, how nice...) Will compare to at most a pentium 160Mhz if that. "Caching blocks of x86 instructions" - WHAT? In order to determine if a fetched "block" is a hit or miss they need to compare this entire "block" to contents of their "translation table" This comparison already increases this poor "block"'s processing time by number of entries in this "translation table"... Please... Also these transmeta guys openly lie about Intel CPUs in their white paper saying that intel translates x86 instructions before executing them. This is true for the 16bit x86 instructions, but not for 32bit... Or do they mean that they only emulate pure x86? Hmmm... I guess it's back to the DOS/win3.1 days for them then... Anyway this whole thing is a bull, since when do emulators run things better or fastern then the real thing? Oh well, I guess slashdotters will be content with a cpu run on water as long as Linus had his hands in it...

    8. Re:Relative performance? by ecampbel · · Score: 3

      I do realize that Transmeta does an excellent job in translating x86 instructions. Probably the best job that has ever been done. But they are still emulating the x86 instruction set. Even with all their fancy technology a 700mhz processor will only perform, at best equivalent, to a 500mhz Pentium. Java's developers have spent years trying to optimize their JIT, and it still is a lot slower than running native compiled C code.

      Their emulation tricks will not work on MMX or other instructions nearly so well. Curose can't make up the time it takes to emulate the instruction set by having better hardware. MMX and the G4 vector processing units utilize many of the same tricks that the entire Cursoe processor utilizes such as long word instructions, parallel processing and other optimization techniques. Also, the hardware of these multimedia units are explicitly optimized to process these special instructions. This means to emulate these instructions would drastically slow down the chip.

      --

      Sig goes here
    9. Re:Relative performance? by drivers · · Score: 1

      Internally, Intel CPUs convert the CISC instructions to some kind of RISC instructions. Great for speed (pipelining, etc.), but in the words of Transmeta, they are doing it by throwing transistors and watts at the problem.

    10. Re:Relative performance? by sirket · · Score: 1

      You need to get your facts straight. All CPU's which came after the Pentium (ie PPro, P][, etc) have done instruction translation. All of these processors have a semi RISC core which they translate x86 instructions to in order to execute them.


      And just because they are doing translation, does not mean they are automatically slower. If the core is 5x faster than an equivalent P!!! core then the processors can still be faster despite doing translations. As you should well know, MHz is not everything.


      -sirket

    11. Re:Relative performance? by Esperandi · · Score: 1

      Heh, you didn't read the whitepaper, right? Didn't think so. Where to start?

      First, the lying. You expose your ignorance (that means lack of knowledge, not stupidity) of how a computer works here. You see, you got the processor, then above the processor you have the microcode, then you have the stuff above it. The microcode is where you send an encoded opcode like ADD AX, 0004h or something similar. The microcode then translates this to something the processor can use. In the case of an ADD that involves memory locations, it translates that into code that moves the stuff from memory into a register, does the add, etc. So pretty much all instructions in a CPU are translated down. Even moreso with things like MMX and Katmai SIMD instructions. So they're not lying, you just don't get their meaning.

      Now, as to cacheing. In that cache, the code is also optimized. Also, that cache is not a cache of x86instructions. It is a cache of Crusoe "molecules". Molecules consist of 4 atoms which are executed in parallel in one clock cycle. During the optimization, the instructions are reordered so that things both execute faster and execute lesser code. The simple example they gave in their whitepaper (which I urge you to read) was when there were 4 x86 instructions. Those instructions would get translated to many micro-ops and sent thru a rescheduler every time they are executed. In a Crusoe, they are translated and optimized once, then every time they get executed they execute less code and execute it faster without re-scheduling. It ends up a lot faster.

      Think about this for performance: DVD playing (another example from the whitepaper). The P3 700 works at 700MHz decoding the DVD keeping a constant framerate. Once the first frame of video has displayed on a Crusoe, the instructions for the DVD decode process are cached and optimized. The Crusoe can clock down to 400MHz and drop its voltage consumption and keep exactly on line with the performance of the 700MHz P3. So no, the performance isn't that of a P120 in realworld use.

      Why does it fail at benchmarks then? Easy, benchmakrs don't repeat code. They do different things very quickly and do not allow any cacheing or optimization to be done. This _never_ happens in real life. Right now I'm typing in this textbox, and the code executing every single time I type a character would be caches and optimized and would be executing many times faster on a Crusoe than on an equivalent x86 machine. It would only slow down if I typed one character, and then clicked used several hundred different features in my browser in the space of a minute. That is never going to happen because I can't type or move the mouse that fast.

      Esperandi
      The Crusoe is the worlds first Real World Processor.
      Oh, and a little PS here... trust me, I am never one to jump when Linus says so, I haven't used Linux in a couple years and I don't plan on switching to it any time soon. I like hardware compatibility and the idea programmers get paid for their work.

    12. Re:Relative performance? by Ryan+Taylor · · Score: 1

      I read the other replies and I think they did a suitible job tearing you to shreds technically. I have something a bit more obvious to point out which they appear to have neglected: The CPU does exist, they have it working in both speeds, the slower one is available for shiping as of today, and they played quake on it today while live on ZDTV.

      If you're gunna be stupid, at least you could keep the edge out of your voice so as not to piss people off.

      Sincerely,

      Ryan Taylor

      --

    13. Re:Relative performance? by Anonymous Coward · · Score: 0

      is transmeta paying you? or are you just easily swayed by marketing?

      The faster model of transmeta CPU doesn't exist yet, so any speed claims are marketing "predictions."

      please try to think critically about WHO is telling you WHAT, and most importantly, WHY.

      until people not associated with transmeta have these, and have had enough time to experiment, any speed claims are suspect to the point of being worthless.

      See Hotspot for a realworld example of this.

    14. Re:Relative performance? by Buaku · · Score: 1
      First off, the 400 MHz chip does exist. It's available "now". Heck, they ran demo's with the thing. As for the 700 MHz chip, we'll see if they meet their expected release date, but that's no different from any other chip company.

      As to performance, the big story here is that it runs on 1/35th the power of a Pentium. If it runs a generation slower, that's to be expected. It's new technology, and thus it hasn't matured yet. Furthermore, I think you're wrong about it being as slow as a Pentium 160 or so. Time will tell, but if by some extremely long chance you are correct about that, then the company won't be around very long in any event.

    15. Re:Relative performance? by Anonymous Coward · · Score: 0

      Yeah, a Paper CPU. Ye of little faith have never seen what the semicon industry can do when they want to. Prototype most likely means they have it at least simulated as hardware in some VHDL. Getting that layed out and into fab at IBM by this summer should be cake. Compare block contents? Crack is bad for you! If each block represented the VLIW bytecodes for a particular instruction or cohesive set of instructions you could tag it at the software level at compare that using a hashtable with minimal lookup times and free translation as long as you reference something not pushed out of the table for lack of reference. you'd be surprised how much locality of reference there is in modern machine code. Intel/AMD both provide RISC cores surrounded by logic to convert x86 to same. The "real thing" sucks a laptop battery dry in under 2 hours when tasked with any meaningful computation. This small hit in processing performance ( if there even is one after you consider the 128 bit and 700MHz specs ) is balanced or enhanced by the ability to scale processor power consumption to what the chip is actually requiring ( no hot logic dissipating heat when not in use ). you can run a notebook with a color display for a full day.

    16. Re:Relative performance? by GregWebb · · Score: 2

      ... and this still isn't going to boot especially fast or work nicely on that small screen.

      They seem to be focussing on webpads to a degree so far. I'd love to see what'd happen with a PDA, as that strikes me as a bigger market.

      Speaking as a Palm III owner, here :)

      Greg

      --

      Greg

      (Inside a nuclear plant)
      Aaaarrrggh! Run! The canary has mutated!

    17. Re:Relative performance? by alexalexis · · Score: 1

      After reading the benchmark papers and what-not, it appears that the current Transmeta chips run x86 software at about 80% the speed of an Intel chip at the same MHz.

      Something interesting to note, however, is that the TM die sizes are quite a bit smaller and much less complex than Intel's -- it's quite possible that they can crank 20% more MHz out of their silicon than Intel can. Also, keep in mind that their power usage is about 1/40th of what Intel uses -- they can further increase MHz by pumping up the voltage. Since their chips are being fabbed by IBM, I have no doubt they also have access to the lastest and greatest materials and equipment out there.

      All things considered, it *really* wouldn't surprise me if they unveiled a 1GHz+ CPU in the next year, geared for performance instead of efficiency. They have a *lot* going for them. I really look forward to testing their products when they hit the market!

    18. Re:Relative performance? by versus · · Score: 1
      Go read Ars Technica article about this issue.

      ...saying that intel translates x86 instructions before executing them. This is true for the 16bit x86 instructions, but not for 32bit...

      Native 32-bit instructions were executed ages ago, on x386/x486. Pentium already has RISC-like core and Pentium Pro/Pentium II added branch prediction.

      BTW, I wonder if there will be Native Crusoe Support for GCC? Despite all this Code Morphing Coolness, native Crusoe VLIW instructions optimized by real compiler should run much faster. GCC++ knows all about program workflow and Code Morphing will tune (rearrange predictions?) it at run-time.

      Furthermore, Transmeta can invent its own extentions like Intel's MMX and implement it real soon. Image tweaking the CPU to 3D card drivers! :-)

      IMHO this technology will beat Intel in the long run.

      --
      Brain is my second favorite organ.
  56. Proprietary hardware sucks by Anonymous Coward · · Score: 1

    Lack of the native ISA also makes reverse engineering their emulation software harder... of course a patent should provide all the details needed to implement that anyway :) (it hardly ever does)

    And if they worry about compatibility license the information, make it usable only for GPL software. With source backwards compatibility on the instruction level isnt much of a concern.

  57. transmeta home page. by Anonymous Coward · · Score: 0

    So why are there six toeprints in the image on their homepage? Is this another hidden message?

  58. www.transmeta.com goes live by god_of_the_machine · · Score: 1

    Its up now (12:05 PST) and it looks good. Though it is very, very, very slow! What... didn't they expect to get hits?

    --

    -rt-
    ** Evil Canadians are taking over the world. Learn about the conspiracy
  59. maybe... by Xtacy · · Score: 1

    they should have spent a bit more time on bandwidth and their server machine while everyone was hyping the product....11:59 PST, normal site, 12:00 PST wow, new site, click on one link and realize its been slashdotted....

    did they think they'd get a small number of hits?

    ugh

    1. Re:maybe... by Anonymous Coward · · Score: 0

      Yep... well, only page I've pulled so far.. to share with those slashed out: Transmeta believes in the future of mobile computing. The Crusoe processor was designed - from the ground up - to deliver true mobility. It runs cool so Crusoe processor-based devices can remain thin and lightweight, deliver high performance, and consume less battery power. Mobile computers with a Crusoe processor are designed to free computer users from the restrictions of desktop systems, without technical tradeoffs. In the near future, we will see many new products in the area of Mobile Internet Computing. Devices such as web pads, like the demonstration prototype shown below, will give users full wireless access to the Internet. Using x86-compatible processors will enable users to run the widest range of available browser plugins, opening up the whole range of multimedia content available on the Internet. Because the web pad shown below uses a Crusoe processor, it is thin, light, quiet, and has excellent battery lifetime. Watch this space in mid-2000 for news about the Crusoe processor in all-new lightweight mobile systems - from ultra-light three-pound PCs to web pads to handheld devices - from top manufacturers, each created especially with your mobile lifestyle in mind. If you would like to receive email notices about new products that feature the Crusoe processor, subscribe to Going Mobile News. -Tranq

    2. Re:maybe... by Dr+Fgets · · Score: 1

      gee do you work for transmeta?

      --
      Dr Fgets Strikes again!
  60. The Doc Sayz by Dr+Fgets · · Score: 0

    I'm not going to wet my pants over this, frankly, it doesn't sound all that amazing.
    Then again, I don't get excited when Linus takes a dump either.

    --
    Dr Fgets Strikes again!
  61. Win2K on my Palm V? by Anonymous Coward · · Score: 0

    Can u imagine gutting a palm V and dumping a crusoe into it and running Win2K? Why? Why not?

  62. Rain in the silver lining by Anonymous Coward · · Score: 1

    Technologically, the Crusoe is a marvellous little device. May be the catalyst to replace our desktops with slimline portable embedded devices.
    But there's a serious flaw I see with all this grandstanding and fanfare...
    It sounds to me like Transmeta is going to focus on their ability to emulate the x86 isn set atop a vliw processor. That's all well and good, but I thought they would focus that as a "feature", not a mainstay of their product. I am disappointed to see the archaic and outdated x86 architecture perpetuated by what I once thought to be a high-tech company on the bleeding edge. AXP, sparc or even MIPS has a better architecture than the x86. Their promise to not release the native vliw instruction set solidifies my opinion against them.

    1. Re:Rain in the silver lining by PhilipKDick · · Score: 1
      Hate posting "me too" replies but this is exactly what I'm doing.
      I really can't grasp how so many slashdot readers don't feel disappointed. I wonder how those replies would look like if Linus wasn't involved in that company. They offer no real edge in terms of performance and the power saving benefits seem dubious to say the least. Everyone realises that it's not just the processor that's power hugry in portable devices.

      Just like yourself I feel that resuccitating x86 is a big mistake. Though they claim they can emulate _any_ instruction set they seem to be suspiciously slanted towards x86. I wonder if they actually have any other set emulated successfully.

      Everyone must realise that what we saw was an interesting approach to cloning the Intel Pentium processor and nothing more.
      They are surely trying to catch the wave of incredible hype they managed to generate but the merits of Crusoe are not obvious to me at all. Can anyone explain what this huge wave of optimism is all about?

    2. Re:Rain in the silver lining by Inoshiro · · Score: 2

      "I thought they would focus that as a "feature", not a mainstay of their product."

      It is a feature. Did you not digest any of the info about the translation unit? It can emulate any ISA you care to specify.

      "I am disappointed to see the archaic and outdated x86 architecture perpetuated by what I once thought to be a high-tech company on the bleeding edge"

      I am disapointed you wasted time whinning about an ISA with a firm foothold in the world. Given an OS like Linux, you can swap processors as much as you want (ie: it's a commodity), but not everyone is using Linux.. In fact, of the 400 million (or so) Windows installations out there, I'm willing to bet a fair number of them are running on a processor that runs the x86 ISA at some point. A lot of other OSes also can run on it, like BeOS, Linux, *BSD, etc. If you don't like that fact, go buy a G4 or an Alpha, and support alternative processors. Whinning about it on Slashdot is like showing up at your local video store, and whinning about how they don't stock Beta, and why VHS is so much worse than Beta..

      "AXP, sparc or even MIPS has a better architecture than the x86."

      *Yawn*..
      ---

      --
      --
      Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
    3. Re:Rain in the silver lining by IntlHarvester · · Score: 1

      AXP, sparc or even MIPS has a better architecture than the x86

      Maybe, in academic theory, or if you write asm, but in the real world, a CPU architecture is only as good as the software that runs on it and how easy it is to integrate it into particular devices.

      Maybe you are thinking about your kick-ass SGI workstation or the humungo Sun E10K in the server room? But, in both those cases, the CPU is the *least* interesting thing about the hardware. It would be awful difficult for the Caruso to emulate SGI's memory bandwidth or Sun's swappable CPUs, and besides, what's the market for a handheld MIPS computer anyway? People don't buy these things for the CPUs, which is exactly why Sun and SGI are covering their bets with support for Intel Merced.

      It is kind of disappointing that the native ISA wasn't released, but x86 is the natural choice, given both the huge amount of software and ultra inexpensive support hardware. (Even Linux users which have a large range of platforms available to them by-in-large install binaries and run on commodity PC platforms)
      --

      --
      Business. Numbers. Money. People. Computer World.
    4. Re:Rain in the silver lining by McKing · · Score: 1

      The reason that they are focusing on the x86 compatibility is not to keep the x86 architecture alive. They made a point of emphasizing the fact that they want their OEM's to have real, working machines in the stores by summer, running all of the latest web software and browser plugins that most people are used to having. That means having a real, full version of Windows (or x86 Linux, yay!), instead of using a less predominant OS and having to get all of the software vendors to port to their platform. The ability to buy a palmtop and run the _exact_ same version of Office, Netscape, Quicken, etc... as on the desktop, instead of some crippled "Pocket" version (a la WinCE devices) from day one is the whole point of this. Similarly, if they wrote a new OS, they would still have to get vendors to port to it, and they would just be another handheld "platform" like Palm, WinCE, Magic Cap, Newton (remember these two?).

      Of course, their tech allows them to upgrade their CPUs to new instruction sets, so if they wanted to they could make a PowerPC, Sparc, or DEC Alpha (I just can't say Compaq Alpha), compatible machine, or come up with the ultimate instruction set, free af all legacy code and able to leap over tall buildings with a single asm instruction.

      The native VLIW set is not really the reason the CPU is so cool. It is the translation engine and the ability to optimize as it translates and adjust the power and speed settings optimally that makes this the CPU that will probably kill Intel in the laptop world. I just can't wait to get one of thesee things!

      --
      If only "common" sense was actually that common...
  63. Transmeta site... by yesthatguy · · Score: 1

    The Transmeta site is understandably slow now. With not only all of /. going to check it out, but many people reading ZDNet, MSNBC, and others, as well as other people in the know, it seems their server was not ready for the traffic...
    ---------------

    --
    Yes! That guy!
  64. MSNBC has the scoop! by Oshu · · Score: 1
    From MSNBC's article:

    Ditzel, a former engineer at Sun Microsystems, was integral in the development of the RISC processor, another competitor for Intel's x86 processor. RISC was faster than Intel's chips, but software designers never picked up on it, and it remained on the fringes.

    I sure hope the Crusoe does better than the RISC!

    --

    That damn time machine set me back 15 years!

  65. Re: Are you a moron?? by Anonymous Coward · · Score: 0

    My chip (Celeron) is running right now at 34 F, and that's plenty hot... That's 2 Degrees above freezing. Room temperature is already around 75 degrees average.. so just sitting there that'd be the temperature of your chip. With power flowing into it, the temperature rises.. even with a fan. The fan is blowing 75 Degree F air acrossed it, so the temp won't be below 75F. Unless you have some refridgerating system hooked up to drop the temperature to 34 (and in which case unless you had it setup right would cause water to condense on the chip and.. well you know how electronics and water work together). I believe you either ment 34C or you're just damn stupid.

  66. Mobile Linux by Matt2000 · · Score: 4

    So now that mobile Linux has been demoed to the world, when is Linus going to release the source? I didn't hear any mention of it during the webcast (although I had to leave part way through). And I guess the other question is, is this a kernel fork in progress, or is it a common kernel with what we've been seeing in development right now?

    Hotnutz.com

    --

    1. Re:Mobile Linux by magg · · Score: 2

      It's not a fork in the kernel tree. Crusoe only 'emulates' x86..

      It is no Mobil Linux it's just some patches for the kernel, made by Transmeta.

      They will be released 'soon'..
      Mvh
      --
      Magnus

      --
      magg
  67. Only relevant to closed source software!!! by Anonymous Coward · · Score: 0

    Well it is, theres other high mips/watt architectures wich are probably competetive. (especially considering the advanced process wich crusoe uses)

    The real innovation as you say is the code morphin, but IMO binary compatibility is becoming less of an issue in the future not more.

  68. Temp? by _GNU_ · · Score: 1

    Hmm, my pIII 450 overclocked to 600 runs at 65C, and my G3/350 (copper 750L) runs at 31C..

    How come Crusoe runs so hot compared to PPC?

    I guess IBM still have an edge on the old-style static register cpu's anyway :P

  69. Crusoe not for Palm devices? by twilight · · Score: 1

    According to this article: http://www.zdnet.co m/zdnn/stories/bursts/0,7407,2423907,00.html These processors are not designed for Palm devices but for other "mobile internet devices". Interesting, perhaps this is due to the existence of low power CPUs in the market already. It would be cool to have Mozilla running on my Palm though.

  70. TM 5400 only for Windows? by Jeff+Lightfoot · · Score: 1

    I keep reading that the TM 5400 will be on notebooks running Windows. Was it announced this way for marketing purposes or will it also have the ability to run Linux?

    1. Re:TM 5400 only for Windows? by Jeff+Lightfoot · · Score: 1
      I guess it was marketing. Their compatibility page has the following to say:


      When designing the Crusoe processors, Transmeta paid special attention to full
      compatibility. This endeavor was in fact made much easier by the unique
      approach of delegating the responsibility for compatibility to the Code
      Morphing software.

      All Crusoe processors are:

      Fully x86 compatible: they run x86 applications just like
      conventional x86 microprocessors.


      PC compatible: Crusoe processors already include portions of the
      traditional PC support chipset, and they run all popular PC operating
      systems.


      Internet compatible: Crusoe processors run the full wealth of
      Internet plugins written for the x86 - a world that is closed to most
      other low-power (non-x86-compatible) processors.

  71. low power linux devices by Anonymous Coward · · Score: 0
    Linux already runs on a fast, ultra-low-power chip.

    ARM Linux on the SA-1100

    While the SA-1100 doesn't have the cool code emulation bit, it does run at 200+ MHz at half a watt. The chip also includes vga, irda, and audio codec support. All of that for a half-watt, max.

    I still think the Transmeta announcement is cool, but not because of the low-watt linux claim.

    --Pat / zippy@cs.brandeis.edu

  72. sounds like a winmodem to me! by myc · · Score: 0

    instructions in software rather than built into the hardware? isn't that basically what a winmodem does? ;-) now all the non-windows, non-linux using crowd can complain there are no drivers for this hardware ;-)

    --
    NO CARRIER
    1. Re:sounds like a winmodem to me! by Frederic54 · · Score: 1

      héhé it's funny :) and that's right, why use windows (or whatever x86 OS) and a morph-emulator CPU? i can use these OS on a true x86 CPU!
      --
      BeRoute

      --
      "Science will win because it works." - Stephen Hawking
    2. Re:sounds like a winmodem to me! by sjames · · Score: 2

      instructions in software rather than built into the hardware? isn't that basically what a winmodem does? ;-)

      More or less, but it's not the same! :-)

      A winmodem takes what the chips on the modem should be doing and dumps the work off on the CPU instead. Normal modems also run in software, it's just that the software (firmware) runs on the modem hardware.

  73. Open hardware by Anonymous Coward · · Score: 0

    Why? Just because Linus works there?

    Intel has shown more support for Linux as Transmeta... Intel's hardware is open, seems like an easy choice from a principle POV.

    1. Re:Open hardware by bssea · · Score: 1

      Intel HAS to open it's instruction set.. otherwise apps won't be made for it. Transmeta doesn't because it CAN RUN the x86 instruction set.

      Intel isn't open by choice; it's by need.

      Besides, I won't buy anything that has a number engraved in it.

    2. Re:Open hardware by Anonymous Coward · · Score: 0

      Then you'd best rid yourself of your ethernet card, if you have one. That is, if you were referring to the P3 spn.

  74. Excitement with reservation by rjamestaylor · · Score: 3
    I agree with your excitement. I've wanted a truly portable Internet-capable device for a long time. the characteristics of this technology are not fully realized (which means there is much potential for growth). I have much hope for this technology (and company).

    My concerns are this:

    since it is maintaining compatability with x86 instruction sets, it will always follow Intel's lead (and require Intel to continue leading) mainstream chip technology.

    It will never run as fast as a native x86 chip would (because it must execute extra instructions). Of course, being smaller and independent to the hardware, these chips may be made significantly faster (clock-wise) than mainstream CISC/RISC chips and comparatively match performance. But not yet.

    No mention was made regarding the connection to the Internet...that was just assumed to be there. But I have yet to hear about any affordable and sufficiently fast connection via mobile unit... How will they address this, or will they just leave it up to other companies to solve this general problem?

    Transmeta touts Internet Compatibility, but the low end Internet appliances are specifically designed to work with Linux. However, Linux does not have a standards-conforming browser (i.e. IE) available until Mozilla is complete. Will Transmeta help push Mozilla to completion? The specific mantra was, "You have to run the cool site of the day" but many sites are becoming dependent on HTML 4, CSS2, DOM2, ECMAScript 2, etc., which, sorry, only are supported to any extent by IE5. How will Transmeta maintian "Internet Compatibility" with Linux-running machines?

    One correction to Hemos, however, Transmeta specifically said they are not targeting cell phones and Palm Pilot-type machines, but rather full-blown Internet compatible multimedia machines (which may be small, but no compromise on feature set).

    :-only kona in my cup-:
    :-robert taylor-:
    --
    -- @rjamestaylor on Ello
    1. Re:Excitement with reservation by Anonymous Coward · · Score: 1

      rjamestaylor wrote:
      It will never run as fast as a native x86 chip would (because it must execute extra instructions).


      I think that you're missing the other important feature of the instruction set morphing: on the fly optimization. So, theoretically, it may not 'run' as fast as a native x86 chip, but the native chip is stupid, while the Crusoe processor optimizes the code, thus getting equal or greater performance.

      Okay, so I don't 100% believe the optimization either, but we'll have to wait and see. I also can't believe they'd do a lot of work for something that isn't terribly effective.


      Zipwow

    2. Re:Excitement with reservation by Anonymous Coward · · Score: 1

      I think they looked for a nice way into the market and found one: with the x86 compatibility they have a huge bunch of software to choose from. This is much easier marketingwise than programming new applications for their processor and they can target the big market. And getting into the market is important, what use is a revolutionary new processor nobody buys?

      As long as they need to they can keep up with intel, implementing intel-instructions softwarewise. But they also have the choice to plug in their own add ons! It's important that Crusoe is not only designed to compete with the Pentium III, with the flexibility of the emulated instruction set it can be tailored to a lot of uses. This makes it possible to integrate a lot of devices into one that can be put to a lot of uses and even enhanced later, this might even answer the question how internet connection is provided, it's just some enhancement of the instruction set and some radio parts. While occupying the processor with additional tasks might not be the ideal approach to play quake in high res i think someone will come up with some uses for this, in the end it will we a question of it's price compared to the prices of the devices it replaces.

      This chip is far more flexible and it's just at the beginning, intel is pushing the PIII technology to the limit, Transmetas far more intelligent solution leaves a lot more possibilities for improvement and development, and it's already competing in a large market. Also note that the instruction set degenerates to just a 'standard' how executable bytecode should look. Now this standard is far more subject to design than before.

      As for Internet Compatibility, i've never seen that defined by the capabilities of the Web Browser, i thought it had to do with the capability to handle IP Packets. This still leaves the issue that Joe normal would find a windows device more attractive, but linux can be tailored to different uses far better than Windows, especially with Linus as a driving force, also it can be run on a slower Processor and still execute tasks with decent speed, note the MHz difference between the Linux- and the Windows- Crusoe.

    3. Re:Excitement with reservation by sgml4kids · · Score: 1

      It will never run as fast as a native x86 chip would (because it must execute extra instructions).


      My Pentium 133 emulates a 6502 at about a bazillion times the speed of a 1 MHz Apple ][+. Technically, there is no reason an emulator can't run faster than the original hardware. Keep in mind, also, that Intel is stretching an outdated technology to the limit -- look at the ridiculous measures Intel goes to keeping its Pentium chips cool. Transmeta doesn't have the obstacles that Intel has in maintaining backward compatibility. Plus the chip is a 128 bit process natively. How long before Intel can provide that?

      It's perfectly feasible that Transmeta could produce faster x86 "emulation" than the chips Intel provides. They also have a lot more room to innovate and provide new features. Plus the concept of a desktop PC is growing passe every day.

      ...I have yet to hear about any affordable and sufficiently fast connection via mobile unit



      Apple just came out with an iBook with an 11MHz wireless connection which is faster than my LAN at work. No it doesn't work any further than 150' or something like that, but it *is* useful and people *do* buy it.

      As for bemoaning that it's not running a "standards-conforming" browser -- well that's a classic example of not seeing the forest for the trees...


    4. Re:Excitement with reservation by Inoshiro · · Score: 4

      "since it is maintaining compatability with x86 instruction sets, it will always follow Intel's lead (and require Intel to continue leading) mainstream chip technology."

      This is like saying that because GM is making cars, they have to follow Ford's lead in how to make them. The x86 ISA is pretty much developed right now, with instructions for anything you'd care to do in silicon (or emulation). Since Intel has said that their next proc is VLIW, it looks like TransMeta's VLIW proc is leading them...

      "It will never run as fast as a native x86 chip would (because it must execute extra instructions). Of course, being smaller and independent to the hardware, these chips may be made significantly faster (clock-wise) than mainstream CISC/RISC chips and comparatively match performance. But not yet."

      The purpose was never performance, or the masturbatory RISC/CISC debate stuff. Its purpose is to provide good performance for insanely low levels of power consumption.

      "No mention was made regarding the connection to the Internet...that was just assumed to be there. But I have yet to hear about any affordable and sufficiently fast connection via mobile unit... How will they address this, or will they just leave it up to other companies to solve this general problem?"

      I think you are refering to the fact that the translation units can be upgraded via software. Software which comes over the internet :-)

      "However, Linux does not have a standards-conforming browser (i.e. IE) available until Mozilla is complete. "

      (sarcasm in good humour)
      Christ, what've I been running? Jeez, I guess it was just Bill himself pushing a cloaked IE through my net connection onto my desktop, as I can't be running something that's not been released..
      (/sarcasm in good humour)

      This will make a very nice "web panel" device, like the one Cyrix promoters were pushing.. I know I'd love to replace my 486 X terminal with a wireless laptop display :-)
      ---

      --
      --
      Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
    5. Re:Excitement with reservation by Buaku · · Score: 1
      since it is maintaining compatibility with x86 instruction sets, it will always follow Intel's lead (and require Intel to continue leading) mainstream chip technology

      Not so. They are emulating x86 off the bat because that is the current standard. However they can emulate other chip technologies with some coding. They mention that they should be able to emulate even a Mac.

      It will never run as fast as a native x86 chip would (because it must execute extra instructions). Of course, being smaller and independent to the hardware, these chips may be made significantly faster (clock-wise) than mainstream CISC/RISC chips and comparatively match performance. But not yet.

      Three answers here. First off, I don't think that one of the goals is super speed. One of the (short-term) goals is decent speed with near zero power consumption and heat production. They achieve this. Second off, if their optimization code is good, they may not lose much in the way of performance. Thirdly, the chip is built on 128-bit architecture. That's a lot of room for growth. They may soon surpass Intel and emulate them faster than they can run native because of that.

      No mention was made regarding the connection to the Internet...that was just assumed to be there. But I have yet to hear about any affordable and sufficiently fast connection via mobile unit... How will they address this, or will they just leave it up to other companies to solve this general problem?

      Ummm, it's called a modem or network card. Transmeta just produced a chip. They mentioned that they've already shipped laptops to OEM's with the Transmeta chip. Just use whatever hardware you have been using.

      Transmeta touts Internet Compatibility, but the low-end Internet appliances are specifically designed to work with Linux. However, Linux does not have a standards-conforming browser (i.e. IE) available until Mozilla is complete. Will Transmeta help push Mozilla to completion? The specific mantra was, "You have to run the cool site of the day" but many sites are becoming dependent on HTML 4, CSS2, DOM2, ECMAScript 2, etc., which, sorry, only are supported to any extent by IE5. How will Transmeta maintian "Internet Compatibility" with Linux-running machines?

      You can install Windows, Linux, FreeBSD, or whatever OS you want that's designed for an x86 machine. They just shipped it with a version of Linux for OEM's to demo it. Did you not read the part about how they ran Office 2000 on a computer with a Transmeta chip? Did you not read the section about how the Transmeta chip will blue-screen like any other chip if you run Windows on it? You can run whatever OS and whatever browser you like.

      It sounds like you got confused and thought that Transmeta produced an entirely different type of proprietary computer. They didn't do this. They aren't producing a new computer a la Amiga, Macintosh, or IBM. They produced a type of chip that can emulate almost any other chip with the right code. In time you might be able to use these chips in IBM PC's, Macintosh PC's, SPARC workstations, etc. In other words, they produced a meta-chip that can transform to look like other chips. Their name Transmeta actually makes sense given what they produced.

    6. Re:Excitement with reservation by Guy+Harris · · Score: 2
      Since Intel has said that their next proc is VLIW, it looks like TransMeta's VLIW proc is leading them...

      ...except that the IA-64 instruction set will be available to programmers and compilers, whilst the instruction sets of the Transmeta chips are not.

      "No mention was made regarding the connection to the Internet...that was just assumed to be there. But I have yet to hear about any affordable and sufficiently fast connection via mobile unit... How will they address this, or will they just leave it up to other companies to solve this general problem?"

      I think you are refering to the fact that the translation units can be upgraded via software. Software which comes over the internet :-)

      Eh? I think he's referring to fast inexpensive wireless Internet access - or the lack of same. There are other reasons to want your mobile device to have fast Internet access than the desire to get upgrades to the binary-to-binary translation software over the Internet (although, unless they've added an "upload new translation software" instruction to the x86 instruction set, and translate that into code to replace the translation code, I'm not sure whether a box with a Transmeta processor would be able to "upgrade the translation units via software" - the technical white paper on the Transmeta Web site implies that, to all the software running on a machine with a Transmeta processor, including the OS and the PROM monitor/BIOS, it looks just like an x86 processor).

    7. Re:Excitement with reservation by JohnFred · · Score: 1


      "I think you are refering to the fact that the translation units can be upgraded via software. Software which comes over the internet :-)"

      Couldn't that result in the security hole to end all security holes, the breaching of which ends up completely changing your CPU instruction set?



      --
      /usr/games/fortune > ~/.signature
    8. Re:Excitement with reservation by rjamestaylor · · Score: 1
      I realize that I wasn't thorough in my points. I did not confuse the fact that Transmeta's Crusoe runs the x86 instruction set and, ergo, every application (OS and otherwise) that runs on x86 instructions (including FreeBSD, Linux, Windows, BeOS[?]). So let me clarify, because I think it will help the discussion.

      First, regarding the Webpad unit that is optimized with Mobile Linux on ROM, how will Transmeta insure Internet Compatibility when currently there is not a shipping broswer that adheres to the W3C standards on the Linux OS? Will Transmeta work with Mozilla or Opera to advance their browsers? Or will they be exposed on this front if Mozilla collapses? (Remember, Transmeta targets Internet Compatibility on the client side!).

      Their Webpad unit, to be truly portable, cannot use a regular modem or ethernet cable: it must use some kind of wireless. Will Transmeta, not just a chip company, assist this or will it depend on others? Either way would be valid, but which will they choose?

      Crusoe has software instruction set. But what motivation would they have to spend another X years to support another microcode instruction set? Afterall, any application, including the MAC OS, can be ported to x86 instruction set. Staying with x86 will maximize their software development investment.

      I look forward to a truly wireless Internet compatible machine I can take anywhere and work/surf. I also look forward to providing a new level of hand-held computer for my customers working in the field.

      My impression yesterday was that mobile computing will never be the same: no more OS/App compromises, abysmal battery life, and, perhaps most important, introduciton of a new standard: Internet compatibility (transcending OS compatibilty).

      :-only kona in my cup-:
      :-robert taylor-:
      --
      -- @rjamestaylor on Ello
    9. Re:Excitement with reservation by Inoshiro · · Score: 2

      Yes.

      Hopefully it will require the physical changing of a jumper on the motherboard + a special bootdisk to flash the translator units.

      Flash rom BIOSes often don't require this, and video BIOSes and SCSI BIOSes are just as vulnerable in the real world, yet they aren't (to my knowledge) often taken advantage of by viruses or trojans in the wild. Any modern operating system restricts access to the hardware, so it'd take a bootdisk to put the machine into real mode before it would likely be able to flash.

      Just a thought.
      ---

      --
      --
      Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
  75. This will be their cash cow I bet by DrSkwid · · Score: 1

    My guess is that this is who they will make their ongoing cash. Charge u for the updates to the processor and let other ppl worry about the end user apps.
    I think their answer about backwards compatibility is a fudge. They said it was so ppl didn;t have to re-compile but it's hardly thw worlds gr8est problem to those ppl who know about such things.
    .oO0Oo.

    --
    There are places where the networks are not touching,and there are places where they are-Boeing's Lori Gunter
  76. But the Question is... by NullGrey · · Score: 1

    Does it run Linu...

    Oh, wait, I guess it does.

    Newest Post!!!

    --
    +-- (Score:-1, Moderator on Power Trip)
  77. Great potential, but what about other uses? by Anonymous Coward · · Score: 0

    Transmeta talked strictly about mobile computing, but what about desktop systems based on this thing? Low power consumption there is good for the enviroment, and good for the electric bill. Also, it would seem that this thing could be programmed to be other, non-x86 (or is it IA32 now) CPUs. It would be great to have a portable Sparc book, without paying portable Sparc-book prices! Or my dream machine... You can select PPC/x86/Sparc/Alpha at bootime, and boot an appropriate OS.

    1. Re:Great potential, but what about other uses? by Anonymous Coward · · Score: 0
      You can select PPC/x86/Sparc/Alpha at bootime, and boot an appropriate OS.
      That's an idea, but the question is whether the morphing code is loadable or if it's masked into the chip during fabrication. The latter case would preclude switching between instruction sets.
  78. Appliance convergence by Anonymous Coward · · Score: 0
    that should read 34 C... I wish it ran at 34 F!

    That would give a whole new meaning to the concept of a computer in the refrigerator.

  79. mainboard by Anonymous Coward · · Score: 0

    my question is what kind of mainboard does this thing have since it runs x86 right now is the mainboard x86 or is it some propritary thing and the code morhping software handles all the requests to the video and IO systems and stuff also couldn't the mainboard be a problem if you say wanted to run say PPC instructions on it, since x86 and PPC boards are different, or does the code morphing software automagicly translate all OS requests into native VLIW instructions and so forth so the main board would basicly be platform independent?? or am I totally bonkers?

  80. Power consumption and more by Espen · · Score: 2

    I'm sorry, 1W descibed as 'incredible' power consumption betrays a serious lack of perspective. The ARM7TDMI consumes 0.6mW per MHz on average (note the 'milli') and these usually run at 66Mhz! That's less than 39mW for a typical processor. 1W means squat until we see the performance figures for this thing. I don't see it making great waves in the mobile device sector unless the power consumption is drastically cut. And what's this about hardware x86 emulation? Have we been tricked by the pre-release chatter? Everyone was talking about this thing being software driven on the emulations side. For now established RISC-based processors don't seem to be challenged.

    1. Re:Power consumption and more by pankaj · · Score: 1

      The 0.6mW/MHz power dissipation figure for ARM7TDMI does not include I/O, any cache memory or anyother peripherals. Basically it is just the core itself - equivalent of about 40K gates.

  81. But the Question is... by NullGrey · · Score: 1

    Does it run Linu...

    Oh, wait, I guess it does.

    Okay, I guess I'll ask the next-to-important question:

    How does this help me get chicks?

    Newest Post!!!

    --
    +-- (Score:-1, Moderator on Power Trip)
  82. Re:Is the microprocessor so important? - yes by Anonymous Coward · · Score: 1


    x86 processors are known to be particularly bad from the power consumption standpoint. When the Pentium II was first launched, it consumed something like 40 watts! There have been several die shrinks that have helped to reduce that quite a bit. I think the typical "maximum" power budget guidelines for a notebook computer is 10 watts for the CPU - the mobile Celeron, PII and PIII CPUs probably come in right around that amount - maybe slightly more 15 Watts or so.

    Now, you look at this Trasmeta CPU that consumes 1 watt of power - that's quite a big difference, like 1/15th the power consumption. (FYI - its not like they are the first people to do this, the DEC StrongARM is a RISC chip that has similar low power consumption - 1 watt or less depending upon operating speed.) What I am really curious to know is how "cut down" is this CPU in order to achieve that power rating? For example, the StrongARM achieves that low power rating (in part) because it leaves out a bunch of processing units typical of a desktop CPU - eg it is integer math only, no floating point unit. My suspicion is that Transmeta must be doing something similar to achieve this.

    As another power benchmark to think about - the PowerPC "G3" chips consume in the range of 5 watts - and these chips have all of the "standard" units (both integer and floating point) - the thing that they do NOT have is instruction level compatibility with x86. The G3 has similar performance to a PII / PIII class x86 CPU so you could infer from this that the PII / PIII must be burning up all that extra power in translating x86 instructions to its native RISC ops in hardware.

    At any rate, it will be interesting to see what exactly this Transmeta CPU does and does not do to achieve such a low power consumption.

    Also, you are correct about the LCD display - the CCFL backlit active matrix LCDs typical of notebook computers are also a major drain on the batteries as is the hard drive though often the hard drive can spin down and "sleep" unless you are constantly hitting it for virtual memory or something.

  83. Corrections by rogerbo · · Score: 3

    -Linus got his ass kicked by Dave Taylor (co author of doom and quake) not the CEO. But the quake3 performance was great if they didn't have a 3d graphics card in there. (i.e. is was running in software mode).

    - The code morphing software WILL NOT be open sourced. It doesn't have to be it's not part of the linux kernel. The code morphing software sits below the kernel translating the x86 instructions into vliw.

    -They will release the "Mobile linux" source code but it looks like all that is is a low memory optimised version of linux with power management and an onscreen keyboard application.
    Nothing earth shaking there.

    But hell, I still want a linux webpad.....

    1. Re:Corrections by Quikah · · Score: 1

      There is no software mode to Quake3, it requires a hardware accelerator. You can run it in software using Mesa, of course it runs at about 3 spf (thats second per frame). :)

      --
      Q.
    2. Re:Corrections by Jburkholder · · Score: 2

      >But the quake3 performance was great if they didn't have a 3d graphics card in there. (i.e. is was running in software mode).


      AFAIK, Q3A does not have a software rendering mode, you either need a Matrox 200, Matrox 400 or a Voodoo 2 to be able to play the game at all using Mesa. I know that some people have been able to get a Q3A 'slide show' by fudging a libMesaGL.so into the game directory, but this gives about a frame every three seconds.

      If something like this was done and it rendered in software at anything near an acceptible framerate, I am _really_ impressed. Seems more likely that there was a Matrox card running some type of GL implementation?

  84. mobile mobile mobile by jrs · · Score: 1

    what about desktop?!?

  85. Native mode by delevant · · Score: 1
    One of the audience asked about Native Mode during the Q&A session:

    . . . it sounds like Transmeta will definitely NOT be releasing the info needed to code directly against their hardware. Their argument hinges around the fact that the two chips they have today both have different code requirements -- so if they let you write directly to the 3120, your code wouldn't work at all on the 5400.

    Moreover, they expect to continue this state of affairs. The result is that if you coded directly for any given Transmeta chip, you'd be writing stuff that didn't work on any other Transmeta CPU.

    Which, naturally, they view as a bad thing.

    Toward the end of their answer to that question, the CEO did hedge a little, so it's still possible that they'll release the info needed to go Native. But it doesn't sound likely.

    . . . or that's what *I* thought they said . . .

    --
    I have no .sig, and I must scream.
    1. Re:Native mode by Anonymous Coward · · Score: 0

      What transmeta have done is build a very fast VLWI processor with a i86 emulator built in software. (Remember the Spectrum Emulator for Linux??).

      Note this is not totally new... WinChip is a RISC processor with x86 translation done in hardware.

      The clever bit is the way that the emulator adapts for the code it is running, this is surely what puts Transmeta ahead of the competition, and is almostly certainly why they will be very protective of their emulation technology

      It's a sad fact that the (comerical) computer industry is x86 based, but that's the way it is. This way they don't have to generate a new market, they just steal a slice of an existing market.

      I wonder if they will ever attempt to produce a plugin x86 processor/emulator say on Socket 7??

      As for native code, as they say applictions will need to be recompiled for different processors - coming from a linux background - 'What's the problem???? Doesn't everyone compile they apps??' My guess is that if you did write native code then you could get some mighty quick apps.....

      Simon Wood

  86. Re:Interesting. This should advance mobile computi by Anonymous Coward · · Score: 0

    Personally I've always wanted one weak enough to give into tempation, come into my office suite and be ready to go, have at least three inputs and a beeper so I could connect whenever I needed =)

    Before you moderate this down (and I couldn't blame you) consider that the thought crossed your mind as well;)

  87. Idea... by Anonymous Coward · · Score: 0

    Someone mentioned that it could emulate 'any' processor architecture.

    Why not have 3 - 4 CPU's in one machine, each emulating a different architecture? Have a hardware switch to swap between them. They have seperate memory blocks but in the same chips.

    Or instead of swapping between them, just slice the screen up into as many sections as you have CPU's and have the mouse live in 'limbo' above the seperate OS's, meaning you don't have to swap from one to the other,they both exist at once on the screen and the mouse can move between them freely, dragging/dropping data where applicable.

    What do you think?

  88. If this is close it only shows we haven't come far by Anonymous Coward · · Score: 0

    A processor for wich its instruction set isnt even given, I can only see it as a step back.

    Sorry I really dont see your POV, please explain.

  89. So what ? by Anonymous Coward · · Score: 0

    A lot of media hype (a very well managed campaign) and then ...

    ... well nothing really.

    They're offering processors which scale well underneath the next big batch of processors from Intel and AMD - and don't forget these Transmeta boys are talking about 0.18 micron so they're gonna have to wait a couple of years to scale to next level. Given that the big 'zillas seem to be releasing new devices every month at the moment I think this little lot is much fuss about nothing.

    As for mobile devices - well frankly nobody really cares if they're x86 compatible.

    Mobile = v.low power, and last time I looked that meant 1W . Something like the ARM ...

    Even more ironic is that with the advent of Linux we're seeing lots of cross platform libraries and technologies. Beating the x86 problem is old hat.

    What this might have been really nice for is a 'native' Java chip.

    You also have to laugh when they say "There's no way a hacker is gonna get into this baby" ... Red rag to a bull time!


  90. Crusoe as hard Java VM by Matt2000 · · Score: 5


    A question: because of the Crusoe code morphing technology does this mean that we could program it to translate Java byte code into the Crusoe VLIW instruction set and get a hardware speed JVM? That would be sweet...

    Hotnutz.com

    --

    1. Re:Crusoe as hard Java VM by hpa · · Score: 4

      Already doing it. It was demoed at the launch.

    2. Re:Crusoe as hard Java VM by Kingpin · · Score: 2

      Wow. No wonder IBM has been all over the place with their Java support for the past year.

      --
      Unable to read configuration file '/bigassraid/htdig//conf/14229.conf'
      Geocrawler error message.
    3. Re:Crusoe as hard Java VM by coaxial · · Score: 2

      Sun has already come out with something that does this; the picoJava microcontroller. Read all about it (alb iet in PDF).

    4. Re:Crusoe as hard Java VM by John+Allsup · · Score: 1

      But picoJava is hardwired Java -- the point about Crusoe is that basically any VM can be soft/hardwired.
      John

      --
      John_Chalisque
    5. Re:Crusoe as hard Java VM by coaxial · · Score: 2

      But picoJava is hardwired Java -- the point about Crusoe is that basically any VM can be soft/hardwired.

      Ture, but it is a hardware implimentation. If all you're concerned with is java bytecode then this would be a cheaper solution.

  91. Davinchi? by jallen02 · · Score: 1

    They may as well bend over and kiss their arse goodbye.

  92. intel schmintel by option8 · · Score: 1

    everybody's saying intel and amd should be quaking in their boots right now, but i doubt even the greatest of expectations coming true could make intel sweat

    i'll tell you who should be concerned - motorola. ibm's been slacking on the ppc front as of late, concentrating on their own ppc variants, and not the altivec/g4 roadmap moto laid out.

    if crusoe can do the heretofore impossible - emulating ppc and running ppc native instructions as fast as a real ppc core, and it seems to take even less power than ppc, it would give the limited risc os market a new place to turn when motorola screws up production again, as it did on g4

    that means next gen amigas, ibm power iron, even macs (!) might take advantage of crusoe to run cooler, faster (?) and cheaper - not to mention have more compatibility to run non-ppc binaries or OSs.

    speaking of mac, why would apple bother to do an x86 port of osX, which has been rumored for quite a while, though never seen in the wild, when they can just optimize for ppc, and tell everyone who wants to run it on non-ppc to get a crusoe?

    --
    - Entertaining Bits from the Ancient Kernel Tree
    1. Re:intel schmintel by VAXman · · Score: 1

      I do agree that Intel will not be the most affected. If "mobile internet handhelds" (or whatever the buzz-phrase is) really do take off and replace PC's, probably the companies most affected would be PC companies such as Dell, Gateway, and Compaq. Intel does have a mobile device which would probably be competitive (StrongArm, though, ironically, not x86 compatible) in such a market. But their bread and butter is in server chips, and this device would only create more servers!

      I personally doubt that "mobile internet handhelds" will replace PC's, though I do not doubt that they would be extremely popular. Most people want a big monitor, and an entertainment/desktop system, which the mobile is not really compatible with. The paradigm will probably similar to modible phones and landline phones, or even PDA's and PC's, where people have a big, powerful machine at home, and a small simple one for the road.

      I also doubt that Crusoe chips will be used in PC's because right now they are more expensive than Celeron, et al. Plus, we do not know the performance. Theoretically, since they are small, they could be cheaper, but IBM doesn't exactly have a good volume manufacturing record.

      Of course, the PC -> handheld revolution COULD be the next mainframe -> PC revolution. However, I doubt this, because usually the really important revolutions are not hyped or predicted, but just happen.

    2. Re:intel schmintel by IntlHarvester · · Score: 1

      One thing to realize is that the "mainframe to PC" revolution didn't shrink the market for mainframes and minis-err-I-mean-servers, it in fact increased it quite a bit.

      The real game is about expanding the market, not playing zero-sum ball with Wintel, DEC, IBM, or whoever the dominant player of the day might be. Of course, MS, Intel, Sun, IBM and everyone else realizes this, which is why they are all making investments into this space.
      --

      --
      Business. Numbers. Money. People. Computer World.
  93. Some interesting similarities to MAJC by ChrisRijk · · Score: 5
    I managed to see most of the broadcast, but missed about the first hour's worth. Anyway, it has some interesting similarities with Sun's MAJC architecture design:

    Been in development for some time, but secretly. (Didn't hear a word from Sun until it was practically complete)

    Has the idea of trying to remove backwards compatability hardware problems and issues. (Crusoe with code morphing, MAJC with Java). This makes it much easier to really optimise for each generation.

    VLIW type design. Sounds like Crusoe is fixed 128bit - like most designs. MAJC is variable - 32-128.

    low power embedded markets. However Sun is more "embedded" than low power (MAJC 5200 is 15W @ 500MHz), but Sun are going for some pretty damn serious performance - eats mutliple MPEG2 streams for breakfast, 100 voice of IP channels at once, or 50-90M triangles/sec for 3D lighting/transform etc - the PlayStation 2 "Emotion Engine" is a similar product (in terms of performance, power, cost) but is rather more conventional.

    Both using IBM fab. Both 0.22 initially, and 0.18 later. (Sun are using copper interconnects, I guess Crusoe is too)

    The point about doing benchmarks for the Crusoe discussed in the annoucement is quite apt too - with Java HotSpot, the longer you run it for, the faster it gets. Normally, you use a real application for minutes or hours, but most current benchmarks don't run that long, so isn't quite so "fair".

    However, Crusoe beat MAJC to being fabbed and sampled. (MAJC should have "taped out" by now, though no official annoucement yet)

    Different markets (MAJC doesn't execute x86 for one, but maybe they could add it later...), though there is some overlap - I think both are going to be very interesting to watch. Both bring some interesting new ideas and applications of things.

    Some architectural differences: Crusoe could do just about any instruction set "directly" through code morphing - you'd just have to code it. However, don't expect them to do many as it would be a huge amount of work for each instruction set. They can also do more than one at a time. Though MAJC is not a Java bytecode executor (and you could port Linux to it as easily as a typical RISC CPU) it only does it's instruction set. They hope to use Java to make things more "portable", which is a lot harder than the code morphing techinique which is basically transparant. Not much details has been given about the Crusoe engine, so it's hard to compare, but it doesn't yet seem like it has hardware/vertical threading support, or chip level multiprocessing support (more than one CPU core on one chip), for example.

    MAJC does have this one thing which similar in terms of complexity and mixing hardware/software though. When running a JVM, you can use a mode called STM (Space Time Computing) which uses more than one CPU to speed up a single threaded Java app (using some interesting thread speculation techniques), which like the Crusoe code morphing engine, is transpart - you don't need to compi

    1. Re:Some interesting similarities to MAJC by Le+douanier · · Score: 1


      "Has the idea of trying to remove backwards compatability hardware problems and issues. (Crusoe with code morphing, MAJC with Java)."

      Nothing prevent Transmeta on providing a Java morphing, thus acting like MAJC, or transforming a bytecode program into a morphed langage (like other compiled programs) thus eliminating the performance drawback of using bytecode.

      The bytecode would then be just one more processor/instruction set.

      --
      "The obvious mathematical breakthrough would be development of an easy way to factor large prime numbers." Bill Gates,
    2. Re:Some interesting similarities to MAJC by Andrew+Gilmore · · Score: 1

      One correction: Apparently the instructions are
      64 or 128 bit wide.

      (Perusing the tech white paper...)

      Enjoy!

      --
      ------ Nope, Not me, you can't prove I said that!
  94. Re:Screw the emulation - 128 bit ? by Anonymous Coward · · Score: 2

    I think I should explain something. What does 128-bit VLIW (Very Long Instruction Word) processor mean? It's not about register size. Neither about data or address buses size. It's instruction word size. The processor itself is probably 32-bit since it's supposed to emulate IA-32, but it could be 64-bit. some numbers: 8088: address bus size:20 data bus size:8 register size:16 instruction word: min=8 max=64 ??? instructions / cycle =~ 0.1 i486: address bus size:32 data bus size:32 register size:32 instruction word: min=8 max=120 instructions / cycle =~ 0.7 pentium: address bus size:32 data bus size:64 register size:32 instruction word: min=8 max=120 instructions / cycle =~ 1.3 K7: address bus size:32 data bus size:64 register size:32 instruction word: min=8 max=120 instructions / cycle =~ 4 crusoe: address bus size:? 32..36 ??? data bus size:? 64 ??? register size:? 32 ??? instruction word: 128 ! instructions / cycle = 3..10 (integer value !)

  95. Linus == CEO whore??? by Anonymous Coward · · Score: 0


    "Linus got his ass kicked by the CEO in quake. =)"

    By the CEO, maybe he was trying not to ridiculate his boss in public ;)

  96. Webcast? by CrAlt · · Score: 1

    Does anyone know where there is a archive of the webcast? I looked on Zdnet's site and didnt see it. Thanks

    --
    I have to return some videotapes...
  97. And thats in .35u by Anonymous Coward · · Score: 0

    Cant really run DVD though unlike the Crusoe :) But once Strongarm hits .18u it should own...

  98. technical GPL violation? immaterial maybe, but... by MattMann · · Score: 2
    Is this a technical violation of the GPL?

    Linus said in the webcast that Transmeta made a mini Linux distribution for its licensees. However, if the licensees could not publish that GPLed software without violating their NDAs with Transmeta, then the GPL forbids Transmeta from distributing that Linux. It doesn't say anything about whether the licensee wished to publish or disclose, nor does it say that the NDA must be amended, it says "if" and "refrain". Hey, I'm not gonna call 'em on it, but what do you all think? :)

    Quoting from the GPL (and careful, don't confuse the example they give with the principle):

    7. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Program at all. For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program.
  99. is the architecture open? by Zooko · · Score: 1

    Is Transmeta going to publish the architecture so that we can make a native Linux port instead of running x86 Linux through the interpreter?

    Zooko
  100. Linus -> figurehead? by Anonymous Coward · · Score: 0

    A few things struck me with the press conference:

    Linux, when asked what his future responsibility at Transmeta will be had to stretch for an answer. Normally this should not be a difficult question for anyone -- you tend to know your job. A job description of "whatever they tell me to do" does not inspire confidence.

    Linus also stated that Mobile Linux was just plain old Linux. Nothing particularily new or exciting.

    My cynical take on all of this is that Transmeta hired Linus because they wanted to inherit all the hype and excitement surrounding Linux. If Linux is the newest coolest thing, surely hiring its creator makes the Crusoe the newest coolest thing by association alone. If this was the case, it sure as hell worked. Every member of the cult of Linux has been drooling for months over what their guru was working on behind those closed doors.

    Quite a nice deal for Linus though -- being paid to do what you love in return for generating hype.

  101. CNET Article review by RobNich · · Score: 1

    I reviewed the cnet article, which is pretty good in most respects. My notes:
    - There is considerable focus on the processer clock speed, without mentioning the fact that it morphs instructions to run them. This reduces the number of instructions, and therefore clock ticks to complete the same amount as a 'traditional' x86 processor.
    - They make no mention of the fact that the processors aren't necessarily competitors only in the x86 market. They can run many other processor's instruction sets.
    - Interesting that they didn't mention the Linus/Linux link to Transmeta and the fact that Linux has already been written specifically for the platform. Also, that the 5400 will run Windows but is not yet in production. Interesting.

    Overall, a pretty good article.


    --
    Hello little man. I will destroy you!
  102. Bye-bye PalmOS by Indomitus · · Score: 2

    Well, I was going to buy a Handspring Visor (running PalmOS) but I guess I'm waiting for one of the Crusoe processor based systems now. :)
    Handheld Quake. Three words: Cool as hell.

    1. Re:Bye-bye PalmOS by Wharper · · Score: 1

      Actually the PalmOS could run on a Crusoe platform. Crusoe does not provide an operating system only native code translation. PalmOS is not going anywhere soon, seeing how no one else seems to be able to create an OS that works on a small mobile platform. (CE Couldn't do it).

  103. Slashdot mentioned on Transmeta site? by Percible · · Score: 3

    Judging by the following URL -

    http://www.transmeta.com/mobile/

    Looks like someone at Transmeta likes Slashdot, too.

    1. Re:Slashdot mentioned on Transmeta site? by McFarlane · · Score: 1

      ha ha!

      Looks like someone's idea of a before and after shot.

      BEFORE: bloated mobile computing = windows, x86 processor in a laptop, entertainment = solitaire

      AFTER: debloatated computing = mobile linux, crusoe processor in a (wireless?) webpad, entertainment = slashdot (viewed in mobile mozilla??)

      --
      [We don't come from a planet. We come from a grid sector.]
  104. Slashdot on the Webpad at Transmeta Site by newial · · Score: 1

    The Webpad shown at the Transmeta Site about the mobile products shows Slashdot on the screen.

  105. Benchmarks by Tim+Behrendsen · · Score: 2

    First, let me say that the chip looks mighty cool. However, I was very disappointed in their not releasing industry standard benchmarks.

    While I understand their wanting to show their big power advantage, trying to mix two totally different measurements such as power consumption and performance into a single rating is the height of marketing bulls**t. But OK, if they want to, that's fine.

    But not at the exclusion of real performance benchmarks. Show me the components that went into the bulls**tmarks or whatever your new benchmark is.

    At one point during the Q&A, someone made this point, and the engineer dude (can't remember his name) said that a 667mhz Crusoe performs like a 500mhz P/III. *cough* I'll believe that a software-based emulator can get 75% of native hardware performance when I see real benchmarks. Until then, all this handwaving makes me very, very suspicious.

    All this having been said, the screens flashed by pretty quickly, so clearly it's not a dog. But Transmeta: if the performance is good DON'T HIDE THE NUMBERS.


    ---

    1. Re:Benchmarks by ocie · · Score: 1

      I'll believe that a software-based emulator can get 75% of native hardware performance when I see real benchmarks. Until then, all this handwaving makes me very, very suspicious.

      As I understand it, the code is not so much emulated as translated on the fly as needed. Once a section of codee has been translated, the processor caches this translated code, which is now running natively. Not an easy thing to do, which is even more reason to admire the work that went into this product line.

      --
      JET Program: see Japan, meet intere
    2. Re:Benchmarks by Anonymous Coward · · Score: 0

      I was very disappointed in their not releasing industry standard benchmarks.

      Oh, they will come out in the reviews. But "standard" benchmarks are useless for comparison here, as are straight "megahertz" comparisons.

      Think of it this way: Q:how fast a CPU does it take to play a DVD movie? A: about 400MHz. So the $300 700MHz Pentium plays it for 1.5 hours before you need to change the battery. The $139 700MHz Crusoe plays it at 400MHz for four hours before you need to pop in a few AA batteries.

      Which would you prefer?

      Point is: benchmarks mean nothing if they are not related to real-world applications and how users really use the machine. Read the PDF files at www.transmeta.com and respond to their benchmarking practices if you believe them to be misleading.

  106. Bleh! by Anonymous Coward · · Score: 1

    So this is it?

    I must admit I'm slightly disappointed after all the hype and excitement. There's no revolution goin on here, folks. Just a small step of evolution.

    Bleh.

  107. so it's a low-power x86 CPU. by Anonymous Coward · · Score: 0

    when all is said and done that is all we have here. ultimately the success or failure the company (as ever) rests on the market for this product. oh, and how the major competitors can crush it.

    the technology implemenentation is nice - transmeta can easily address instruction set changes by intel & amd simply - and fix any bugs they have via firmware upload. however, the ultimate performance of the chip might not be as high as 100% silicon (but for a mobile market, this is not so important). as the silicon is simple and cheap, transmeta are likely to be able to handle lower prices from competitors.

    the hardware approach is interesting. a full VLIW design. these things never took off because any hardware change means rewriting a compiler and compiling *all* application software. not really acceptable! but as this one only ever runs one application - an x86 emulator - the problem goes away. transmeta do absorb writing a compiler for each chip generation, as it is part of the CPU development budget.

    sun's MACJ is similar in philosophy, being another VLIW. that is more of a traditional processor in that they will preserve the instuction set across generations. but in that case they are really aiming to run a VM on it - Java in that cae, rather than x86. it's a different beast, but perhas trotting in the same direction.

    but please remember guys that it's success depends on the what it is. a low power, low cost x86 clone. it saves you a fresh battery at mid-day. so in the scheme of things it is not so exciting.

    an ARM (and they admitted this) uses less power.

    1. Re:so it's a low-power x86 CPU. by Anonymous Coward · · Score: 0

      SUN couldn't find their ass with both hands. They've been selling UNIX for years, and they still let MS/Intel walk all over them. Java is going to loose out when compared to Linux and Perl. Oktober

  108. All TM chips can run all x86 OS's by AJWM · · Score: 1

    The TM 5400 is optimized for Windows but it'll run any x86 OS.

    The neat thing, that I haven't seen anyone here pick up on yet, is that the Mobile Linux based web pads, etc, will have about a six-month lead over any Windows-based Crusoe machine -- that is, if you want a web pad (and they are cool looking machines), you get Linux. The 3120 is in production now, the 5400 won't be for another few months. (Oh, I suppose some manufacturer might decide to do a Windows-based machine on the the 3120, but it'd be more expensive: Windows license cost, hard drive for the OS (the Linux is ROMable), added development cost).

    --
    -- Alastair
  109. PalmOS? by IHateEverybody · · Score: 1


    Maybe with the right code-morphing softare it can emulate the PalmOS. Or even run the Palm GUI on top of Mobile Linux kernel. Maybe? Pretty please?

    Hmmm, a 400 MHz Palm Pilot. Maybe I'll try that color screen after all.

    --
    Does this .sig make my butt look big?
  110. x86 emulation comparable to what MHz PIII? by Anonymous Coward · · Score: 0

    PowerPC was nice, but they never ran the old 680x0 code worth beans. Wheather or not Crusoe is the "Pentium/Athlon Killer" rests on how fast it runs in x84 emulation/translation/whatever-they-wanna-call-it mode.

    1. Re:x86 emulation comparable to what MHz PIII? by Anonymous Coward · · Score: 1

      It already is an P3-Killer because it blasts them off in terms of power-dissipation.
      And do not even talk about the power dissipaton of Athlons.
      Just remember that part is aimed at NOTEBOOKS .. not desktops or servers ... power matters a lot there

    2. Re:x86 emulation comparable to what MHz PIII? by Anonymous Coward · · Score: 0

      The 700Mhz version of the chip is supposed to be around the same speed as a 500Mhz Pentium III.

    3. Re:x86 emulation comparable to what MHz PIII? by IntlHarvester · · Score: 1

      PowerPC was nice, but they never ran the old 680x0 code worth beans.

      Actually, even a lowly Performa with a 603e/120 can run 68K code in emulation faster than the fastest real 040 Macintosh Quadra, which topped out at 40Mhz, I think.

      The key to emulation is to have a CPU that's fast enough to make the emulation transparent. Early on, the PPC guys predicted their 'scalability' would be so great that they would soon be running x86 in emulation faster than any real x86 processor.

      Of course, Intel wasn't about to let that happen, so Transmeta is taking the wise approach at aiming for a different market (laptops and handhelds that Intel couldn't touch without losing tons of performance.)
      --

      --
      Business. Numbers. Money. People. Computer World.
    4. Re:x86 emulation comparable to what MHz PIII? by Anonymous Coward · · Score: 0

      Quadra 950 came in 66Mhz 68060

    5. Re:x86 emulation comparable to what MHz PIII? by IntlHarvester · · Score: 1

      No mac shipped with an 060, so yer wrong there.

      When the Q950 was shipping, Apple sold it as a 33Mhz CPU (just as Moto had labeled it). Some point later, Apple retroactively doubled the clock speed on all 68040s for marketing reasons. It was a debatable decision, most people quote the old clock speeds.
      --

      --
      Business. Numbers. Money. People. Computer World.
    6. Re:x86 emulation comparable to what MHz PIII? by Anonymous Coward · · Score: 0
      Actually, even a lowly Performa with a 603e/120 can run 68K code in emulation faster than the fastest real 040 Macintosh Quadra, which topped out at 40Mhz, I think.
      It wasn't the emulation speed that was so much of the problem, it was the compatibility of applications. Quite a few programs didn't work on the PPC when it was first released.

      The offical Apple line was... "if the application was written within our guidelines then it will be compatible with the PowerPC".

      Eventually everything was rewritten or patched or just forgotten about.

      It was, however, a growing pain for the Apple platform during the days the CISC was believed to be at the end of their capabilty to expand further.

      Intel has never waived from the original x86 instruction set (ala 1979 or whenever) but that additional baggage is a bit of a pain at times.

      The whole concept behind Transmeta is that you can loose that baggage without causing a rift in your developer or user communities.

      Although I can't say I'm happy with their lack of industry standard benchmark results...

  111. Code morphing vs emulation?? by Julian+Morrison · · Score: 1

    Can someone explain what the diff between this code morphing and an emulator written in assembler is? Is it that the emulator binary is loaded into a fast on-chip cache, or something?

    1. Re:Code morphing vs emulation?? by gwyndaf · · Score: 1

      Code morphing is compilation, not interpretation. The x86 (or whatever) code is translated into an equivalent program for the Crusoe.

    2. Re:Code morphing vs emulation?? by trey · · Score: 1

      does this mean we take a performance hit?

      --

      he who has the fastest cart always has the best lie.
    3. Re:Code morphing vs emulation?? by Guy+Harris · · Score: 2
      does this mean we take a performance hit?

      The first time a chunk of code is run, there's a performance hit as it's translated to the native instruction set for the particular chip on which it's running. The translated code is cached, so the next time it's run, if it's still in the cache of translated code, there's no performance hit for translation.

    4. Re:Code morphing vs emulation?? by rew · · Score: 1

      The first time a chunk of code is run, there's a performance hit as it's translated to the native instruction set for the particular chip on which it's running.

      Nope. the first few times it is executed it takes a performance hit as if it is interpreted. As that's exactly what the transmeta CPU is doing.

      Only after a few times the code is going into the compiler. And after a few hundred (my guess) times the optimizer is kicked into action....

      Roger.

  112. I like recompiling... by Anonymous Coward · · Score: 0

    http://www.userfriendly.org/cartoons/archives/98de c/19981201.html

    Theres good arguments to not do a native version of windoze for it, the arguments for Linux are bunk. They decided not to release the instruction set, the fact that we wont see a native Linux is a result of that not the other way around... pity Linus has resorted to just towing the company line.

  113. Da Im only to and a half years old by CrAlt · · Score: 1

    32F / 0C is freezing.

    --
    I have to return some videotapes...
    1. Re:Da Im only to and a half years old by Bartmoss · · Score: 1

      Hey, how am I supposed to know? Nobody besides those pesky Americans uses that outdated system. But I did know that 0C was freezing, thank you very much. ;)

  114. transmeta.com /.'d by famfurnell · · Score: 1

    Seeing how that appears to have happened (like we ever doubted it wouldn't!), the pdf docs (the important stuff) has been mirrored at http://printf.net/transmeta/

    1. Re:transmeta.com /.'d by dattaway · · Score: 3

      And a mirror of the mirror is mirrored at http://www.attaway.org/~ dattaway/www.printf.net/transmeta/ as well as the irc transcripts. Enjoy!

  115. What's wrong with Transmeta by LinuxParanoid · · Score: 5
    I learn things through critique. Given all the hype, what is the counter-case? In the interest of eventually reaching a balanced perspective, here's what looks wrong with Transmeta to me:

    Power consumption of the chip is lower, but power consumption of the chip is only 20-30% of a notebook, limiting the value of this "revolution."

    Further power reductions require either A) giving up a hard disk (aka Linux in ROMs) or B) integrating more than just the CPU and chipset (what about 2D/3D just for starters, not to mention sound, fast ethernet, modem, wireless etc.; note that some of these require analog circuitry not just digital and pinouts start getting complicated)

    Sure, Intel's SpeedStep power circuitry is less dynamic, more of a step-function static approach to power management. But is it good enough? If not, will the next generation of their technology in 2-3 years be good enough? Not much of a market window here, in the big scheme of things. Remember, Intel only loses when CPU power is an issue; it can pursue the same no-hard-disk and system-on-a-chip approaches as Transmeta. No patents there.

    In terms of integrating 3D, Intel has a huge lead over Transmeta in terms of patent licensing and technology development.

    So what about Transmeta in the embedded space, a la cell-phones? This appears to be a backup strategy not articulated yet for one simple reason: the TM processors are still less power-efficient than, say, StrongARM.

    Did I mention the difficulties Transmeta would have keeping up with Intel's clock rates and performance? There's not a clear win here today, and this is only going to get worse before it gets better. It's relatively easy to release one innovative product that hits the market sweet spot once; it takes a totally different set of skills to keep up development of an ongoing stream of products that is always competitive with what's in the market. You can see this in the 3D space over the last four years, and AMD also illustrates the ups and downs of playing challenger.

    Wireless internet is cool, but I find it hard to be optimistic about the per-month pricing over the next 3 years at reasonable bandwidth rates attracting serious (5+ million) consumers. Guys putting up towers and satellites are the bottleneck here, as is the degree of competition.

    This is all very innovative, and perhaps Transmeta OEMs will sell a few million units of handheld notebook/palmtops, with Transmeta gaining reasonable market share over the short term, IPO'ing to incredible hype, and three years from now realizing that well, they don't have the market position needed to really compete when Intel puts the squeeze on. Their technology value-add that I've seen is too slim that it can't be embraced and extended by some means. I see enough value add for them to survive, to live well and cash in on some sweet stock options, but I don't see them becoming a big or significant player 5-10 years out. Long term, well after the IPO and honeymoon period are over, they only make sense combined with someone like AMD with a much broader product line and established consumer reputation.

    How's that for thought provoking? ;-)

    --LP

    1. Re:What's wrong with Transmeta by MadAhab · · Score: 3


      This isn't a bad critique, but to me it looks like the improvements Transmeta has acheived in power consumption could really enable smart, powerful, application machines that are worth something. PC's that now cost $1000 or less can be replaced with tablets that have browsers and a few office apps on top of Linux. Let's face it, that would meet the computing needs of 90% of the populace, and having no fans makes them much more home and SOHO friendly. Storage is the only remaining obstacle to having nice little machines that do all this, and there is hope on that front as well (I seem to remember reading about a technology for 2300GB, solid state, deck-of-cards-sized, $200 drives in the next two years - just imagine your portable MP3 collection!). This might be a better way for Linux to bury windows - start with WinCE and work from there.

      --
      Expanding a vast wasteland since 1996.
    2. Re:What's wrong with Transmeta by bhurt · · Score: 3

      On the other hand, Intel is going to try to switch architectures here in a year or three, from the old x86 to the new IA64. This is a vulnerable time for Intel- if you're moving onto a new architecture, why not move onto the best one available? This is the best time in the last couple of decades to launch a new CPU.

      I'll agree that the odds are still against them. Too many people buy on the basis of MHz rating and "Intel Inside". But they have a shot...

    3. Re:What's wrong with Transmeta by ebbv · · Score: 1


      your post is well written, but anyone could make speculations like that and come to any conclusion they wanted. that's the problem with speculation, if you try hard enough you can find facts to back up any probable outcome.

      i prefer not to bother with attempts at playing psychic, and just wait until i can get my hands on a product made with crusoe.

      right now, crusoe is not all that impressive, yes it can run any OS an x86 can run,.. so can an x86 chip :P but, what will be impressive is when they start releasing other VMs,.. you can have a laptop with linux, irix, solaris, be os, macos, whatever
      you want,... all happily working together,.. that's what i'd like to see developed with this technology. longer-lasting battery life is well and good, but i'm not the type who travels much, so if i buy a laptop, at least half the time it's going to be plugged into an outlet anyway. (though i know this is not the case for most of you, i'm a selfish bastard.)

      the ideal implimentation would allow switching without rebooting, let's all click our heels and hope this is what comes to pass... i want laptops that only boot when you first take them out of the box! ;)

      a low power consumption x86 clone is neat, but not as impressive as what is to come.
      ...dave

      --

      Think different? I'd be happy if most people would just think...
    4. Re:What's wrong with Transmeta by tj2 · · Score: 3
      Your points regarding the Transmeta chips are well taken. However, a couple of points:

      1. Power efficiency is a lot, but it isn't everything. I think the big win here is lower power draw than Intel, while offering x86 compatability that StrongARM doesn't have.

      2. Clock speeds are becoming less and less of an issue. Let's face it, there's damn little software of interest to the average consumer that's actually going to need 800+ MHz. Yeah, yeah, I know, wait until next year, today's screamer is tomorrow's piece of junk. But Transmeta is also free to increase clock speeds and improve their chips.

      Most of the market they seem to be aiming for is concerned with web access and email, not video editing. I think these chips will have the power for that.

      3. As for mobile access to the net, check out www.metricom.com. They make the Ricochet wireless modem, which is currently in the process of rolling out a 128kbps network to about 46 cities. Metricom had a successful beta of the network last year and secured about $600 million in financing from Paul Allen and MCI/Worldcom for the rollout. I've used the older Ricochet modems (28.8kbps), and they worked like a charm. And they use flat-rate pricing, which is critical for mobile users, IMO.

      4. Why do they have to combine with AMD or someone else? They're not a chip fab, they're a design shop. I expect that they'll partner with whoever wants to make the chips, which will depend on consumer demand. We'll see.

    5. Re:What's wrong with Transmeta by spinkham · · Score: 1

      Except that Palm has already buried WinCE...

      --
      Blessed are the pessimists, for they have made backups.
    6. Re:What's wrong with Transmeta by IkeTo · · Score: 1

      > Power consumption of the chip is lower, but
      > power consumption of the chip is only 20-30% of
      > a notebook, limiting the value of this
      > "revolution."

      The remaining goes to hard disks and display. For hard disks, we have solid state HD which might eventually get it down another 20-30%. Anybody know about what we can do about display efficiency in terms of power?

      > In terms of integrating 3D, Intel has a huge
      > lead over Transmeta in terms of patent licensing
      > and technology development.

      I'm an idiot here. What's wrong to just imitate the 3D instruction set of Intel? Would "fair use" to obtain "interoperability" guarantee that such approach won't be successfully sued?

      > Did I mention the difficulties Transmeta would
      > have keeping up with Intel's clock rates and
      > performance?

      I see the software approach for Crusoe a big win here. The complexity to increase clock speed is non-linearly proportional to the complexity of the hardware, not the software. Whether Intel will take Crusoe approach in its next chip (or buy out Transmeta) I can't be sure.

    7. Re:What's wrong with Transmeta by LinuxParanoid · · Score: 2

      Thanks for your well-thought out reply. I think it fairly reiterates the strengths of Transmeta in points 1 and 2. I could quibble with #2, but in
      the end, I basically agree.

      The pointer to Metricom is interesting-- unlimited wireless Internet at 28.8 kbps for a flat $29.99/month is better pricing than I expected and suggests that 128kbps and higher services from Metricom will also be priced at consumer levels, sooner or later. Keep in mind though that once users taste high-bandwidth wireline (DSL/cable,) it'll hard to go back to slower wireless until wireless speeds up a fair bit. Not an issue for email, but a big issue for web browsing technology adoption.

      The AMD comment was the most speculative of my comments and it is *way* too early for any "serious" talk along these lines. To explain more fully however, its not an issue of fab partnerships; Transmeta's current strategy with IBM fabbing makes perfect sense to me. And Transmeta doesn't *have* to combine with AMD. I mention the AMD combination only because it appears to me to be the most viable exit strategy if Transmeta gets stuck in a niche of low-power x86-compatible CPUs and a bigger player like Intel starts fiercely targeting that market as well. Transmeta can slug it out, but in that case might be better off combined with AMD's greater volumes, brand recognition, leverage across other product lines, and other factors.

      Low-power x86 is a healthy sized niche, and Transmeta has a nice technology lead. But I don't see barriers to entry that Intel can't surmount given some time and focus. Intel may be a little sluggish, but they're not stupid. They may not match Transmeta's leading power usage, but if they can cut it down to a modest enough level, Transmeta has very little recourse that I can see. I don't see any other Transmeta lock-in that would inhibit Intel from retaking 50+% of Transmeta's business away in 3 years. Sure, patents will inhibit some avenues of power-reduction technology development for Intel, but it's unlikely it'll block them all.

      All this being said, Transmeta has a bright future ahead of them in the short term. Sorry if I get nervous or paranoid about the future; it's part of the skillset I use for companies in real life. Hopefully Transmeta management will be seriously thinking about these sorts of issues and come up with creative, effective solutions.

      --LinuxParanoid

    8. Re:What's wrong with Transmeta by Cyno · · Score: 1
      Thought provoking? You suck! You failed to mention anything about the real technological advantages of the crusoe processors. They are basically an FPGA. Imagine a processor you could rewrite the Linux kernel for, so that it could optimize the instruction set on the processor to match the needs of your code at that time.

      Plus it runs faster than any current mobile intel chip and consumes less power. What more could a person want?

      I can't believe the sensors gave you such a high score for that rant. So here's mine... who knows maybe I'll score a 5 too. ;)

      Take it bitch!

    9. Re:What's wrong with Transmeta by LinuxParanoid · · Score: 2


      Your points are fair ones. As I said in a reply to someone else, Transmeta has a shot at opening up a whole new category of low-power, low-cost "mobile internet" as they call it PCs.

      To sum up my basic concern in a single sentence: I don't see enough barriers to entry to prevent Intel from crushing them in 3 years with a "good enough" power-optimized x86 solution. Does Transmeta have a strategy to deal with this?

      --LP

    10. Re:What's wrong with Transmeta by Anonymous Coward · · Score: 0

      That would be Paul Allen, major investor in Transmeta.

      Da na na naaaa....

    11. Re:What's wrong with Transmeta by ElitistWhiner · · Score: 1

      Don't forget... NCR has bubble memory which is stable and uses very little power. If alternate storage technologies scale into commodities, price points will converge with traditional hard drives.

      -r

    12. Re:What's wrong with Transmeta by Kvan · · Score: 1
      Anybody know about what we can do about display efficiency in terms of power?

      Electronic ink. All the major manufacturers are developing it; the main obstacle right now is update speed. My projection: We'll see monochrome E-Ink displays on the market for e-book type appliances within 2-3 years. The beauty with this technology is that they can achieve 300 DPI easily, and the speed isn't very important in this type of machine. Whether we'll se LCD-speed color E-Ink displays any time soon (if ever) is still a big question.


      "A *person* is smart. People are dumb, panicky, dangerous animals and you know it."

      --

      "A *person* is smart. People are dumb, panicky, dangerous animals and you know it."
      - 'K' in Men in Black.

    13. Re:What's wrong with Transmeta by Anthony+Kilna · · Score: 1

      Of course you then update the code morphing software in your flash ROM to support IA-64. :) In reality of course, there will be significant differences in how x86 and IA-64 (Itanium... UGH) comunicate with memory and peripheral busses, so it'll probably not be as easy as that. But if intel unleashes something like an updated MMX instruction set, there's no reason the TM processor couldn't be software upgraded to handle those instructions.

      --
      s/[BW]ill(y|iam)?( H\.?)?( G(ate|8)(s|z))?(,? ?v?(III|3)(\.\D)?)?/Girly-man/gi
    14. Re:What's wrong with Transmeta by extagboy · · Score: 1

      I would have to agree. We are very quickly becoming a "handheld" society -- palm pilots, internet ready cellular phones and most everybody has a laptop. Sure everybody wants that fast intel processor for home or work but when it comes to portable crusoe comes out way ahead. It wont take long before manufacturers will be using this processor for top-of-the-line models. The point is that they have the right idea for all mobile computers. How can it help but be the leading force in this market. With the very reasonable prices why wouldnt they be the first choice for manufacturers? Not to mention the "geek attraction" and "new coolest thing" which they would have to have patents on. Now add on Linus and the Slashdot effect -- how can they lose?

    15. Re:What's wrong with Transmeta by extagboy · · Score: 1

      I was going to add this at the end of my post but I forgot. Sorry I had to share. One of those coffee on the keyboard things.

      ...and the Number 1 Change at Microsoft as a Result of AntitrustCharges...
      1. Tables turned in jail, where Bill Gates has no choice but to have "Big Louie's Inmate Explorer" installed against *his* will.

    16. Re:What's wrong with Transmeta by Anonymous Coward · · Score: 0
      The word you want is "censors"

      "sensors" are those pads stuck to your head with the wires coming out of them.

      Can't read? Thank a teacher!

    17. Re:What's wrong with Transmeta by GregWebb · · Score: 2

      I'd have to agree, here. AMD could win very nicely in the desktop arena on this one.

      Why? Intel are pushing IA64, which ISN'T compatible. It can pretend, but not by default. AMD, OTOH, are pushing x86-64, which is a superset of IA32. You just plug it in and use it.

      Which would you bet on? :)

      Greg

      --

      Greg

      (Inside a nuclear plant)
      Aaaarrrggh! Run! The canary has mutated!

    18. Re:What's wrong with Transmeta by Corrado · · Score: 1
      Power consumption of the chip is lower, but power consumption of the chip is only 20-30% of a notebook, limiting the value of this "revolution."

      That's true, but since it runs so much cooler it does not need a fan or a super huge heat sink. That cpu fan reduces battery life and the heat sink adds weight.
      Later...

      --
      KangarooBox - We make IT simple!
    19. Re:What's wrong with Transmeta by LinuxParanoid · · Score: 1

      By integrating 3D, I meant integrating the whole 3D pipeline (think "integrating Nvidia's GeForce 256"), not just the CPU instructions like 3DNow or SSE which allow floating point vector processing for early stages of geometry transformations and lighting.

      You point about suits and licensing of SSE is a valid one; it's not clear to me whether the lack of SSE so far is a technology development issue or a legal isse. Transmeta has added MMX (which is widely crosslicensed) but not yet added SSE. I'm no expert on this but the fab also may need a license; i.e. IBM may have a license to fab processors with MMX technology but not SSE?

      --LP

    20. Re:What's wrong with Transmeta by Anonymous Coward · · Score: 0

      "They are basically an FPGA"

      They are basically sand. What's your point?

    21. Re:What's wrong with Transmeta by TimoT · · Score: 1

      Why? Intel are pushing IA64, which ISN'T compatible. It can pretend, but not by default. AMD, OTOH, are pushing x86-64, which is a superset of IA32. You just plug it in and use it.
      Which would you bet on? :)

      IMO, It's time to leave the old crappy IA32 behind. Basically the instruction set sucks, totally. Backward compatibility wouldn't be such an issue if software companies were doing things right ie. portable software and you get all ports when you buy the software (and new ones, when they appear). But software isn't portable and this is why we are in this situation. IA32 won't die soon and the only thing that will kill it is changes in the way software is made.

      Most software should be architecture (and OS) independent with a simple recompilation required for a port. I think that the GNU tools go a long way towards accomplishing this, since they're quite standardized. I wonder what happened to the ideas of portable GUI toolkits, hardware and OS abstraction layers ? At least M$ isn't aware of them... So here's a free software idea: an industrial-strength portable (yes, that includes Windows98/NT) hardware/GUI/OS/net abstraction library, which looks the same from the application programmer's standpoint no matter where you use it (like an OS compability wrapper). There might already be one, but I haven't heard about it...

    22. Re:What's wrong with Transmeta by copito · · Score: 2

      They are far different than an FPGA. FPGAs are essentially matricies of programmable lookup tables, and muxes. Reprogramming them is relatively slow but they are good for prototyping ASICs or low volume production. The Crusoe processor is a VLIW processor with some extremely fancy translation software and power saving features. It quite far from an FPGA.
      --

      --
      "L'IT c'est moi!"
    23. Re:What's wrong with Transmeta by LarsG · · Score: 1

      So here's a free software idea: an industrial-strength portable (yes, that includes Windows98/NT) hardware/GUI/OS/net abstraction library, which looks the same from the application programmer's standpoint no matter where you use it (like an OS compability wrapper).


      It's called "Java". ;-p

      --
      If J.K.R wrote Windows: Puteulanus fenestra mortalis!
    24. Re:What's wrong with Transmeta by TimoT · · Score: 1

      It's called "Java". ;-p

      And no one uses it because of efficiency concerns, the instability of the language spec and the large bulk of existing C/C++ code. This should be language agnostic ie. written for C (and then interfaced from other languages). Nice try though...

  116. Significant Items by CodeShark · · Score: 1
    Like everyone else, here I am tossing in my 2 cents worth about Transmeta's press conference...
    1. Reference machines...

      What this means is that Transmeta does the R&D for a device (such as the web pad which they displayed), then provides the specs to manufacturers, who can then produce them without requiring startup R&D to design and build Crusoe based machines. Smart move.

    2. The Code Morphing piece -- raves here!! It received a brief mention at the end, but if Intel, etc. add an instruction to the x86 set, Transmeta can use the CodeMorphing piece without requiring a new fab mask to be redesigned into the chip.
    3. Full x86 compatibility, with two different models. The more expensive, faster chip will support all x86 apps including 16 bit ones, the less expensive one dumps the 16 bit because (as Transmeta made clear) there really isn't a set of legacy apps out there.

      I like this alot -- if I'm really committed to Linux, why would I want to be stuck with the overhead from what was originally a fairly bad chip design (8088,8086 up to 80386)?

    4. The LongRun design -- which automatically sets the chip(s) to use power maximally well.
    5. Showing Transmeta chips running both the Windows version and the Linux version of Quake concurrently. 'Course, Linus coulda used a bit more practice before the demo, but at least he mentioned that when he lost, it reflected his skill and not that of the OS... ;)
    I expect this chip to succeed big for reasons mentioned in a previous post.

    --
    ...Open Source isn't the only answer -- but it's almost always a better value than the alternatives...
    1. Re:Significant Items by Guy+Harris · · Score: 2
      Full x86 compatibility, with two different models. The more expensive, faster chip will support all x86 apps including 16 bit ones, the less expensive one dumps the 16 bit because (as Transmeta made clear) there really isn't a set of legacy apps out there.

      If it dumps support for the 16-bit instructions, it doesn't offer "full x86 compatibility".

      However, the Crusoe processor family page says that the TM3120 does offer full x86 compatibility:

      The TM3120 is compatible with the complete range of x86-based operating systems, including those offered by Microsoft and Linux suppliers. Transmeta expects that Linux will be the primary operating system for mobile Internet devices.

      so they don't "dump" the 16-bit instructions in the less-expensive chip. It may not handle 16-bit code as efficiently the more-expensive chip, just as a product from another maker of x86-compatible chips punted on handling 16-bit code efficiently (but said maker had to fix that in the Pentium II because the P6 was to be used in PC's, not in appliances without the problem of legacy apps; the PPro might have been able to get away with it as it was primarily intended for high-end machines running NT and Win32 apps).

  117. Beowulf or SMP by Anonymous Coward · · Score: 0

    Not being a technogeek but a geek-of-all-trades wouldn't it be more powerful to have a single device with these cheapo chips and Linux's SMP? Am I stating the obvious or the ignorant.

    In my imagination I would like to have a multi-TMchip pad with pen and vkeyboard. An engineer could carry a full-blown 3d & raytracing CADD package one handed into the field rather than a bulky, shortlived notebook. (Provided mainboards or the equiv are dev'd.) Hell, at this kind of price (which will drop) why not have four or eight in your desktop. So what if 400MHz is slow right now? That will go up.

  118. Re:technical GPL violation? immaterial maybe, but. by mce · · Score: 1
    I have not seen the webcast and any Transmeta related web site is incredibly slow at the moment, so I've not been able to look for more detailed accounts, but I would guess that it all depends on the definitions of "made a mini Linux distribution for its licensees" and "distribute".

    IMHO this part of the GPL can be worked around temporarily by being very careful in what the contracts between (in this case) Transmeta and the licensees say. One way might be to state "we'll make it for you now and you can come and inspect it and play with it at our premises, but we will only deliver the product to you on January 19, 2000 etc."

    Beware: IANAL.

    --

    --
    Linux user since early January 1992.
  119. *I* operate at 48C. (Las Vegas in Mojave desert) by Anonymous Coward · · Score: 0

    Wimps! It hits 48C (even 50C) a few days during summer here in Las Vegas, Nevada and 40C can stick around for many weeks in a row, with temps not going below 30C even at "coldest" time of the night (just before sunrise)! And as the A/C breaks from time to time, I have to worry about how CPUs will run in ambient 48C air, which the ads for chips/PCs never tell you. Crusoe will be a godsend for desert dwelling geeks like me,if it runs as cool as its specs say it does.

  120. Powered by Crusoe: by Anonymous Coward · · Score: 1

    http://www.qubit.net/hardware.htm

  121. Of Hype and Performance by tdsanchez · · Score: 4

    Being one of a few (if not the only) negative poster, I'm likely to get branded (and moderated) as a troll/flamebaiter, but please hear me out...

    I'm wondering if I watched the same presentation as the rest of the posters here... Deitzel and co. effectively skirted the performance/Mhz question, which says to me that they don't have much to brag about in the performance area, otherwise- They would've bragged about performance/Mhz.

    I could've sworn I was watching an Microsoft/Apple/Intel love-in/press-conference at times. A quote of note: "Crusoe will be a low power internet platform for the future". What the fsck does that mean? There was lots of 'marchitecure', but little in the way of hard performance numbers.

    Looks like Transmeta's smartest move was to hire Linus, 'cause the whole of Slashdot is believing the (and feeding) the hype without knowing all the facts.

    There's an interesting double-standard on slashdot... Announced and unshipping products that are !linux are vaporware, yet announced and unshipping products that Linus smiles on are "the next big thing" and "A new paradigm in computing".

    And they say Mac advocates are fanatics...

    -t

    1. Re:Of Hype and Performance by Smack · · Score: 3

      Well, I'm sure this will fall on deaf ears, but here are some responses.

      I think their concept is low power, similar performance. So why would they brag about their similar performance if that's not their selling point?

      Did you watch the technical briefing? The one that followed the press conference? It was much less jargony. Sometimes companies have to play to their audience. Quotes like you gave are the only things the reporters understand sometimes, and most people don't get their news from Slashdot, so the reporters have to understand.

      Also, Transmeta isn't quite as vaporware as you make it sound. Having production silicon from IBM is a lot different than just having software simulations (like that Russian company a while ago). Since they aren't selling to end users, it's real hard to tell whether these are actually obtainable right now. If no one release a product using them before the end of Feb, then you can start shouting Vaporware.

      And that cheap fanatic comment at the end does make you a flamebaiter. Really, it does. Look in the mirror.

    2. Re:Of Hype and Performance by sgml4kids · · Score: 1

      You're right... it might be hype. But for the first time in living memory, the hype includes Linux.

      They did compare it to Pentium performance. Even if it falls short, I'm pretty confident that it'll run faster than a DragonBall or a Psion. Which makes me happy. I pen-based interface makes me happy too.

      Also recognize that /. is a technically savvy audience. The architecture makes sense to us and it ran the most godawful of all Linux applications: Netscape.

      There's nothing fanatical about being intrigued by the possibilities of this new platform.


    3. Re:Of Hype and Performance by Felinoid · · Score: 1

      And they say Mac advocates are fanatics...
      They say the same thing of Linux fanatics.. only they call them Linux Zelots to reflect the extream nature of the worst Linux fanatic.
      There are also Amiga fanatics... not as many but they still exist...
      OS/2 supporters.. mostly they go ranting on IBMs handling of OS/2 rather than "The greatness that is OS/2" you get from Linux,Mac,Amiga fanatics..
      Finnaly there are Windows fanatics.. They refuse to belive anyone using Windows dosn't love Windows.

      :)

      --
      I don't actually exist.
    4. Re:Of Hype and Performance by The+Man · · Score: 1

      Nothing wrong with a negative post, but your complaint about lack of hard numbers is unjustified. Read the two benchamrking white papers, one on method and the other on specific results against a Mobile P3 @ 500. Not bad at all, really. If they can deliver, I don't think performance will be an issue.

    5. Re:Of Hype and Performance by acarlisle · · Score: 1

      Transmeta has benchmarks on its website.

      -AC

    6. Re:Of Hype and Performance by BenHmm · · Score: 1

      the performance isn't really an issue - this is mobile computing not some tower built for hardcore video editing. It's for laptops running normal biz apps - browsers, wp, spreadsheets. the comparison between, say, the crusoes 700mhz and a 800mhz PIII is irrelevent to most ordinary users.

      Battery life, on the other hand, is key.

  122. Optimized real time interpreting? by crush · · Score: 1

    So, if what's going on is that instructions for an x86 are being translated on the fly into instructions for Crusoe and this is done so well that there is a marginal difference between using the actual Crusoe instruction set what does this bode for independent byte-code implementations of Java etc? Doesn't it open up the possibility of keeping libraries of pre-compiled, pre-interpreted , ANSI C, on the hand-held, then a nice little dynamically compiled x86 gets sent over the net, looks up it's stubs locally and away it goes? Am I being really naive here? (Actually, I don't want an answer to exactly that, I want to know what my naivete is !. I suppose there would be security issues.

  123. Is it just me... by Anonymous Coward · · Score: 1

    ...or have the moderators been smoking crack recently? Maybe the moderators should read the article before moderating. How was that offtopic? (Hopefully its been fixed by the time I post this.)

    To respond to yours - yes, I think this is the next generation, but only for wireless computing. I really don't know what to make of it as far as computing in general goes (nobody ever knows where computers will be in even 5-10 years)... Given all the big names Transmeta has, though, I imagine it will at least be phenomenally successful amoung tech junkies. It sounds incredible and I figure it has to suffer somewhere because it sounds too good to be true (don't let me down, Linus!). Maybe its really expensive? Well, I will have to buy one of those webpads so I'll find out soon enough how nice it is.

    Perhaps the revolution your referring to is the one (insert name here - forget where I read it, probably a Crichton novel) was talking about when discussing ultimately "invisible" computers - when computers are in everything and we don't have to think about them anymore. Different "computer" tasks will be handled by miniaturized devices whose sole purpose is to browse the web, or check your email, or just make your coffee exactly the way you like it. This new chip could potentially help set such a trend in motion.

    1. Re:Is it just me... by ocie · · Score: 1

      ...or have the moderators been smoking crack recently? Maybe the moderators should read the article before moderating. How was that offtopic? (Hopefully its been fixed by the time I post this.)

      Tell me about it. Perhaps there should be a request for meta-moderation button next to articles. If you think an article was unfairly moderated, push the button and the most requested articles are the most likely to be meta-moderated.

      --
      JET Program: see Japan, meet intere
    2. Re:Is it just me... by alphaparticle · · Score: 1

      "Invisible computer" by Don Norman? I think he probably not on the same track here with Transmeta.

    3. Re:Is it just me... by wnissen · · Score: 1

      That's a great idea! Someone moderate this up so it gets to Rob! Although I would restrict it so that AC's couldn't ask for meta-mod. Also, maybe people who assiduously point out bad moderation (moderation marked unfair in MM) would get moderator points more.

      Walt

      P.S. Is this Ocie of HMC?

  124. Transmeta could be pulling out leg here by Squeezer · · Score: 1

    I haven't reviewed the spec sheet yet, but I understand the chip uses one watt of power. Well, I think transmeta might be out to make this chip look better then it really is. When running electronics off of batteries, the watt rating doesn't matter (W = V x I). So you can have a high electical drain an a low voltage and still come out at 1 watt but your batterys wouldn't last long due to the current draw (I'm hoping its more like a low current draw and only uses a few volts and comes out at 1 watt). Can anyone that has the spec/data sheets on it tell me what the milliamp hour rating on it is, so that I can get a better picture of the actually battery life on this chip?

    --
    Does the name Pavlov ring a bell?
    1. Re:Transmeta could be pulling out leg here by hpa · · Score: 1

      Since V is fixed by the battery technology, you have a proportional relationship between P (the symbol for power is P, it's measured in W; just as the symbol for current is I and it's measured in A) and I. You don't use linear voltage regulation in a laptop; it is far too wasteful.

  125. but will it by ArchieBunker · · Score: 1

    run FreeBSD?

    --
    Only the State obtains its revenue by coercion. - Murray Rothbard
    1. Re:but will it by Guy+Harris · · Score: 2
      run FreeBSD?

      FreeBSD runs on x86-compatible processors, does it not?

      The Crusoe processor family page on the Transmeta Web site says of the lower-speed processor:

      The TM3120 is compatible with the complete range of x86-based operating systems, including those offered by Microsoft and Linux suppliers. Transmeta expects that Linux will be the primary operating system for mobile Internet devices.

      and says of the higher-speed processor:

      The TM5400 is compatible with the complete range of x86-based operating systems. This includes all versions of Linux, as well as Microsoft's popular Windows 98, Windows NT, and Windows 2000 operating systems.

      They don't explicitly mention FreeBSD - or NetBSD, or OpenBSD, or BSD/OS, or SCO UNIX, or SCO Unixware, or SCO Xenix, or OS/2, or BeOS, or Solaris, or Coherent, or... - but those are all part of the "complete range of x86-based operating systems".

      As far as the BIOS/PROM monitor, the OS, the applications, etc. are concerned, it's just another x86 processor.

  126. A Chip that has it all by Dungeon+Dweller · · Score: 1

    Cool running...
    Fast...
    128 bit! Goodness!
    You could make a handheld with these things that is both cheaper and faster than the webserver where I work!

    --
    Eh...
  127. Transmeta Reads /. by Anomalous+Canard · · Score: 3

    Just judging from the screenshot in the lower right. That is *if* you can get through to the site.
    Anomalous: inconsistent with or deviating from what is usual, normal, or expected

    --
    Anomalous: deviating from what is usual, normal, or expected
    Canard: a false or unfounded repor
  128. Crusoe is heavily biased towards x86 by Kaa · · Score: 3

    And here is an example (quote from the datasheet for the 400MHz processor):

    Other than having execution hardware for logical, arithmetic, shift, and floating point instructions, as in conventional processors, the Crusoe Processor has very distinctive features from traditional x86 designs. To ease the translation process from x86 to the core VLIW instruction set, the hardware generates the same condition codes as conven-tional x86 processors and operates on the same 80-bit floating point numbers. Also, the Translation Look-aside Buffer (TLB) has the same protection bits and address mapping as x86 processors. The software component of this solution is used to emulate all other features of the x86 architecture.

    So all the people that were thinking about 128-bit floats are SOL. I think that emulating non-x86 architectures on Crusoe is going to be possible, but noticeably harder and slower than x86.

    Kaa

    --

    Kaa
    Kaa's Law: In any sufficiently large group of people most are idiots.
    1. Re:Crusoe is heavily biased towards x86 by albalbo · · Score: 1

      Your forgetting - the HW don't matter, it's decoupled ;))

      --
      "Elmo knows where you live!" - The Simpsons
    2. Re:Crusoe is heavily biased towards x86 by IkeTo · · Score: 1

      > emulating non-x86 architectures on Crusoe is
      > going to be possible, but noticeably harder and
      > slower than x86.

      Did I read something that is really weird? Emulating x86 is the worst of the worst in terms of emulating a processor using another processor. It is simply too complicated.

  129. Transmeta promotes Slashdot by rjamestaylor · · Score: 1
    On Transmeta's "Going Mobile" page there are two systems, the larger laptop running Windows and the sleeker unit running Linux.

    Of course, the Windows unit is shown running the most popular Windows program (sol.exe).

    And Linux is shown running Netscape with Slashdot.

    Gotta love it!

    :-only kona in my cup-:
    :-robert taylor-:
    --
    -- @rjamestaylor on Ello
  130. Crusoe is heavily biased towards x86 by Kaa · · Score: 5

    And here is an example (quote from the datasheet for the 400MHz processor):



    Other than having execution hardware for logical, arithmetic, shift, and floating point instructions, as in conventional processors, the Crusoe Processor has very distinctive features from traditional x86 designs. To ease the translation process from x86 to the core VLIW instruction set, the hardware generates the same condition codes as conven-tional x86 processors and operates on the same 80-bit floating point numbers. Also, the Translation Look-aside Buffer (TLB) has the same protection bits and address mapping as x86 processors. The software component of this solution is used to emulate all other features of the x86 architecture.



    So all the people that were thinking about 128-bit floats are SOL. I think that emulating non-x86 architectures on Crusoe is going to be possible, but noticeably harder and slower than x86.


    Kaa

    --

    Kaa
    Kaa's Law: In any sufficiently large group of people most are idiots.
  131. Ultimate Web Server by rotten_ · · Score: 1

    I am thinking that this processor / platform could end up being the ultimate web server. Built in Linux stuff, ultra low power consumption. It all means that you can build an ultra small, ultra cheap, headless web server that can probably be powered by a medium sized UPS for like a day in case of power outage. Think of the amazing density you could get out of a server farm.

  132. Re:but. by toast0 · · Score: 1

    i was trying to burn off karma

    --
    Need a Catering Connection
  133. In Comparison by Shewmaker · · Score: 1
    I saw the webpad in the CNN article and was wondering to myself what processors other webpads like Qubit's are using and how they stacked up against Crusoe. Anyway, I thought other people may be interested in reading about National Semiconductor's GXLV processor.

    --
    "For the Snark was a Boojum, you see." -From the Hunting of the Snark: An Agony in Eight Fits, by Lewis Carroll
  134. Re:Interesting. This should advance mobile computi by Anonymous Coward · · Score: 0

    Personally, I've always wanted: One, week enough to give in--two, temptation! Come into my office, sweet, and be red; eat to go; have, at least. Three, input-sand a beeper. Sew I could, connect wen ever Rhiney dead.

  135. to avoid backwards-compatibility hassles? by will · · Score: 1

    my guess was that the native instruction set might just be held under wraps until it settles down. the two chips so far revealed differ in their optimisation, for example (the 5400 is slightly better suited to running 16bit code), but once you reach the abstract layer of x86 instructions they are identical.

    with this approach nobody writes code for the VLIW instruction set, so nobody cares if they change it to something entirely different next time around when they think of a new hack.

  136. I guess variable CPU speed Linux problem solved. by Anonymous Coward · · Score: 1

    From my understanding, variable speed CPUs (used on many laptops) that throtle up and down freak out Linux. If crusoe does this, Linux must have resolved this issue.

  137. http://www.transmeta.com/images/padsmall2.jpg by MadAhab · · Score: 1
    Slashdot is there! Look closely.

    It's only redundant if you can actually read it already. Given the 20 min laggin g I'm getting on each page, I'm sure it's out there by now. I pity the fool.

    --
    Expanding a vast wasteland since 1996.
    1. Re:http://www.transmeta.com/images/padsmall2.jpg by Megane · · Score: 1

      The interesting part is they didn't log in.

      I guess that means they're "open-source Anonymous Coward petrified grits" compliant?

      --
      #naabhaprzrag, #sverubfr-000, #agi-fcbafberq, negvpyr[pynff*=' negvpyr-ary-'] { qvfcynl: abar !vzcbegnag; }
  138. What about Beowolf ? by Anonymous Coward · · Score: 0

    I just thought of something -- this new chip is rather cheap (considering Slashdot figures are right); the chip consumes very little power, it is nonetheless fast and needs no fan.

    It's perfect for Beowolf-type clusters. Just think about it -- cost is lower than with other chips, less power consumption, less noise, etc.

    Perfect.

  139. Tech Links by Space+Cow · · Score: 1

    I am surprised to see links to news sites but no tech links. Guess I will post them.

    PDF Tech Sheet - Not specs, just more details of the technology.
    PD F Benchmark Info Whitepaper
    PDF Benchmark Report
    And finally for those of you wanting to develop for the Crusoe, the developer registration page: Register Today!
    Have a good one.

  140. No shortage of pointy hairs at Transmeta by Anonymous Coward · · Score: 0

    The only real code written to a specific chip is a compiler backend nowadays... its clear open source is not a Transmeta concern.

    If I have the source and a native compiler I could give a toss about what processor Im running on.

    I have the suspicion that its just a way to more closely protect their translation code.

  141. Power Comparison by TheCarp · · Score: 1

    I tried to goto www.intel.com and look at how
    the 1 watt power consumption looks compared to
    the Intel Celeron.

    Unfortunaly, Intels web designers decided that
    people who leave Javascript (and java) turned off
    in their browser don't deserve to be able to look
    up intel product info.

    Nice guys.

    Anyway, does anyone know how the Intel chips
    compare power wise?

    -Steve

    --
    "I opened my eyes, and everything went dark again"
  142. Picture of webpad by suraklin · · Score: 1

    Did anyone else notice the screen on this webpad?

  143. I want a "Powered by Transmeta" icon on my webpage by Anonymous Coward · · Score: 0

    ...so I can replace the "powered by AMD" icon on my web site. (no Intel product will ever enter my home. Intel *always* runs too hot.)

  144. Let's not blow this out of proportion.. by -tji · · Score: 1
    A. Exotic emulation technology allows it to run any architecture. Maybe.. But the only architecture they have done anything with, shown any interest in, and advertised as a key benefit is x86. If you only want x86, there are more efficient ways to do it.

    B. Low Power Consumption. Yes, lower than x86. But, not lower than other architectures like the ARM. They acknowledged this in the webcast.

    C. Compatibility. Yes, x86 compatible. I can dual boot it between Linux x86, NT, Win98. My Celeron does this today.. wow.

    What we have here is an x86 processor with good power consumption. In the webcast they were saying something around 20-30% better power consumption than existing chips. That's nice, but not exactly revolutionary..

  145. comments, speculations by Anonymous Coward · · Score: 0

    I like what I hear, and what I read between the lines with the Crusoe family of processors.

    It's too bad they are not making a chip for a Cellular phone or Palm-like device, only HPC and larger form factors. One device that would really benefit from such a chip would be the Nokia Communicator line... that would kick some serious tail.

    Things I'm excited over:

    1) The Code Morphing Software (CMS): enables the user to have any x86 OS on the HPC, not just windows. Also, they can make a CMS that emulates SPARC, Alpha, etc - very cool. The optimization of the code on the fly, and with a translation buffer - very cool - plus, performance improves over time. Bugs fixed within weeks without need for a hardware fix.

    2) App development - any x86 app and OS runs now! very, very cool. If you don't like windows, you can develop linux or BeOS apps, and it will run. Out of the box, this thing is running everything you have and will come to have. And, the possiblities inidicate being able to run different OS apps at the same time (Java and Win running simultaneously was mentioned - I'm sure more is possible).

    3) Power Management - geez, could you ask for a better performance. Sure, your disk drive, cd-rom may eat up energy... but saving power is saving power - why eat up more than you need? Also, with appliances - that will probably use flash cards. Since its software-based - they can continue to improve it.

    4) These two are just the first in the family... can't wait to see what the next ones will be like. If they are releasing 700 mHz in production now, they must have close to 1 gHz in development for release by years end. Seems to me, they stated they were conservative in the initial product debut, and that they are capable of much more in terms of power, performance, and optimization.

    5) Intel and Sun doesn't have much to worry about on the desktop and server side - but, they'll face a good fight from Transmeta on the portable side... this is going to get interesting.

    1. Re:comments, speculations by Guy+Harris · · Score: 2
      The Code Morphing Software (CMS): enables the user to have any x86 OS on the HPC, not just windows.

      True of a PC running just about any x86 processor, not just a Crusoe (as long as the system doesn't have features that some OSes don't support and that have to be supported, e.g. USB keyboards, which aren't, as far as I know, supported by one well-known PC OS...

      ...Windows NT 4.0).

      App development - any x86 app and OS runs now! very, very cool. If you don't like windows, you can develop linux or BeOS apps, and it will run. Out of the box, this thing is running everything you have and will come to have.

      True of a PC running just about any x86 processor, not just a Crusoe.

      And, the possiblities inidicate being able to run different OS apps at the same time (Java and Win running simultaneously was mentioned - I'm sure more is possible).

      Did they really claim that they could do this any better than any other x86 processor? You can run Java apps on Windows on x86 (or Linux on x86, or...) now - as far as I know, you can even do it with a JIT compiler.

  146. Crack open the Champagne by DrXym · · Score: 1
    Put up the bunting, stoke the barbecue and have a street party!

    Why? For today a new microchip was released! How can I possibly contain myself?

    If it weren't for Linus, this processor wouldn't even have garnered 1/50th the interest that it has now.

  147. Under 1 watt -- that's bogus by Kaa · · Score: 3

    Again, some quotes from the datasheets for the processors:

    for the TM5400: (the future one)

    while playing a DVD: 1.8 W
    while playing a MP3: 1.0 W

    for the TM3120: (the current one)

    while playing a DVD: 2.9 W
    while playing a MP3: 1.4 W

    Kaa

    --

    Kaa
    Kaa's Law: In any sufficiently large group of people most are idiots.
    1. Re:Under 1 watt -- that's bogus by thecap · · Score: 1
      The "under one watt" was referring to typical usage. You might always listen to mp3's while on your computer but most people do less power intesive tasks such as browsing the web, working on spreadsheets, etc.

      This chip saves power by slowing down when not actually doing any computations. For example spreadsheet programs spend most of their time waiting for user input.

      distributed.net claims that these idle cycles are wasted by most people but soon you will be able to buy a CPU which will just slow to a crawl when it doesn't have any work to do.

    2. Re:Under 1 watt -- that's bogus by Anonymous Coward · · Score: 0
      distributed.net claims that these idle cycles are wasted by most people but soon you will be able to buy a CPU which will just slow to a crawl when it doesn't have any work to do.

      That's good; I can hardly wait. I'm tired of every couple of months having to take off the covers of my computers and vacuum out those extra cycles that were discarded. Running SETI@home has helped reduce the build-up, but this should help even more.

  148. the competition - something to think about by kwashiorkor · · Score: 2
    I wonder how this competes with the strongARM processor:
    • The strongARM only requires about 400 milliwatts vs. the crusoe's 1 watt.
    • The crusoe has compatibility with existing software (not just x86) via the code morphing whereas the strongARM is a platform unto itself.
    • The crusoe might have a faster clock cycle, but are the extra megahertz used up by the CodeMorphing?
    Whatever the verdict, when coupled by emergent display technologies (OLED and SSD), it looks like the future of ultra compacts is all bright and shiny right now.

    I'm also wondering what the overclocking potential of such a low heat dissipating CPU must be :-)! Imgine one of these babies in a cryotech tower (this is more or less a joke - laugh, damn it!).

    -- kwashiorkor --
    Pure speculation gets you nowhere.

    --
    -- kwashiorkor --
    Leaps in Logic
    should not be confused with
    Jumping to Conclusions.
  149. Tech detail highlights (from the white paper) by AJWM · · Score: 3

    Some highlights from the technology whitepaper on the Transmeta website. It should answer some of the FAQ's raised here so far. (My comments/observations in italics).

    Code Morphing software will "typically reside in standard Flash ROMs on the motherboard". This implies it could be in RAM, and potentially dynamically reconfigurable or switchable, on a suitable designed motherboard. (Elsewhere the paper implies that this ROM software is or can be copied to RAM at boot up for faster execution.)

    The VLIW engine has "two integer units, an [FPU], a memory (load/store) unit, and a branch unit". These can operate in parallel.

    All VLIW code, both translated x86 (or whatever) code and Code Morphing code, live in a separate memory space invisible to x86 code. The size of this memory space can be set at boot time or the OS can make the size adjustable. This last implies that the OS can somehow see this memory, so it's either not totally invisible to x86 code or the OS has some VLIW code hooks.

    "The Code Morphing software includes in its arsenal a wide choice of execution modes for x86 code, ranging from interpretation [...] through translation using simple-minded code generation, all the way to highly optimized code [...] A sophisticated set of heuristics helps choose among these execution modes based on dynamic feedback information gathered during actual execution of the code." This sounds a lot like Sun's "HotSpot" technology for Java VMs. Either way it sounds cool..

    "the translator adds code whose sole purpose is to collect information such as block execution frequencies, or branch history."

    There's hardware support to help the code morphing, ie support for exceptions, speculation, optimization of memory and for self-modifying code. All x86 registers are shadowed, there's a working and a shadow copy. As blocks of x86 code get translated, that page's entry in the Crusoe MMU (for the translation cache) is marked as "translated" so that it doesn't get translated again. A write (by x86 code, indicating self-modifying code) into that block causes that bit to be cleared.

    The Crusoe processor voltage and clock (at least on the 5400) are accessible to the Code Morphing software, which can adjust them on the fly to optimize power/performance for the running app.


    --
    -- Alastair
    1. Re:Tech detail highlights (from the white paper) by Anonymous Coward · · Score: 0

      Fully optimizing compilers (dynamic or otherwise) take a lot of memory. To do x86 optimization on-the-fly, you'd need something roughtly the size of the GCC backend, several megabytes. I don't see how this would fit in the L1 cache, so it probably would have to go in main memory.

    2. Re:Tech detail highlights (from the white paper) by rew · · Score: 1

      (Elsewhere the paper implies that this ROM software is or can be copied to RAM at boot up for faster execution.)

      It is. The ROM is serial and about 8Mbit. So the CPU will spend a significant number of milliseconds of emptying the serial ROM into the SDRAM.

      Roger.

  150. Re:technical GPL violation? immaterial maybe, but. by Anonymous Coward · · Score: 0

    From the Transmeta FAQ:

    20. What is Mobile Linux?
    Transmeta is creating a Linux distribution to support its OEM customers called Mobile Linux. Mobile Linux is designed for systems without hard disks, such as Mobile Internet devices (for example, Web pads, clients). The principal enhancements for Mobile Linux will be in power management and in the reduction of the memory footprint.

    Back to top


    21. Is Transmeta getting into the Linux distribution business like RedHat?
    No. Transmeta does not intend to support end users.

    The purpose for creating a Mobile Linux for OEM customers is to provide a total solution including the Crusoe processor, the Code Morphing software, all the required driver support for our motherboard platform and the Mobile Linux operating system. This will provide our OEM customers with the best combination of features and time to market for the emerging Internet device marketplace.

    Back to top


    22. Does Transmeta intend to release Mobile Linux to the open source community?
    Yes.


  151. question about crusoe-vliw ? by *bjorn* · · Score: 1

    I have this potentially stupid question. Would it be possible to write assembly code in crusoe's vliw instuction set. Or to write a compiler that produces "crusoe-instruction-based" code?

    Couldn't this make the processor even faster as there would be no need for software code morphing?

    Or doesn't the code morhing slow things down at all?

    Ok, I will stop smoking crack... honestly

  152. The Code-morphing-sotware by Anonymous Coward · · Score: 0

    What happens when the code-morpher-software crashes ?? Anybody has a clue about this ...
    Processor panic? Code morpher protection fault?
    TSOD (Transmeta Screen of Death)?

  153. what does this mean for palm pilot people? by zonker · · Score: 0
    okay, so what does this mean for the palm pilot et al? Will the Transmeta chips run palm stuff (prolly not... I dunno)? I like palms... Does this mean convergeance or stiff competition w/ the Palm (architecturally speaking)?


    / k.d / earth trickle / Monkeys vs. Robots Films /

    --

    Large print giveth, and the small print taketh away

  154. Whoops, missed a couple of important links by Space+Cow · · Score: 2

    TM312 0 Data Sheet
    TM540 0 Data Sheet
    FAQ

    Looks like they beefed up their web backend in the last few minutes, the slashdot effect that was going on has disapeared for me. Slashdot seems to be more /.'d than Transmeta is right now! LOL

  155. Re:technical GPL violation? immaterial maybe, but. by LoveBear · · Score: 1

    According to Transmeta's FAQ, they're planning on releasing the source code for Mobile Linux.

  156. i'm missing two things in the announcements by arnim · · Score: 1
    1. what do they mean by "internet plug-ins" ? the can't mean that a CPU runs netscape or IE plug-ins.

    2. where are the q3-benchmarks ? they're mentioned in the introduction of crusoebenchmarkreport_1-18-00.pdf but i can't find them anywhere. looks like crusoe is lousy in floating point.

    1. Re:i'm missing two things in the announcements by jopasm · · Score: 1

      >1. what do they mean by "internet plug-ins" ? the >can't mean that a CPU runs netscape or IE >plug-ins.

      It's partially a marketing thing, partially a
      "slam" against WinCE. I don't think the
      version of IE on CE can run "standard" IE
      plug-ins, and there's no netscape for it.
      A machine running on a Crusoe processor
      could (in theory) run Linux+Netscape
      (or Win98+IE, or Be, etc) and use "standard"
      plugins. This would "open up" more web sites
      to handheld/palmtop devices.

      --

      ObTagLine: The more you run over the 'possum, the flatter it gets.

    2. Re:i'm missing two things in the announcements by Anonymous Coward · · Score: 0

      Not just plug-ins -- you could run the regular versions of MS Word or GNU Emacs, instead of "PocketWord" or "PocketEmacs".

    3. Re:i'm missing two things in the announcements by Guy+Harris · · Score: 2
      I don't think the version of IE on CE can run "standard" IE plug-ins

      Given that IE plugins tend, as far as I know, to be x86 Win32 code, and that not all CE machines have x86-compatible processors in them, at least some of them almost certainly can't run those plugins.

      A machine running on a Crusoe processor could (in theory) run Linux+Netscape (or Win98+IE, or Be, etc) and use "standard" plugins.

      A machine running an x86-compatible processor from a supplier other than Transmeta could do that, too - but presumably the idea is that such a machine might have higher power consumption than one might like (although there is at least one mobile device with an x86 processor in it - the Nokia 9000 family of mobile phones on steroids - although I think it's an embedded x86 and may not be capable of running most x86 OSes; I think the 9000's run GeoWorks).

  157. Great product, but announcement fell short. by MartyJG · · Score: 1

    This is going to be a really great product, and several great products are going to come from this. I can already see how the web-pads can make my daily at-work life much easier.

    However a couple of points that the formal talk (which I watched live in its entirety over the webcast) which I felt were missing:

    1. Focus on x86 translation. Obviously the technology could be taken much further, but the whole talk focused on just x86 code. What about translating other code?

    2. The differences between the two chips weren't made clear (the website helps). From the talk you were left wondering, 'Why would I want anything less than the 700mhz chip?'.

    3. No mention whatsoever of desktop implementation. They made it perfectly clear that they were focusing on mobile products, but what do you call those machines Linus and Taylor were playing Quake on? Light laptops are great, but I still want a desktop machine with ultimate power!

    4. Mobile Linux? Isn't Linux meant to be free? There was no mention of how or when we get to find out about the changes/improvements in this new version.

    5. How do they propose these webpads get and stay connected all day? Are they preaching wireless lan technology?

    Anyway, I'm really looking forward to getting my grubbly little mits on one of these machines. And now we all know why IBM chose this time to announce its adoption of Linux.

    --
    insignificant sig
  158. Backwards compatibility isn't the important part. by Christopher+B.+Brown · · Score: 5

    The implication that they consider the low level stuff part of this their business, and, as you suggest, the "cash cow." It is part of their "competitive advantage" to be able to do that which nobody else can, which is to customize the processors in this way.

    ON THE OTHER HAND. Not giving out the underlying instruction set means that they never have to apologize if they change the instruction set. They claimed that there were different instruction sets for the 3120 and 5400 models; if that be the case, it would be no surprise at all if later models are different again. If people are interactive via "middleware" instruction sets, then all Transmeta need do is to make sure that the microcode continues to support the "middleware," whether that be IA-32 or JVM.

    Vision for the Future.

    There is a really cool thing that this offers... Wouldn't it be a neat idea if Linus were to create what we might call Linus' Favorite Instruction Set, with all the features that he wishes there were to make Linux as fast and robust as possible?

    Alternatively, the Lisp Machine people might find it a slick idea if Transmeta provided microcode to provide a Lisp-oriented instruction set that (notably) provides a really tightly microcoded set of garbage collector instructions. That would let them both benefit from MHz enhancements as well as from generational enhancements, perhaps simultaneously having some IA-32 emulation going on so that they have a virtual machine alongside providing PC compatibility services...

    --
    If you're not part of the solution, you're part of the precipitate.
  159. Naming scheme a bit confusing? by fluffhead · · Score: 1

    I noticed that their chips' names (TM3120, TM5400) are somewhat similar to Texas Instruments' naming scheme for DSP's:

    TMS320C6000 (TMS320C6X or 'C6000 for short)
    TMS320C5000

    etc.

    Not too much alike - theirs starts with TM for TransMeta, obviously, but is much shorter; ours start with TMS - but might be enough to confuse somebody (or get somebody's lawyers' panties in a bunch). ;-) I don't know where the TMS came from here, I'm pretty new to TI. Maybe "TI Microprocessor Solutions" or some such.

    All the usual disclaimers apply, IAAISL (I Am An Inactive Status Lawyer), etc.

    #include "disclaim.h"
    "All the best people in life seem to like LINUX." - Steve Wozniak

    --

    #include "disclaim.h"
    "All the best people in life seem to like LINUX." - Steve Wozniak
  160. Endianness? by BinxBolling · · Score: 1

    So does the VLIW substrate have an sort of intrinsic endianness? If it is little-endian in nature like the x86 that it's so good at emulating, then this could result in a noticable performance hit when emulating a big-endian CPU like the 68k or PPC.

  161. Slashdot got a reference on Transmeta's site... by zonker · · Score: 0
    http://www2adm.transmeta.com/mobile/

    Check out the webpad...


    / k.d / earth trickle / Monkeys vs. Robots Films /

    --

    Large print giveth, and the small print taketh away

  162. Re:PGP by sjames · · Score: 2

    You and me both!

  163. So older crusoe chips are worth more cash? by toppk · · Score: 1

    I guess they are worth more cause the chips get smarter over time ;)

  164. Transmeta FAQ by Smack · · Score: 2

    Hidden away in the Press section (along w/ some unimpressive pics of the chip) is the Transmeta FAQ:

    http://www2adm.transmeta.com/press/faq. html

    Interesting stuff.

  165. Transmeta effect. by plaa · · Score: 1

    Judging by the speed of Slashdot today, Slashdot has been crushed under the Transmeta Effect.

    --

    I doubt, therefore I may be.
  166. Has anyone considered how this affects competition by ragnar! · · Score: 2
    I think the idea is pure genious. Add new cpu instructions by downloding new code. Runtime optimizations. Intelligent power consumption.

    The thing that disturbs me is that we have benefitted as consumers from the competition between intel, AMD and others. If what transmeta has done in fact turns out to be a significantly better design approach, they may never see any real competition.

    Intel has to publish it's instruction sets to get people to write software for the CPUs. Nothing legally prevents other companies from designing CPUs that offer the same instructions. Thats why AMD and others are in business.

    During the announcement, transmeta indicated that it has been granted (or was it sought?) numerous PATENTS, not just copyrights, on concepts related to software defined instruction sets. If they are upheld, would that not keep any other company from designing a hardware / software combination that does a similar thing from competing with them, even if the hardware and software were designed from scratch?

    If someone reverse engineers a crusoe cpu and builds one that is hardware compatible, and transmeta refuses to sell the software component without the accompanying cpu, isn't that like the old apple machines who never licensed the bios so there could never be a (legal) market for clones?

    Yes, this post is another thinly veiled rant against intellectual property rights law, as it impedes growth and competition in many industries.

    "Monopolies cannot come into existence without the assistance of government." - Ayn Rand

    I don't want the government to break up monopolies when they become large and vicous, I want it to stop creating them.

  167. Correction ! by Anonymous Coward · · Score: 0

    Contrary to the summary above, the ZDNet article states:

    "Crusoe will not be targeted at personal organizers such as the Palm, or at cell phones, executives said."

  168. Linux's favorite instruction set is C by Anonymous Coward · · Score: 0

    Its written in it, so it is its ultimate instruction set. Anything on a lower level of abstraction is a compromise, and a compiler can deal better with that compromise than an emulator.

    With closed source this is definetely a very efficient way to be able to scale processing power. For open source it is mostly irrelevant to start with if the architecture changes. Given the amount of man years wich go into making a processor porting the kernel is trivial.

    Funny how this processor was touted as the ultimate Linux processor and it turns out to be the ultimate (low power-) Windows processor...

  169. It's real slow by Anonymous Coward · · Score: 0
    They grudgingly admitted at the press conference that their TM5400 700Mhz was slower than a 500Mhz PIII. I suspect a 800Mhz PIII or Athlon would bury Transmeta's chip.

    This was after they proposed a new set of benchmarks, because the old ones aren't any good. :)

    It's really nothing to be excited about, a slow CPU that doesn't consume much power. A niche product at best.

  170. fo, yfpa by Francisco+d'Aconia · · Score: 1

    Why doesn't crap like this get moderated down?

    Slashdot should run all posts through an additional filter:

    $defaultmoderation = -1 if $post =~ /[Ff][Ii][Rr][Ss][Tt]\s*[Pp][Oo][Ss][Tt]/;

    It won't stop all first posting nonsense, but it will help.

    ---------
    Once in a while you get shown the light,

    --

    ---------
    Once in a while you get shown the light,
    In the strangest of places, when you look at it right -
  171. Full Press Release by Cycon · · Score: 1

    For those of you who can't get through to transmeta's hosed site, here's the full press release:

    (I apologize before hand for any wrapping problems that may occur)

    Transmeta Breaks the Silence, Unveils Smart Processor to Revolutionize Mobile Internet Computing

    For Immediate Release

    Contact:
    Carla Cook
    Ketchum Thomas Public Relations
    650 596-2281
    carla@thomaspr.com

    SANTA CLARA, Calif. (January 19, 2000) - Transmeta Corporation today ended four and a half years of
    secrecy with the introduction of CrusoeTM , the world's first family of smart microprocessors. Designed to
    create a new category of Mobile Internet Computers, the Crusoe processor family (www.crusoe.com) is based
    on a breakthrough software approach that will revolutionize the field of mobile computing. Crusoe delivers on
    the market's need for "all day computing" with a PC compatible solution that is unmatched in performance with
    low power.

    The Mobile Internet Computing Market
    The evolving class of Mobile Internet Computers includes a rich set of products that spans from Web pads to
    ultra-light (less than four pound) Mobile PCs that share the common need of x86 software compatibility and
    long battery life. It represents a significant shift from today's mostly stationary laptops or incompatible
    handheld devices to a platform that offers greater mobility and access to the Web from most anywhere at
    anytime.

    Ultra-light Mobile PCs operating with the Windows operating system and Microsoft Office applications can
    take advantage of the Crusoe processor's low power to increase the average user's productivity by operating
    on a single battery for up to a full work day.

    Crusoe-based Internet devices such as Web pads and mobile clients can take advantage of the Mobile Linux
    operating system to create a robust yet economical machine that can handle all the required Internet plug-in
    applications. Mobile Linux offers an additional advantage in that it is an operating system that can be stored in
    solid state Flash ROM thus removing the need for an expensive hard disk drive.

    "Cellular phones became more pervasive when they were made smaller and provided greater battery life," said
    Dave Ditzel, Transmeta's CEO. "We believe that Crusoe will bring about a change of similar magnitude in
    Mobile Internet Computers."

    Commenting on the current state of the mobile market, analyst Martin Reynolds of the market research firm
    Dataquest, concurs with the need for a new mobile processor. "When people build mobile computers today,
    they use what's basically a desktop processor in a different package," he said. "There's definitely room for a
    fresh approach."

    "Our customers are telling us that significant battery life improvement is the most requested feature by a
    margin of two to one. That's why Crusoe's low power is so important," said Transmeta's Jim Chapman, vice
    president of sales and marketing. "The current mobile market needs to evolve from today's heavier (six to ten
    pound) laptops to lighter weight, high performance mobile PCs. Crusoe will help propel that change."

    Re-thinking the Microprocessor
    In a radical departure from traditional microprocessor design, Transmeta made innovative use of software to
    implement many functions that had previously been implemented in hardware. This approach gives Crusoe both
    the high performance and low power required for today's demanding mobile computing environment.

    The key to Crusoe's unique architecture is its Code MorphingTM software. Code Morphing software surrounds
    a simple Very Long Instruction Word (VLIW) silicon engine to deliver a fully PC compatible processor. It is
    this software that provides the compatibility by "morphing" (i.e. translating x86 instructions) to the underlying
    hardware engine.

    Crusoe is a smart processor that "learns" about an application while it runs and uses that experience to greatly
    extend battery life. Using a new Transmeta invention called LongRunTM power management, Crusoe
    continuously adjusts its operating speed and voltage to exactly match the needs of the application workload.
    LongRun can make adjustments hundreds of times per second, which can dramatically extend battery life. This
    is in contrast to other processors that run at a fixed operating speed on batteries, needlessly wasting battery
    life.

    LongRun also provides a solution for today's strenuous multimedia applications that typically drain an
    ultra-light PC's battery in as little as an hour. With LongRun, it is possible to design a light-weight mobile PC
    that plays a DVD movie for three hours or more.

    "We rethought the microprocessor from the ground-up," said Ditzel. "Crusoe is the first processor to deliver
    all three of the key requirements for Mobile Internet Computing: low power, high performance and full PC
    compatibility. Now manufacturers have the ideal solution for true mobility."

    The Crusoe Product Family
    The Crusoe processor family consists of two solutions, the TM5400 and the TM3120, for the Mobile Internet
    Computing market.

    The model TM5400 is targeted at ultra-light mobile PCs running Microsoft Windows and NT operating
    systems. These PCs will take advantage of the TM5400's high performance (up to 700MHz) and LongRun
    power management to create the longest running mobile PCs for office applications, multimedia games and
    DVD movies.

    The TM3120, operating at up to 400MHz, is designed for economical Web pads and mobile clients. With the
    Crusoe processor and the Mobile Linux operating system, users can expect a complete Internet experience,
    including access to the full range of plug-in applications. Transmeta provides Mobile Linux assistance to
    OEMs looking to accelerate their time to market with new mobile Internet devices.

    Pricing and Availability
    The TM3120, available immediately, is economically priced for Linux-based Web pads and devices selling for
    $500 to $999. The 333MHz version sells for $65 while the 400MHz version sells for $89.

    The TM5400, sampling now, will be offered in a range of performance levels from 500MHz to 700MHz to meet
    the needs of ultra-light mobile PCs selling for between $1200 and $2500. Transmeta expects that
    Crusoe-based systems with these attractive price points will be available in the marketplace by mid 2000. The
    500MHz version will list for $119, while the 700MHz version will list for $329.

    About Transmeta Corporation
    Founded in 1995, Transmeta is a privately held company based in Santa Clara, California. Transmeta develops,
    in concert with its OEM customers, platform solutions for the Mobile Internet Computing market. Transmeta
    markets and sells the Crusoe processor solution as the engine for a new class of computers. Crusoe is the
    only x86 compatible processor solution built to run the large installed base of PC software applications with
    high performance and extremely long battery life.

    --
    Your Brain + EEG + LEGO Robots = Brainstorms
  172. Slashdot on the Mobile Page by len_ux · · Score: 1

    I like the Screen on one of the mobil computer, it shows the Slashdot Site Coool

  173. Is it just me... by walflour · · Score: 1

    This thing is going to eat memory for breakfast, lunch, and dinner? Those translated caches have got to take a lot of memory. Its going to make NTs' memory requirements look trivial.
    --

    --
    When she told me I was average, she was just being mean.
  174. so cold, then... by BenHmm · · Score: 1


    if this chip runs really cold then what would be the effect of overclocking it? or can't you?
    I'm not a chip designer, so give me a break if I'm being daft


    finally! a laptop that doesn't burn my knees off! yay!

  175. Why you won't be seeing it emulate RISC by ecampbel · · Score: 5

    Emulating CISC based hardware is vastly easier than achieving acceptable performance emulating RISC based hardware. With CISC based hardware, each instruction does a lot, and might take several clock ticks to execute. The Transmeta emulator can come along and translate the instruction into equivalent Crusoe instructions, and achieve roughly comparable performance. This is why PC emulators can be run on PowerPC hardware, and why the MacOS's 68k CISC emulator was so successful for running old applications on PowerPC based Macintosh's.

    RISC emulation is vastly different. Each instruction doesn't do a whole lot, but runs extremely fast. So, basically, to emulate a RISC processor, you have to translate each instruction into one or more Curose instructions that don't benefit from RISC hardware's pipelining and other efficiencies. You're going to end up with a vastly slower PC.

    Also, this chip will not be able to emulate multimedia enhancements like 3DNow or the G4's velocity engine as fast as the native version. These chips are optimized to run the special instructions in a highly efficient and parallel manner. The multimedia enhancements in hardware basically utilize almost all the hardware technology that Transmeta has in their chip, and tehy don't have the overhead of translating intructions before being able to execute them. The Transmeta chip's software layer simply won't do as good a job as dedicated vector processing units.

    Please remember that 700mhz doesn't mean shit. You need to know how fast the processor actually runs x86 software, not how fast it runs the Crusoe transcoding software. Its the same with all emulators. Would you rush out and buy a Macintosh with a hypothetical G4 chip running at 1GHZ that only emulated x86 software? Of course not! You'd know that a x86 native processor running at 700mhz would probably be faster.

    --

    Sig goes here
    1. Re:Why you won't be seeing it emulate RISC by Esperandi · · Score: 1

      "RISC emulation is vastly different. Each instruction doesn't do a whole lot, but runs extremely fast. So, basically, to emulate a RISC processor, you have to translate each instruction into one or more Curose instructions that don't benefit from RISC hardware's pipelining and other efficiencies. You're going to end up with a vastly slower PC"

      The way a Crusoe would handle RISC "emulation" would be very very efficient when applied to RISC processors, much more so than for CISC processors. How? Well, in a RISC chip, since the instructions are smaller, there is more repetition of them. What would be four or five CISC operations gets translated into one Crusoe molecule and is then cached and optimized. When it repeats, it doesn't have to re-translate, it simply executes the optimized version that is cached.

      They've got lots of catches in the hardware of the chip that let it execute things out of order with no penalties, thereby allowing a lot more optimization. Now when they decide to re-write the code morphine module to handle a RISC instruction set it will probably take 8 or 16 RISC instructions and translate them into a single molecule which is executed in paralell.

      The same applies to 3DNow and such. When a 3DNow instruction goes to the processor it has to be translated into micro-ops and executed, and then re-scheduled. Every single time. The Crusoe would perform slower the first time this instruction is executed, but afterwards would perform much faster...

      Esperandi

    2. Re:Why you won't be seeing it emulate RISC by ecampbel · · Score: 3

      RISC code is extremely optimized for a specific processor architecture, and optimizing RISC code is a very processor intensive job. Taking highly optimized RISC code (most of the time significantly reorded to exploit parrel and pipelining efficiencies), and trying to generate equivelent instructions in another architecutre would be very slow. The Crusoe architecure is a lot better at taking a few correctly ordered CISC instructions, expanding them to Crusoe's instruction set and highly optomizing them, rather than trying to convert and optomize 8 or 16 out of order optomized RISC instructions.

      If RISC code was indeed easier to emulate, then JAVA would compile to RISC instructions, not CISC bytecodes. It compiles down to CISC bite codes so that the JIT on a particular platform can take one or two bite codes and expand them into the native instruction set of the operating system, and optomize them from there.

      Another reason, there has never been a RISC emulator that has performed at anywhere close to a resonable speed. Any emulation of the Mac done on the PC only handles 68k CISC code. The reason for this is that RISC emulation is just too slow.

      Their emulation tricks will not work on MMX or other instructions nearly so well. Curose can't make up the time it takes to emulate the instruction set by having better hardware. MMX and the G4 vector processing units utilize many of the same tricks that the entire Cursoe processor utilizes such as long word instructions, parallel processing and other optimization techniques. Also, the hardware of these multimedia units are explicitly optimized to process these special instructions. This means emulating these instructions would drastically slow down the chip.

      --

      Sig goes here
    3. Re:Why you won't be seeing it emulate RISC by Esperandi · · Score: 1

      I didn't mention this in my first post because I wasn't sure how right I am, but I assume that if Transmeta was to produce an architecture to emulate RISC, their pipelining would be re-done and such.

      But, excepting that, when taking highly optimized RISC code, assuming the Crusoe couldn't optimize it any more, its advantages wane but still exist. The 128 bit instruction words pack in 4 instructions in a clock cycle, that's an advantage there. Also, I'm not sure how many RISC instructions can go into a single Crusoe atom, but if its more than 2, that's the equivalent of doing 8 times as many RISC instructions right there.

      I can see your points about RISC optimizied code being re-optimized... its like compressing compressed data when you first look at it, but the Crusoe processor will "understand" or will "learn" the given optimizations and re-do them ONCE for its own pipeline architecture... I think the cahcheing will give it a great advantage, but I guess it will have to wait for a real world trial since the Crusoe is designed for real world applications where repetitive code that is continually re-scheduled for normal processors is re-scheduled once and saved really shows an advantage.

      Esperandi
      I'm glad you understand emulation CAN be faster though, lots of people refuse to admit that ;)

    4. Re:Why you won't be seeing it emulate RISC by ~k.lee · · Score: 2

      Emulating CISC based hardware is vastly easier than achieving acceptable performance emulating RISC based hardware. With CISC based hardware, each instruction does a lot, and might take several clock ticks to execute. The Transmeta emulator can come along and translate the instruction into equivalent Crusoe instructions, and achieve roughly comparable performance.

      In principle, this is correct. In practice, Intel engineers made a deliberate design decision in the P5 and P6 generations to spend time making the RISC-like instructions fast (using deep pipelines, register renaming, and other magic), and the CISC-like instructions slow. Implementation of rarely-used, complex CISC instructions was simply too much of a performance drag.

      As a result, modern optimizing compilers for x86 hardware generate very RISC-like code: mostly loads, stores, branches, and ALU ops. Look at the x86 assembly emitted by gcc -S sometime. Sensible compilers these days simply don't emit REP MOVSB or other behemoth CISC instructions left over from the 386 days.

      Any putative performance gain from translating "complicated" instructions to many optimized Crusoe instructions is probably marginal. Hence, your argument about the relative "inefficiency" of Crusoe with regard to translating from RISC native code is incorrect.

      BTW: for the interested, check out this Ars Technica article, which provides a fairly accessible discussion of why the RISC/CISC distinction isn't a very useful means of characterizing processors anymore.

      ~k.lee

      --
      (remove nospam for email)
    5. Re:Why you won't be seeing it emulate RISC by Omnifarious · · Score: 1

      Hmmm... VLIW instruction sets actually have the potential to seriously outperform RISC. A VLIW architecture essentially makes the processor 'multithreaded'. All the instructions in a word are executed at the same time, and if two of them affect the same registers, well, the behavior is undefined.

      This is in contrast to RISC, in which, in some sense, all the instructions in the pipeline are being executed simultaneously, but dependencies between registers and pipelines stalls from branches are handled by lots of logic that heats things up and slows things down.

      Of course, writing code that always has every instruction slot in your VLIW filled with an instruction is very hard. It's even very hard to write a compiler that will do it for you. VLIW actually very strongly resembles microcode.

      So, I'm not sure if your evaluation is totally on base here.

      --
      Need a Python, C++, Unix, Linux develop
    6. Re:Why you won't be seeing it emulate RISC by Omnifarious · · Score: 1

      Actually, the reordering to avoid register hits should help with translating to VLIW. Every VLIW molecule has to have instructions that are independent of eachother. It should be easier to collect a bunch of these instructions together from a heavily optimized RISC program than it would be for CISC.

      CISCs emulation advantage of each instruction doing a lot will be what helps optimization for those instructions. A different reason than RISC.

      --
      Need a Python, C++, Unix, Linux develop
    7. Re:Why you won't be seeing it emulate RISC by ~k.lee · · Score: 4

      The fact that ecampbel's comments in this thread have been moderated so high is proof positive that when you referee technical opinions by popular vote, you get lousy technical opinions.

      RISC code is extremely optimized for a specific processor architecture, and optimizing RISC code is a very processor intensive job.

      So is optimizing CISC code, or, for that matter, translating CISC code to VLIW and then optimizing the VLIW instructions. In fact, students of compilers and architecture history know perfectly well that optimizing RISC code is actually much easier than optimizing CISC code. One of the prime motivations behind the uniform register file and instruction set of RISC architecture was ease of compilation.

      Taking highly optimized RISC code (most of the time significantly reorded to exploit parrel and pipelining efficiencies), and trying to generate equivelent instructions in another architecutre would be very slow.

      This might have some relevance, except that modern Intel architecture code usually consists of simple, RISC-like operations that are just as idiosyncratically reordered to exploit the pipelining/superscalar efficiencies in modern Intel chips. If translating RISC code is slow, then so is translating modern x86 code. See my other comment in this thread.

      If RISC code was indeed easier to emulate, then JAVA would compile to RISC instructions, not CISC bytecodes. It compiles down to CISC bite codes so that the JIT on a particular platform can take one or two bite codes and expand them into the native instruction set of the operating system, and optomize them from there.

      Java, nee Oak, was designed for embedded applications; the original motivation for using CISC-like bytecodes (and a stack architecture, of all things) was code compactness. Ease of translation into native machine code for JITs was not one of the primary concerns when the bytecode format was selected.

      Another reason, there has never been a RISC emulator that has performed at anywhere close to a resonable speed. Any emulation of the Mac done on the PC only handles 68k CISC code. The reason for this is that RISC emulation is just too slow.

      No; the reason is the lack of market demand. The x86 architecture's relative poverty of registers, and their nonuniformity, make it rather difficult to emulate, say, PowerPC code efficiently, but adequate performance could probably be achieved.

      ~k.lee

      --
      (remove nospam for email)
    8. Re:Why you won't be seeing it emulate RISC by ecampbel · · Score: 2

      Here is quote from your own citation:
      Both the Athlon and the P6 run the CISC x86 ISA in what amounts to hardware emulation, but they translate the x86 instructions into smaller, RISC-like operations that fed into a fully post-RISC core. Their cores have a number of RISC features (LOAD/STORE memory access, pipelined execution, reduced instructions, expanded register count via register renaming), to which are added all of the post-RISC features we've discussed. The Athlon muddies the waters even further in that it uses both direct execution and a microcode engine for instruction decoding. A crucial difference between the Athlon (and P6) and the G4 is that, as already noted, the Athlon must translate x86 instructions into smaller RISC ops.

      The Pentium does this in hardware too. Both are emulating x86 instruction with RISC instructions. This is exactly how the Transmeta chip does it, except the translation happens in highly optimized software. In both cases, it's a CISC-RISC translation.

      Ease of translation into native machine code for JITs was not one of the primary concerns when the bytecode format was selected.

      You're wrong; Sun did care about performance. Here's a quote regarding SUN's Java, nee Oak, compiler from Sun's Java site:

      The Java compiler does this by generating bytecode instructions which have nothing to do with a particular computer architecture. Rather, they are designed to be both easy to interpret on any machine and easily translated into native machine code on the fly... The bytecode format was designed with generating machine codes in mind, so the actual process of generating machine code is generally simple. Efficient code is produced: the compiler does automatic register allocation and some optimization when it produces the bytecodes.

      Basically, nothing you've said has convinced me or anyone else that RISC translation is nearly as efficient as CISC translation.. Please provide an example of software or hardware that does RISC-CISC translation or even does RISC-RISC translation to support your claims that, "adequate performance could probably be achieved."

      --

      Sig goes here
    9. Re:Why you won't be seeing it emulate RISC by ~k.lee · · Score: 1

      The Pentium does this in hardware too.

      Incorrect. The Pentium, a.k.a. P5 core, does not have a RISC chip under the hood. It is superscalar and pipelining, but it does not translate x86 code into 32-bit-wide microops, as the P6 (Pentium II/Celeron) core does. Interesting how you feel free to pontificate about processor technologies when you don't know the difference between them.

      Sun writes: The bytecode format was designed with generating machine codes in mind, so the actual process of generating machine code is generally simple. Efficient code is produced: the compiler does automatic register allocation and some optimization when it produces the bytecodes.

      This is Sun's hype, not technical truth. Current JIT compilers perform only the most basic of optimizations when translating to machine code, because they operate under extreme time constraints. The machine code generated is "efficient", but only in comparison with the code emitted by a circa 1980 C compiler.

      The choice of a high-level bytecode format made naive translation into machine code very simple, but it placed optimization between a rock and a hard place. You can't optimize load/store reordering, register reuse, or anything even vaguely machine-dependent at bytecode compile time, because bytecodes don't express those operations. On the other hand, you can't perform sophisticated optimizations such as the more complex forms of dataflow analyses during JIT compilation, because you don't have enough time. Hence, optimization suffers at both ends.

      In any case, it's not clear to me how bytecodes are particularly "CISC". The term CISC was retroactively invented by RISC partisans to designate a particular microprocessor instruction set architecture, of which the 386 is the prime example. The JVM is not a microprocessor; it is a virtual machine, and there are features in the JVM instruction set that would never have appeared even in a CISC processor.

      Basically, nothing you've said has convinced me or anyone else that RISC translation is nearly as efficient as CISC translation. Please provide an example of software or hardware that does RISC-CISC translation or even does RISC-RISC translation to support your claims that, "adequate performance could probably be achieved."

      I suppose that it's difficult to use technical facts to convince those who have no technical background. However, that is not the entire population.

      Perhaps you missed my point about native P6 code emitted by modern compilers, and its similarity to RISC code. I suggest you reread it. If RISC emulation is so slow, then x86 emulation should be slow for all the same reasons. Translating modern x86 code to RISC code is, in effect, a RISC to RISC translation. And the Crusoe demo seemed to indicate that yes, it is fast.

      ~k.lee

      --
      (remove nospam for email)
    10. Re:Why you won't be seeing it emulate RISC by TheFalken · · Score: 1

      > [3dNow!] slower the first time this instruction > is executed, but afterwards would perform much > faster... Now that would be *amusing* in terms of FPS... Starts of at 5 fps, by the end of the deathmatch it's nearer 24 :-)

    11. Re:Why you won't be seeing it emulate RISC by TheFalken · · Score: 1
      > [3dNow!] slower the first time this instruction
      > is executed, but afterwards would perform much
      > faster...

      Now that would be *amusing* in terms of FPS... Starts of at 5 fps, by the end of the deathmatch it's nearer 24 :-)

  176. Technically maybe, but still a lot to prove... by loom · · Score: 1

    With all the hype surrounding this, I still think that Transmeta has a lot to do in terms of asserting itself.

    I mean sure it was a great idea to hire Linus and make a big secret about everything, but because of that people are going to be VERY sceptic about everything they are doing from now on, and somehow I don't see how they can benefit from it.

    Sure the PR stunt is achieved, but once people get down to the hard facts, Transmeta still has a lot to prove, and has NOT sold a single product that we know of in 5 years ! This is a big risk for a company, whichever size it is. All the best to you Transmeta, we'll be watching :)

  177. Strongarm is on a stoneage process by Anonymous Coward · · Score: 0

    Intel has been severely dragging its feet getting SA-2 ready...

  178. That makes no sense at all by Anonymous Coward · · Score: 0

    Read up on active powersupplies.

  179. So, it's like... by DrJolt · · Score: 1

    So, it's kinda like Digital's FX!32, but translating to their VLIW instead of alpha, and implemented at a much lower level, combined with the sort of power management that you have in a palmpilot?

  180. Re:technical GPL violation? immaterial maybe, but. by Anonymous Coward · · Score: 0

    Is this a technical violation of the GPL?

    You're forgetting IAONAL (I Am OBVIOUSLY Not A Lawyer). Linus can license Linux to anyone under any terms he wants, because he holds the copyright.

  181. CodeMorphing: borrowing from Russians? by olg · · Score: 1
    http://www.elbrus.ru/press/mprep-p2.html

    Impressive Performance Claims

    Babaian's claims for the E2k would sesm unbelievable, if not for the credibility of the Elbrus team. In a 0.18-micron six-layer-metal process, he says, the E2k will run at 1.2 GHz and deliver 135 SPECint95 and 350 SPECfp95, yet it will require only 35 W of power and 126 mm^ of silicon (with 256K of on-chip L2 cache). We protect that in a similar process Merced will operate at 800 MHz and deliver 45 SPECint95 and 70 SPECfp95 in 300 mm^ of silicon at 60 W. Merced, however, is ahead of the E2k in development by at least a year. Even more amazing, Babaian claims the E2k processor will be x86 binary compatible and, after a few tweaks, IA-64 (Merced) compatible as well. To achieve this feat, Elbrus will rely on binary compilation assisted by emulation hooks in the processor, a strategy which, not coincidentally, is similar to the tack that Transmeta is apparently taking (see MPR 12/7/98, p. 9). The similarity might arise from the fact that Transmeta cofounder and CEO Dave Ditzel spent several years at Sun working with Elbrus. Babaian believes, however, that his company's binary-compilation technology is more advanced than any other on the planet.

  182. So what?? by Anonymous Coward · · Score: 0

    Im sure extreme zealotry and fanatism dont get much worse than this. So, just because Linus is a Transmeta employee we must jump in excitment and kneel in front of the 2 processors. Gee, there must be a lot of hardware designers reading Slashdot; otherwise, Id not understand that nobodys asking about the REAL systems using these microprocessors. At least a motherboard to be able to roll your own machine. But, since its Linus, I think everyone forgot that little detail. Do you guys remember the Media processors? What about the Transputer? No? Well, that figures. Unless there are machines or at the very least motherboards using the chip, it will never be any better than a museum feature. So who are the manufacturers designing the machines and/or mobos? How long will it take for them to show up in stores? How is this different to the Microsoft hype? So, fellas, cool down and sit. Stop all the cheering. This is not the Second Coming and Linus is not a Messiah. Gee, I thought Steve Jobs was the only guy with a reality-distortion field. I know better now. Moderators: If you want to moderate something, why dont you moderate this? Let there be flames!

  183. *I* want native Crusoe apps by Anonymous Coward · · Score: 1

    Can you imagine the performance increase you'd see if you compiled directly into the native Crusoe instruction set, so it could bypass all that code-morphing (emulation) software?

    If you want a REAL paradigm shift, make the Linux compilation process transparent to the user and easy enough for Joe Sixpack's grandma to do. It would work like this: Joe Sixpack's grandma downloads a source code file, and the first time she double-clicks the file the OS automatically compiles it for her, then launches the newly-created binary. In the future when she double-clicks the same file, the OS is smart enough to just run the binary.

    She would never have to worry about which version of a binary to download (Linux x86, LinuxPPC, Linux Alpha, Linux Crusoe, etc.). Just get the source code and she's golden.

    If recompiling was this easy, then who cares that you'd have to recompile when a new Crusoe comes out. If she moves her source code file onto a machine a newer Crusoe, or a totally different processor for that matter, the first time she tries to run the app (assuming an appropriate compiler is installed on that machine), the recompiling would be taken care of.

    1. Re:*I* want native Crusoe apps by Zach+Baker · · Score: 1
      Can you imagine the performance increase you'd see if you compiled directly into the native Crusoe instruction set, so it could bypass all that code-morphing (emulation) software?

      Compiling to native instructions before execution is for wimps.
      I want to see this processor running C bytecode. =^)

    2. Re:*I* want native Crusoe apps by Esperandi · · Score: 1

      "Can you imagine the performance increase you'd see if you compiled directly into the native Crusoe instruction set, so it could bypass all that code-morphing (emulation) software? "

      I think the advantage would be very small. In fact, I'm not sure if there would be much at all. A native optimizing compiler would work just like the code morphing layer.... hmm... this is working my brain ;) On one hand, you bypass the cacheing structure, there would be a very small increase. On the other hand, if its not optimized for the pipeline as well as the code morphing layer is, there would be a sizeable performance hit. Where do you believe the advantage would come from?

      Esperandi
      Emulators aren't always slower. Look at UltraHLE. Look at UltraHLE under Wine which runs faster than UltraHLE does under Windows...

    3. Re:*I* want native Crusoe apps by odaiwai · · Score: 2

      > I want to see this processor running C bytecode. =^)

      Huh, I want to see this translate a program into several languages, compile the all and see which is the fastest language to use.

      dave

    4. Re:*I* want native Crusoe apps by cave76 · · Score: 1

      In addition to what you've said in terms of the CMS code optimization vs optimiztion gained from code written specifically for the hardware, you would lose many of the other features of the Crusoe. For example, you lose the ability to add new instruction sets as they are developed and you lose ability to adjust the cpu frequency according to the demand as the LongRun power management is able to do (unless there's some way to put this ability into hardware that I'm missing which is, of course, quite possible). In otherwards, you'd lose the ability to dynamically change the instruction set and lose the battery life, which were pretty much two of the main reasons the processor was developed.

      Mike

    5. Re:*I* want native Crusoe apps by spinkham · · Score: 2

      Actually, the performance would probably be *WORSE*.
      Why?
      The Crusoe doesn't have out of order execution.
      That is part of how the power requirements are so low..
      A lot of the circutry that other cpu's have is handled by the software driving the processor. So, you would still need a crusoe to crusoe translator to resolve dependancies and do OOO.

      --
      Blessed are the pessimists, for they have made backups.
    6. Re:*I* want native Crusoe apps by rew · · Score: 1

      So, you would still need a crusoe to crusoe translator to resolve dependancies and do OOO.

      No. You need a compiler. Some things are hard to do. For example:

      *p = a * *b;
      *q = a * *b + 1;

      here a normal compiler will have to assume that a and *b can change by the assignment to *p. It therefore will have to reevaluate the a * *b expression for the assignment to *q. However, on a TM CPU, you can ask the CPU to raise an exception on the case that that *p happens to hit on a or *b!

      There are several different areas where a compiler would benefit just as much as code-morpher does...

      But the power of the Transmeta CPUs is that they have an efficient bytecode that they pretend to interpret which in fact they compile, optimize and cache. That bytecode happens to be industry standard x86 code.... They can change the internal architecture without having to bother with a large installed base of programs. There is just one 64k instruction program that really matters: The code morpher.

      Roger.

  184. Re:technical GPL violation? immaterial maybe, but. by Robert+S+Gormley · · Score: 2

    And obviously neither are you, because Linus only owns the copyright on what he wrote. He didn't write the OS himself. Other people have copyright on their contributions, and licensing of their own. Just how do you think he would retroactively be able to change *someone else's* licence?

    --

    Open Source. Closed Minds. We are Slashdot.

  185. lithium batteries by Redundant() · · Score: 1

    AA Lithium batteries will probably work with these chips. They can handle 1 watt burst loads without heating up and browning out. I am assuming these chips were developed for AA battery powered devices, it is too big a market to ignore. I Wonder if the price of lithium batteries will come down?

  186. What a disappointment! by jms · · Score: 1

    A closed chip, running an object-code only, proprietary compiler, translating to an undocumented machine language, all to emulate an X86 chip, oh, and "other architectures", if the high priests at Transmeta decide to grace us with interpreters for them.

    Of course, once Transmeta writes an interpreter for a certain instruction set, there is no guarantee that they will support that instruction set for future chips.

    How nice of them to "protect" us from having to recompile by simply forcing us to use the obsolete X86 instruction set.

    I'm so underwhelmed.

    We've fought too long and hard to get away from proprietary, closed-source hardware and software to go back now.

    This might have been "cool" 10 years ago.

    I Think I'll pass. Thanks for trying.

    All that hype for nothing.

    1. Re:What a disappointment! by juri · · Score: 2
      I don't know, perhaps that was sarcasm and I'm just being thick, but that last line in the article just happened to struck a nerve; it's one of my recent pet peeves: "All that hype for nothing."

      Exactly what hype are you talking about? Yes, the press has been talking about the stuff and people on slashdot have been speculating far more than they should have, as usual.

      But there has been absolutely no hype, at least none generated by Transmeta. Of course, you could argue that their marketing department is just smarter and stealthier than that of most other companies. It might even be true, but still, all the so-called hype that has reached the consumers has been speculation and rumors. Not anything that Transmeta has said directly. You can hardly argue that Crusoe has been vaporware or something.

      It's quite possible that you expected more from Crusoe. But I think it's kind of silly to hold Transmeta responsible for not living up to expectations people have built up based on rumors.

      // Juri

    2. Re:What a disappointment! by jms · · Score: 2

      You're right. I didn't phrase that well. There's been an awful lot of hype from the rest of the community ABOUT what Transmeta has been working on, and that's what I'm referring to.

      I was just expecting more then a low-power x86 clone, and that's all these chips are to the end user.

  187. Software upgradeable... and downloadable code by loftwyr · · Score: 2

    So, since we could get updates to the CPU off the web as they said in the presentation, what's stopping people from creating a program that turns my Crusoe chip into junk a.k.a. a Crusoe virus.

    How do I fix a chip that doesn't have any instructions in it?

    1. Re:Software upgradeable... and downloadable code by sterwill · · Score: 2

      Was I the only person who read the technical white paper that clearly said the Code Morphing (TM) software would be stored in ROM (Read Only Memory, for those new to computers) with each revision of the chip?

      --

  188. Why they don't give out the instruction set by Dr.+Sp0ng · · Score: 1

    Jesus people, quit bitching about something that you don't know anything about. Transmeta won't open the instruction set for the Crusoe because they want to have the ability to change it from CPU to CPU. That way they don't end up building up tons of obsolete crap (like Intel from the 8086 days) They can start each processor with a fresh design, and the only software that needs to be ported is the Code Morpher thing. It's not an anti-OSS thing like you people are making it out to be. Hell, if they were doing that, Linus probably wouldn't be working for them.

    It's for the same reason that they don't give the information needed for random people to write morphers - they would end up with things like "SPARC morpher, only for the XXYYY version of the Crusoe." It makes sense for them to only let people write software for the other CPUs and have it run emulated (especially since there seems to be such a small loss in speed) They want the freedom to design each new chip as they see fit without building up tons of backwards-compatibility crap, the lack of which is exactly what allows the Crusoe to run so cool and use a such a little amount of power (as well as make the chip cost so little to make)

    Sorry if this post doesn't make sense or rambles, I haven't slept in 2 days. But I'm sure you can get the main idea.

    "Software is like sex- the best is for free"
    -Linus Torvalds

    1. Re:Why they don't give out the instruction set by jms · · Score: 2

      It isn't that I don't understand their argument, I just don't accept it. It's the same old closed-source, proprietary argument.

      Here's the software analogy of your argument.

      "We don't give out the instruction set because we don't want to give the information needed for random people to write compilers. They would end up with things like "Linux, only for the Sparc." It makes sense for them to only let people write software for interpreted languages like BASIC, or any other future languages that we decide to allow people to write in."

      What they are really doing is locking people out of the market for interpreters for other architectures. Want a morpher for the IBM 390 instruction set? Transmeta decides whether or not you get it. If it doesn't suit their business plan, you won't get it.

      This is the antithesis of the open source philosophy. You can't "scratch your itch" if you're locked in handcuffs.

      This is just a cheap, low power x86 clone. All the "cool stuff" is meaningless if it's not available to the interested programmer.

      As I said, underwhelming.

      - John

    2. Re:Why they don't give out the instruction set by Benabik · · Score: 1

      Looks like, to me at least, that Transmeta is starting out the safe way. They don't yet want to do the really cool stuff /.ers want to because that's a risky market. Maybe in a year (if they're still there) they'll release a processor that allows you to switch ISAs on the fly. Untill then it's kinda pointless, since the Crusoe's VLIW set might change before then.

  189. TM vs. TPM -- worth the wait by ocie · · Score: 1

    Two things that have been hyped by secrecy. I was really looking forward to both The Phantom Menace and whatever Transmeta was working on. So far, I am impressed. Now where can I get one?

    --
    JET Program: see Japan, meet intere
  190. What does this mean for compiler optimization? by RebornData · · Score: 3

    I took a compiler class in college, but only remember enough to be curioius, and am hoping someone who knows what they're talking about could answer.

    Compilers do a lot of compile-time reasoning to try to predict what happens at run time, optimizing both register use and instuction order. However, this is not done solely based on the ISA- there are significant assumptions about how the processor actually executes instructions.

    Crusoe, in optimizing execution, has the benefit of *knowing* what's happening at run time, as opposed to a normal compiler that has to guess about it at compile time. Also, because the x86 ISA is implemented in software, they could do all sorts of crazy stuff beyond instruction reordering. For example, if the compiler guessed wrong about what set of values to keep in the registers, the code morphing software might be able map hardware registers in Crusoe to values that were being sent to memory in the original code.

    Which is all well and good. But given that there now are two astonishingly different execution models for x86 code, what are compiler writers supposed to do? Is the amount of information that a run-time optimizer has so much better than a compile-time optimizer that the compiler guys should just give up? Or do the constraints on the complexity of run-time algorithms (they've got to be *fast*) mean that compiler optimization is still worthwhile? If so, how will compiler authors cope with the potentially vast differences in optimizing for Intel vs Crusoe? Will we be seeing a Crusoe-opimization flag in gcc?

    More questions than answers...

    1. Re:What does this mean for compiler optimization? by EvilKevin · · Score: 1

      Well the compiler guys certainly shouldn't give up. You've touched on a good point though. Many potentially profitable code transformations cannot be performed statically because it is impossible to prove they won't alter a program's semantics. Often, the transformation is prevented because the value of some variable at a given point in the program can't be determined. Think aliases.

      The Crusoe code morpher has an entirely different problem. It must make optimization decisions based solely on a local set of machine instructions. It doesn't have any of the high level information about the program's structure -- and some of this potentially useful information can't be inferred directly from the instruction stream. So, yes, the Crusoe code morpher has more information (values of variables, etc.) and less information (types, etc.) than the static compiler.

      So it makes sense that static and dynamic optimization should benefit from one another and each exploit its own strengths.

      - Good static optimization can reduce the amount of work that the dynamic optimizer has to do, thus yielding better performance (through lower translation latency) and reduced power consumption.

      - Static optimizations can perform translations on larger chunks of code than the Crusoe code morpher. For instance, through analysis it may be possible to show that a loop can be reordered to improve cache performance w/o altering the original program semantics.

      - Static optimizations can be more agressive since their cost is amortized across all program executions.

      And there are many other arguments for good static optimization technology.

      Now, I imagine that Transmeta has arranged it so that codes optimized for Intel processors run well on Crusoe under morphing. But I also imagine that Transmeta might publish a set of guidelines for compiler writers that suggest how one might generate better code for Crusoe targets. The trick will be doing so w/o exposing too much information about their top secret morphing technology.

    2. Re:What does this mean for compiler optimization? by Trent+Took · · Score: 1

      > Crusoe, in optimizing execution, has the benefit of *knowing* what's happening at run time, as opposed to a normal compiler that has to guess about it at compile time

      But this is not insurmountable. I know of at least one compiler that is available in the embedded market where you can optimize based on real execution results with an instrumented compile.

      The clever thing I've seen is the talk about dynamically changeing the power drain (by changing speed) depending on how loaded the processor is. How does it know if its is running an idle loop or a boring iterative calculation?

      --
      See old Atari's come to life stonx.sourceforge.net
  191. WHY WOULD ANYONE WANT TO RUN NATIVE?!?!?! by Tjl · · Score: 3
    I don't understand why everyone wants so much to run on the "native" 128-bit VLIW.


    The point is, if you compile C code for the VLIW, it will likely run slower than the same code compiled for the x86 architecture and then run through the code morphing software.
    The reason for this is simple: read their tech whitepaper. In it, they talk about memory protecting for out-of-order loads and stores.
    If you have C code like

    b=a[d]; *c=5; h=a[d];


    then this code will run slower unless you can do the out-of-order check and run-time recompilation: if you compile it yourself for the 128-bit VLIW, you lose: your C compiler is NOT ALLOWED to perform the same optimizations since you could get incorrect results.


    The really new innovation by Transmeta is those small bits of simple hardware that allow them to trap violations of optimization assumptions, not the 128-bit VLIW part: the latter could have been done by anyone at any time.


    Well, stopping the rant now and beginning to wonder when I can get my hands on one ;)
    With the cheap and non-heating parts, you could build a pretty big beowulf or SMP machine inside one tower case --- now that would be something.

    1. Re:WHY WOULD ANYONE WANT TO RUN NATIVE?!?!?! by Anonymous Coward · · Score: 0

      Don't drink the marketting cool-aid... yes there are cases morphing can lead to less instructions which will of course be shown in material produced by the company :-) In general they execute less then 1 x86 instruction for every p95 instruction so a 400MHz Crusoe is like a 233MHz Pentium class machine running standard software. When independent test labs do the work you will see. Xtransmeta

    2. Re:WHY WOULD ANYONE WANT TO RUN NATIVE?!?!?! by earthy · · Score: 1

      Actually, if you were to compile C code to the 128 bit VLIW of the Crusoe, you *could* be benefiting from the runtime aliasing, as you would also have access to the ldp (load-and-protect) and stam (store-under-alias-mask) instructions. The problem however is in the fact that Transmeta reserves the right to change the instructionset underlying the CodeMorphing software with each and every revision of the processor. And that would mean rewriting the compiler...

  192. Do you mean "Celsius or Fahrenheit"? by mangu · · Score: 1

    Get it? Cel s ius and Fa h renheit.
    Anders Celsius, swedish astronomer, born in 1701 in Uppsala, Sweden, died in 1744 in Uppsala.
    Gabriel Daniel Fahrenheit, German physicist, born in 1686 in Danzig, Germany, died in 1736 in The Hague, Holland.
    Each invented the temperature scale that bears his respective name.

    1. Re:Do you mean "Celsius or Fahrenheit"? by Anonymous Coward · · Score: 0

      Spelling flames are anal, bud.

  193. Yes, in fact, we said it here last September! by srp · · Score: 2
    This was the first question that the person from Slashdot (forgive me, I can't remember who it was). The response was there was a demo there (not shown on the ZDNet broadcast) of a machine running X86 byte code and Java bytecode natively, at the same time.

    This is much along the lines of what we discussed here in a post from last September. See the Chips that change on the fly thread (starting at post #44) from September 23rd.

    Too bad they didn't give out prizes for guessing right. :-)

  194. I hope this is as cool as it sounds by Anonymous Coward · · Score: 0

    I've been putting off getting a Linux compatible portable for about a year. Now I finally know why.

    Anyone have any ideas when we can start seeing machines running these?

  195. Where's Linus? by Anonymous Coward · · Score: 0

    http://www1adm.transmeta.com/im ages/company_large.jpg

    It's like Where's Waldo, except more geeky!

    1. Re:Where's Linus? by Anonymous Coward · · Score: 0

      Didn't you notice the Copyright 2000 by Linus Torvalds under the picture? He took it!

  196. Very interesting! by Anonymous Coward · · Score: 0

    It seems to me that Transmeta has choosen exactly the right mix of design , speed , power consumption. And marketing it at exactly the right time. Handheld market is hughe. Waiting for Nokia:s next product series :)

  197. Business Strategy by Anonymous Coward · · Score: 0

    holy shit! you mean IBM has a real business strategy and a strategic vision of the future?

  198. The Athlon!!! by Anonymous Coward · · Score: 0

    The athlon technically does emulation of the x86 instruction set, allbeit 1/3 of the transisters are dedicated to it. Really no modern app uses it anymore, at least they use very few x86 instructions. They all use mmx, sse, 3dnow and other non named sets that have been added, mmx, sse, 3dnow are all simple instructions, even though as a set there not really a "reduced instruction set" but I like the way processors are going, all simpler faster instruction and alot of them to keep programming relatively easy

    1. Re:The Athlon!!! by Anonymous Coward · · Score: 0

      that's not true. most code is x86, sse and mmx are only marginally used

    2. Re:The Athlon!!! by Anonymous Coward · · Score: 0

      You technically don't know what you're talking about, do you, O Great One?

    3. Re:The Athlon!!! by Anonymous Coward · · Score: 0

      Absolute rubbish. Where did you get this information? Virtually all software still works on my old 486 (albeit slowly!) and therefore uses purely commands in the x86 instruction set. The only possible exception is the latest 3D-accelerated games.

      ac.uk

    4. Re:The Athlon!!! by Anonymous Coward · · Score: 0

      Mmx,sse,3dnow are all used very little and only in massive floating point calculations (3d graphic or signal processing). The old (386) instruction set is still very used althroug modern compilers generates only instuctions that are known to be the fastest for a given operation on the processor they are targeted. So, gcc when targeting a 486 generates the "leave" instruction while when optimized for pentiums uses "movl %ebp,%esp ; popl %ebp". With time most exotic instructoins (bcd,...) remained only for compatibility with older applications. New compilers generate equivalent code using a much restricted set of instruction, when they find it appropriate (for pipelines). But this has nothing to do with MMX and its friends. Take a look to the assembler output of a c program (linux kernel) and count how many MMX instruction you find. Also count how many different instructions there are. Ix86 still isn't a risc inst.set not because the quantity of instruction but because of rich adressing modes that requires complicate decoding. A riscish machine will have separate instructions for loading and storing registers in memory and all other operations are done in registers.

  199. Will this stunt Portable Linux on other platforms? by wildernapt · · Score: 1

    We now know that Mr. I-Decide-What-Gets-In-The Kernel (Mr. Torvalds) is a partisan player on one particular low-power hardware platform. Will this come to mean that kernel development for other low-power platforms (the StrongARM comes to mind...) will suffer? I certainly can't see Transmeta being overly enthusiastic about paying Linus for further optimisation-for-portability changes that benefit their competitors.

    Maybe it's a moot point, though, as NetBSD is a more finished OS for the StrongARM anyway, and for embedded work doesn't drag along the GPL baggage.

  200. Not Impressed by Anonymous Coward · · Score: 0

    Gee, this is almost exactly the same thing I had to do 30 years ago - load the microcode into the old IBM 360/30. Big difference is that this is the size of a postage stamp rather than the size of a VW bus. (Also I don't think you'll need rotary switches to suck the code in off of 8 inch floppies) On the other hand, it might be a good platform to build a system that runs JAVA bytecode. I can see why they're targeting x86 stuff, but hopefully they'll eventually do something interesting..

  201. Virus warning. by Anonymous Coward · · Score: 1

    As much as im amased over the processor im thinking of the possibility of upgrading hardware with software. We know what happens then dont we (the CIH virus). Virus writers will definately make atleast one virus that will change the instruction set or maybe even infect the processor itself (is possible, but probably hard) as virus writers love these kind of stuff. If not them a skilled hacker och cracker (reverse enginerer) will figure it out and then its not long until a virus is developed for it. The only thing lowering the risk is exactly what they are doing, a different instruction set for every processor. /Bhunji

    1. Re:Virus warning. by Anonymous Coward · · Score: 0

      What's weird is now it's probably possible for the virus writers to make things even more subtle - stuff like 2 + 2 =4 but 3 + 2 = 1

  202. Code morphing by Nyarly · · Score: 1
    I hate it when what looks like a very general purpose technology gets reduced to what's most profitable. Because very often something will be advertised as less than it might be, and never gets to be what it could be.

    To wit: the Crusoe chip is x86 compatible. Goody. Don't run windows, couldn't care less about clock-speed obsessed, power hungry Intel/AMD chip designs. But code-morphing makes it sound as if I could throw Solaris on the little bastard. Or OS X. Or BeOS. Hell, I might even boot to a Guru Message. :)

    I mean: that what it sounds like. Is that possible? Is it reasonable to expect? If so, I need new shorts.

    If not, ho hum, another Intel knockoff. Nice power consumption, and some over-time optimization, but nothing spectacular. Which would really start to stunt my inner geek child.

    --
    IP is just rude.
    Is there any torture so subl
    1. Re:Code morphing by Guy+Harris · · Score: 2
      But code-morphing makes it sound as if I could throw Solaris on the little bastard. Or OS X. Or BeOS.

      Definitely true for Solaris and BeOS, given that there are versions of both of those OSes that run on the instruction set from which Transmeta's binary-to-binary translation code translates.

      MacOS X is another matter, unless Apple release an x86 version of it.

      Perhaps the underlying VLIW engines on the two chips aren't so specialized to serve as targets for translation from x86 that SPARC-to-native or PowerPC-to-native compilers could be written, but I wouldn't hold my breath waiting for those translators - I'd punt on MacOS X and get x86 versions of Solaris or BeOS instead.

    2. Re:Code morphing by Nyarly · · Score: 1
      MacOS X is another matter, unless Apple release an x86 version of it. Or, more amusingly, a Crusoe version of it. Just because x86 will go to Crusoe doesn't mean you have to still code for the x86. And Apple has no incentive do so. Perhaps the underlying VLIW engines on the two chips aren't so specialized to serve as targets for translation from x86 that SPARC-to-native or PowerPC-to-native compilers could be written, but I wouldn't hold my breath waiting for those translators - I'd punt on MacOS X and get x86 This is my point though. If all it will do is be a Crusoe and fake an x86, that isn't very exciting. If the morphing is such that the ability to write a compiler to Machine X -> the ability to write a Morph from Machine X to Crusoe, that's incredibly exciting. Why buy any other processor? Apart from cost.

      For instance, Java to Crusoe is then reasonable. And, if it's all the whiz-bang it sounds like, then it could probably optimize the safeguard redundancy away and make Java a realistic application development platform.

      To summarize: Low power is nice, but the G4 does a gigaflop on 4 watts. Code morphing is where the Crusoe is at. Or if it isn't it should be, or else it's a downhill run to a non-event. Linux or not.

      --
      IP is just rude.
      Is there any torture so subl
    3. Re:Code morphing by Guy+Harris · · Score: 2
      Or, more amusingly, a Crusoe version of it.

      What would "a Crusoe version" of a platform be? Something built to run on a particular Crusoe chip (the two chips Transmeta have mentioned apparently do not have the same instruction set)? Or something built for some instruction set implemented atop various Crusoe chips (with different translators)?

      Low power is nice, but the G4 does a gigaflop on 4 watts.

      ...but doesn't run Windows. If the market for low-power machines that run Windows (or other OSes not available for mobile machines with other types of processors) is sufficiently large, and if other x86 processor vendors don't get (as others have said) "good enough" power consumption, and if a lower-power CPU makes enough of a difference if the other components have the same power consumption no matter what processor is used, then they might have a chance. Time will tell.

      As for the ability to run multiple instruction sets on the same machine, that may be of interest, at least for some OSes, only if you can run in a "virtual machine" environment, so you can run Windows and MacOS on the same machine, unless Transmeta relase to the vendors of one or the other of those OSes enough information to let them use the instruction-set switching directly and those OS vendors are willing to modify their OS to be able to run applications for other processors/OSes without just running the entire other OS inside a virtual machine.

  203. No, they're releasing the source by Dominic_Mazzoni · · Score: 1

    According to their FAQ, they are planning to release the source, but they are not getting into the distribution business.

    1. Re:No, they're releasing the source by MattMann · · Score: 2
      when they give the code to their licensees, they are in the distribution business according to the GPL.

      As I said above, source has nothing to do with the objection I'm raising. It has to do with putting restrictions on redistribution by your licensees.

  204. Old Technology Renews by Anonymous Coward · · Score: 0

    Not to slight their work as the chip sounds wonderful, but the concept of dynamicly changeable microcode isnt new... DEC did it 20 years ago with the pdp8 ( i belive ) ... AMD and TI had their bit slice components at least as long ago if not longer... but its still great chip work with the silicon.... i wonder if portable linux is gpl?

  205. The tense wait could have been avoided by forensic · · Score: 1

    Am I the only one that found Transmeta's patents in the US Patent database? http://members.cotse.com/newz/stupi d_media.html

    1. Re:The tense wait could have been avoided by orcrist · · Score: 2

      No.

      But, you're probably the only one who didn't notice that those patents have been discussed, analyzed, rehashed, etc. for months in just about every technical publication on the web, as well as the subject of several Slashdot articles :-)

      Chris

      --
      San Francisco values: compassion, tolerance, respect, intelligence
  206. I was hoping that the Transmeta chip... by Anonymous Coward · · Score: 0

    would be able to turn Lacey Chabert, Natalie Portman, Jessica Alba, and Britney Spears TO STONE!!

    Why couldn't they do this???

  207. Re:I guess variable CPU speed Linux problem solved by Zurk · · Score: 1

    nope. from what i gather this just goes to sleep and back up again at full speed with noops/hlt for power saving. variable speed asynchronous machines cant be handled by linux due to its bogomips setting..basically it loops when booting up and stores the cpu rating which it uses for timing. Linux cant handle asynchronous machine architectures yet (and indeed no solution was presented other than checking the do_hardware_timer() in the last kernmel discussion on the subject - check the mailing list - and this solutions was rejected since it was too costly in terms of cpu time). BTW, i'd love to know whether this chip can run the java VM at full CPU speed - i'd love to crunch my servlets on a non dedicated java vm emulating processor which apache runs x86 instructions as usual.

  208. Maybe not anti OSS, but it is pro closed source by Anonymous Coward · · Score: 0

    OSS software does not suffer that much from architectures not being backwards compatible. Designing a processor is a couple of magnitudes more work than writing a compiler for that architecture... and if you have the source and a compiler there is no need for backward compatibility.

    If thats truly their only concern why not make the specs available with a license mandating use of the information be constrained to developing OSS?

  209. Compiler can generate code to do that too... by Anonymous Coward · · Score: 0

    No it cant be done at compile time... but whats to stop the compiler from generating code wich uses the same mechanism to change the code during run time?

    Anything wich can be done with x86+emulator can be done with code generated by a native compiler, but not vice versa.

    1. Re:Compiler can generate code to do that too... by Tjl · · Score: 1
      Not really: there are so many different assumptions that might fail that generating code for all the combinations would be impossible.

      Of course, the compiler could compile into some sort of intermediate format from which at run time the optimizations are done and discarded as the program runs... but then you've suddenly completed a full circle: that's exactly how the Transmeta CPU runs x86 and Java code and is not what you refer to as "running native"

      One may argue that x86 assembly is not the optimal bytecode for something like this. However, given the existing body of software and compilers, it is an obvious choice.

      We *may* see (if the speed advantage thus obtained is large enough) a special "transmeta byte code" in the future but it's not that likely.

  210. Why not solve the problem another way? by IpSo_ · · Score: 1

    The way I understand it, is Transmeta got stats from the general "mobile computing" public,
    which stated people were not happy with the length
    of time batteries last in there "computing appliances". (Duh!) Why not, instead of investing 5+ years and tons of money etc... solving the problem for only the "mobile computing" market, solve the problem for every battery user in existance by making a better battery?

    This would make more sense to me :)

    --
    Open Source Time and Attendance, Job Costing a
    1. Re:Why not solve the problem another way? by hesiod · · Score: 1

      entre Motorola
      ---------
      Thus Spake Dave
      Meine Hühner lachen Nicht!

    2. Re:Why not solve the problem another way? by orcrist · · Score: 2

      Why not, instead of investing 5+ years and tons of money etc... solving the problem for only the "mobile computing" market, solve the problem for every battery user in existance by making a better battery?

      What do you think companies have been trying to do for years now? Progress has been made, sure, but longer battery life can be achieved at 2 sides of the equation:

      1. More power
      2. Less power-consumption

      Now if you're a processor designer and you've got a team of engineers who specialize in processors, which side are you going to focus on? Furthermore, they benefit immediately from the side-effects:

      1. Less heat (I like the idea of not having a lap-warmer in summer)
      2. Less noise (no fan? or at least a quieter one?)
      3. Less weight

      Leave the battery design to people who specialize in that sort of thing.

      Finally, I have to admit, saving power appeals to me much more than packing more and more power (a.k.a. heat-emission) into a device; think green :-)

      Chris

      --
      San Francisco values: compassion, tolerance, respect, intelligence
    3. Re:Why not solve the problem another way? by deaddeng · · Score: 1

      If you think that the Transmeta chip only has implications for notebooks, think again. Internet aware cell phones; PDA's; remote monitoring equipment; embedded controllers. Virtually any device that requires low power consumption, low heat, and remote administration can adopt this technology. The fact that a remote administrator can update the software instruction set--never mind that it can execute x86 code--is going to drive people nuts. The telecom, power generation, petroleum, and many other huge industries are going to like this a great deal, as will the US defense and intelligence communities.

      --
      --- .085 as cool; proving that a little knowledge is dangerous
  211. x86 only?? by Anonymous Coward · · Score: 0

    from reading some of the white paper on the chips it sounded to me like it would be difficult for them to use other instructions besides x86 for one thing the system bus, PCI/memory are geared for x86, would it be possible to run PPC instructions when your core logic set is geared for x86? is this something the code morphing software would handle, or would you have to have an OEM intergrate different core logic sets (like sawtooth mobo equivilant) into the device and then be able to run PPC (or other) instruction sets?

    1. Re:x86 only?? by Anonymous Coward · · Score: 0

      You can do others... gee in all the information on the web they don't mention the second sdram interface that is used to store the morphed code that doesn't fit in the cache. Anyway IBM had a Java compiler to P95 oops Crusoe but canned the project when they dropped Transmeta as a partner and just became a fab... Xtransmeta

  212. Re:Interesting. This should advance mobile computi by Anonymous Coward · · Score: 0

    werd

  213. summer in the city (Offtopic, but oh well) by CmdrPinkTaco · · Score: 1

    I live in the Las Vegas area also and know EXACTLY what this AC means about high ambient temps. In my room I have 2 computers, a large stereo and a 27" TV. With all of these on and in full power they have been known to raise the temp of my room as much as 12F. During the summer here this can be quite annoying since the A/C doesn't always cool the house down to comfortable levels during peak temp hours (about 2:00pm until 5:30pm). Writing code in my room can be quite a perspiring experience. I think that a desktop grade Crusoe will be very nice, and it will help me keep my elec bill down at the same time.
    --------------------------------------------

    --
    Please give your mod points to others, Im at the cap. They will appreciate it more
  214. handheld's dont need windows by Anonymous Coward · · Score: 0

    If you dont need windows you dont need x86. If you dont need x86 you dont need Crusoe. If you dont need Crusoe's you can consider yourself lucky.

  215. Man Friday ? by Anonymous Coward · · Score: 0

    Any bets as to what that could be ? The Miscreant

  216. (ignore this language flame) by Jamie+Zawinski · · Score: 2

    Inoshiro wrote:
    Ahh, the penultimate question of the ages

    ``You keep using that word. I think that you do not know what it means.''

    Penultimate is not just a fancy way of saying ultimate. It means second-to-last.

    1. Re:(ignore this language flame) by Inoshiro · · Score: 2

      "Penultimate is not just a fancy way of saying ultimate. It means second-to-last."

      Would you, in this day and age, ask if it ran Quake first?

      I'd first ask if it ran Quake 3: Arena, then Unreal Tournement, etc, on down the list.. Until we hit Quake. After Quake, I'd ask if it ran Doom. So, it really is second to last ;-)

      [Being flamed by you is nifty, kinda like that 48 hour period where I read your all of website]
      ---

      --
      --
      Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
    2. Re:(ignore this language flame) by odaiwai · · Score: 2

      > ``You keep using that word. I think that you do not know what it means.''

      ITYM "You keep using that word. I do not think it means what you think it means." HTH, HAND.
      From tPB, of course.
      dave, being a pedant.

  217. Man Friday ? by Anonymous Coward · · Score: 0
    Try that again...

    Any bets as to what that could be ?

    The Miscreant

  218. Re:Interesting. This should advance mobile computi by Anonymous Coward · · Score: 0

    to your mama

  219. But We Digress.... by Anonymous Coward · · Score: 0

    'nuff said.

  220. could you be missing the point? by Anonymous Coward · · Score: 3

    firstly, two points to think about:
    1. the chip is not only more efficient but also colder, hence you save also the fan (power, noise, moving parts).
    2. there is such a thing as enough power. we have passed the stage where you wait for applications - now they wait for you. this will power application machines. I seriously doubt it will power games, graphics, etc' machines.

    now for the neat part ("paradigm shift", I guess):
    computers have been stuck in two ways till now - one is that the smallest unit is still the pc, and the other is that the possibility of backwards incompatability has kept us stuck with x86-intel-microsoft, er, technology ;-).
    what we get now is there is another layer - a "transformation layer" between the hardware and the software, allowing improvement in the hardware without losing compatability and also running code from different architectures together. this is a step forward, and a paradigm that have definite advantages that should allow it to survive and compete, despite being slower and (probably) not doing risc well.
    the second neat thing (which actually refers to my first "being-stuck" point) - this could possibly realize the niche of smaller-than-pc computers. until now it just didn't work well enough, can we now expect a small notebook/diary/phone/organizer/whatever to be the "common pc" (let's say for every member of the family/business) with fewer, more powerful computers being the centers of local networks and/or doing the real heavy work (and games). well - just think of the possibilities...

    this is a new step, all you are saying is that it is possible to not use it well - but just think of what can happen if it is!

    the AC $0.02, hope it was worth your time...

    1. Re:could you be missing the point? by LinuxParanoid · · Score: 2


      Your points are fair ones. Transmeta has a shot at opening up a whole new category of low-power, low-cost "mobile internet" as they call it PCs.

      To sum up my basic concern in a single sentence: I don't see enough barriers to entry to prevent Intel from crushing them in 3 years with a "good enough" power-optimized x86 solution. Does Transmeta have a strategy to deal with this?

      --LP

  221. Yes, StrongARM can play movies by Anonymous Coward · · Score: 0
    The SA-1100 is absolutely fast enough to play MPEG-encoded movies. I've seen it.

    --Pat / zippy@cs.brandeis.edu

  222. Sounds nice, but I can do all this with my Psion. by Colin+Smith · · Score: 1

    ARM can pretty much do all this anyway, though not 128bit (does it matter). The bonus I suppose is the Ix86 compatability.

    BTW The ARM share price went mental today...

    --
    Deleted
  223. This is a sell-out by Linus. by Anonymous Coward · · Score: 0
    I am incredibly disappointed. Here was a great opportunity to REALLY change the entire chip industry, and the mighty Linus has struck out. And sold out.

    Sigh.

    What I'm talking about is GPL'ing (no - not Open Sourcing) the CPU. That really would've kicked Intel in the pants, the same way that Microsoft is being kicked by Linux.

    But no, in the end the mighty Transmeta has the usual closed solution.

    I'm amazed that someone of Linus' supposed image could even be associated with this. But I guess with everyone else making huge bucks with Linux, he decided to sell out for the quick buck.

    Perhaps someone will hire RMS, and really kick butt in the chip industry.

  224. suppose they codemorphed mesa? by rogerbo · · Score: 4

    You can run quake 3 in software mode under mesa as you say at about 3 frames per second.

    But you're missing something here. This is transmeta we're talking about and that was Dave Taylor, the SAME dave taylor that once leaked a document onto usenet ranting about the inferiority of hardware graphics accelerators and that what he really wanted was a generic parallel processing chip that could do arbitary transforms.

    (anyone got the link to that usenet posting on deja that dave taylor tried to cancel?)

    GEE, a lot like the crusoe chip can do?

    Isn't it feasible that they have put hooks into their code morphing software that optimises specially for 3d transforms and mesa/opengl?

    Especially in the linux version? Where they have all the source code to linux and mesa?

    Hmm, what fancy optimisations could those clever brains come up with.

    Maybe those transmeta laptops WON'T need 3d accelerator ships?

    And it would completely defeat the purpose of a low power laptop to put a big,hot,power sucking 3d chip in it. So I'm assuming that demo of quake3 they showed WAS running in software mode.

    Someone prove me wrong?

    1. Re:suppose they codemorphed mesa? by Jburkholder · · Score: 1

      Well, that's a really good point and you may very well be right. I'm willing to be very impressed. I didn't mean to come off skeptical if that's how I sounded.

    2. Re:suppose they codemorphed mesa? by Anonymous Coward · · Score: 0

      There's one flaw to your theory. If they did all this, why would they be quiet about it? It's not something to be ashamed about...

  225. The Clock Speed Race by Esperandi · · Score: 3

    Transmeta doesn't really have to participate in the clock speed race (BTW, if you factor price into it, Intel isn't even close right now)... A 700MHz P3 and a 700MHz TM5400 CPU do not perform the same. In a repetitive section of code (which is most of the code being executed in real world applications), the P3 just executes the code over and over re-doing the instruction translation, the scheduling, and all those things. The TM5400 "learns" what the code is doing and optimizes it down to the point where it executes that code at the same speed when stepped down to 400MHz as the P3 700 does running at full speed and full power consumption.

    The example they give is a DVD player, by the time the first video frame has loaded the entire DVD-decoding process has been translated, stored in a translation cache (so it is never translated again), and optimized. If it continued running at 700MHz it would be performing like a 1GHz P3, but instead it steps down intelligently. If stepping down would lose frames or slow execution speed, it wouldn't do it... so until code really really needs over 700MHz in real world applications, Intel will not have outpaced Transmeta. Ithink they'll be around for awhile. And I don't expect this to be the only thing they do. They say they're going to be a professional R&D lab, i doubt they're just goin to keep adapting these processors to different things, they'll move on...

    These facts eliminate a few of the bad things you pointed out... like what is the point of integrating 3D if all of the 3D instructions are cached and optimized? There wouldn't be much of a speed improvement if any....

    Esperandi

    1. Re:The Clock Speed Race by Anonymous Coward · · Score: 0

      so says the marketing department. wait till someone not associated with the company gets to play with one.

      If you believed Sun's marketing, HotSpot is faster than C++ for exactly the same reason Transmeta gave.

    2. Re:The Clock Speed Race by LinuxParanoid · · Score: 2


      I'm OK with most of what you say up until the last paragraph. 3D circuit design is a very complex beast. More complex than any other digital circuitry save CPUs, IMHO, and that's saying a lot. If you understood all the stages of the graphics pipeline and how they could be executed in parallel, you'd understand that it's a lot more complicated than saying you can "cache and optimize the 3D instructions."

      If your "3D instructions" are SSE or 3DNow, sure I agree with you, but if you're talking about integrating and caching/optimizing the actual hardware geometry transformations and lighting code and various rasterization stages in a flexible programmable way, that's a huge multi-year architectural project. The difference in datapath design between CPU circuitry and 3D circuitry is enough to require wholly different sets of optimization techniques. It'd be easier to just slap a fixed 3D ASIC design in one corner of the chip, although once you're doing that, you're losing much of the benefits of the Transmeta approach. Plus there is analog circuitry such as the RAMDAC video output that presents additional not-major but not-trivial engineering complications for complete 2D or 3D integration.

      --LP

  226. Bluetooth by Anonymous Coward · · Score: 0
    Any news as to what the status of Bluetooth is in Linux ?

    I'm figuring that could become vital in all of this.

  227. Re:technical GPL violation? immaterial maybe, but. by MattMann · · Score: 2

    I don't want to make a big deal out of it (I said a "technical" violation) but it is not an open source question. It is a question of putting no restrictions on redistibution.

  228. Neat idea... by Graymalkin · · Score: 2

    even if it has been done already. That sounds like something a troll would say but it is true, several companies tried this before. I think Crusoe will become popular because it has Linus working on it (free publicity) and it is compatible with current software. Of course mobile users will be all over this but it also has some applications in thin clients and terminals and probably POS terminals. The power consumption and low cost would work to save companies oodle of dollars, not to mention the chips are software upgradable so their useful lifetime is probably 50-75% longer than a regular processor. Does anyone think they might release a code translator for IA-64?

    --
    I'm a loner Dottie, a Rebel.
  229. The biggest power sucker is the screen. by Colin+Smith · · Score: 2

    The LCD screens are one of the largest power suckers on palmtops (no moving parts, so no HD). That's why I get 30hours battery life from my greyscale Psion series 5 and those palmtops with backlit colour screens (HP) get 2 hours.

    --
    Deleted
  230. Code Morphing is FX!32 by General+Breetai · · Score: 1

    It seems to me that DEC (now Compaq) already has the claim on "code morphing". FX!32 Did the exact same thing for the Alpha line of processors. FX!32 would translate on the fly and save the resulting translation. The result was that the more you used the program, the more code got translated and native code is used. The only thing Transmeta did is to integrate it into the processor.

    Based on this, it seems that Transmeta's only real advance is on the power consumption front. This is probably the x86 compatible processor with the lowest power consumption.

    I wonder how long it will take DEC (Comapq) to sue for patent infringement.

    It should be interesting...

    1. Re:Code Morphing is FX!32 by Guy+Harris · · Score: 2
      It seems to me that DEC (now Compaq) already has the claim on "code morphing".

      Digital have done a binary-to-binary translator that (as I remember) can instrument the translated code (or otherwise observe its behavior) and retranslate based on that, but:

      1. they weren't the first people to do binary-to-binary translation;
      2. they don't necessarily have a patent on the instrumenting or other observation, or retranslation based on feedback, ideas.
  231. All of the hype and this is what we got! by tjstork · · Score: 1

    All of this transmeta hype and the end result is yet another intel clone, and a slow one and that. Sure it uses low power... but screens and drives us a lot of power on a notebook, so, no matter how good the guts of it are, it's just another Intel clone. Not impressed.

    --
    This is my sig.
  232. What a disappointment... by fuerstma · · Score: 1

    C'mon.. two thousand plus NDA's and this is all we get. Talk about the hype and rumor not even coming close to living up to the truth. I can't imagine a more disappointing announcement. Low power? I'll take PowerPC.. C'mon C'mon I know that the Crusoe will blow it away in terms of wattages, but isn't the PowerPC cool enough? Ok, given the fact that it's so cool, it could easily scale into faster speed ranges, but every chip on the market is claiming that they can scale into the multi-gigahertz range. What's new? What's exciting? This sounds like another StrongARM! Everyone that is excited is just so since they don't want to appear to detract from their lord 'n savior, Linus... fair enough... but call it what it is..

    --
    www.jackasscritics.com
  233. Links for performance evaluations by Ryan+Taylor · · Score: 3

    The first describes the benchmarking technique, the second describes the results! =) I would have had it up sooner, but either /. is flakey today, or my connection is. Share and enjoy.

    Mobile Platform Benchmarks (http://www.transmeta.com/crusoe/download/pdf/Benc hmarkWhitePaper_1-18-00.pdf) (.pdf, 93 KB)

    Transmet a Mobile Benchmark Report (http://www.transmeta.com/crusoe/download/pdf/Crus oeBenchmarkReport_1-18-00.pdf) (.pdf, 116 KB)

    Sincerely,

    Ryan Taylor

    --

  234. How ironic ... by |DaBuzz| · · Score: 5

    - Company XYZ patents software, they are denounced by the "community" as greedy profiteers
    - Transmeta patents software-morphing and they are revolutionary geniuses

    - Company XYZ cracks down on trademark use in domains and is demonized by the "community"
    - Linus cracks down on the use of "Linux" in domains he doesn't "approve" and he's a God

    - Company XYZ produces a closed source OS that gains 95% market share and they are considered the devil by the "community"
    - Transmeta produces a closed source emulation software and they are the holy grail

    How ironic indeed.

    1. Re:How ironic ... by Anonymous Coward · · Score: 0

      Technically, that's just hypocritical, not ironic... :P

    2. Re:How ironic ... by Evil+Spammer · · Score: 1

      The difference is simple: trust. The community trusts Transmeta to sell the technology for a reasonable price, always play "fair" in its business tactics, and withhold source code only because to do so will ensure future compatibility.

      The trust is there because the community is generally optimistic, especially because of Linus' involvement. But that trust can be broken easily--look at the way people talk about Microsoft, Sun, and Intel after each company did things that were looked down upon by slashdot.

      Transmeta will be able to retain their reputation the best if they stick with the technology they're best at: Crusoe. To expand into web pads, mobile computers, and other applications of the chip would increase complexity and decisions would have to be made between pleasing either the geeks or the shareholders. Soon the trust would be broken and the community would start looking for a new holy grail.

  235. Transmeta can design a whole new instruction set by ebcdic · · Score: 2
    Several people have commented on the possibility of writing native Crusoe programs, and how this is just what Transmeta don't want - because they want to be able to change the hardware incompatibly. But unlike almost any other company now, they can do something else: design a new instruction set.

    This would be an instruction set chosen to be particularly efficiently translatable into the Crusoe native instructions, but would remain constant as the processor changed.

    Why can they do this when others can't? Because they can have x86 compatibility at the same time, and cheaply. The system could simultaneously run both formats, just by having multiple translators.

    And they could do other things, in particular they could join with AMD and implement its as yet unrevealed 64-bit instruction set.

  236. DeCSS implied in products by MadAhab · · Score: 1

    Let's see:

    1. Demos of systems playing DVDs
    2. Demos of systems running Linux
    Methinks DeCSS fits very nicely into the product lines we'll be seeing... Has Linus offered any observations on the knowledge-is-illegal DVD suits anywhwere?
    --
    Expanding a vast wasteland since 1996.
  237. anti-dogma rant != "+1 Interesting" by Frac · · Score: 3
    Being one of a few (if not the only) negative poster, I'm likely to get branded (and moderated) as a troll/flamebaiter, but please hear me out...

    Oh please. You're not one of the few negative posters around slashdot, but you're certainly among the many negative posters that got moderated up - not because your comment was interesting, but you included the "I'm might get moderated down for this..." disclaimer.

    I'm wondering if I watched the same presentation as the rest of the posters here... Deitzel and co. effectively skirted the performance/Mhz question, which says to me that they don't have much to brag about in the performance area, otherwise- They would've bragged about performance/Mhz.

    I'm sorry to be rude, but duh. That's how marketing works. You emphasize on the fine points of your product, and you skim over the weak points. Even a moron would not try to shift that focus in their own product release.

    I could've sworn I was watching an Microsoft/Apple/Intel love-in/press-conference at times. A quote of note: "Crusoe will be a low power internet platform for the future". What the fsck does that mean? There was lots of 'marchitecure', but little in the way of hard performance numbers.

    I sure as hell don't konw what's a "Microsoft/Apple/Intel love-in/press-conference", but if you're referring to the catchy marketing phrases, you might want to remind yourself that you ARE listening to a product release.

    Transmeta is a privately-funded company. It's not a university research laboratory. They aren't presenting a research paper. They are a company, and they are trying to make money to make a return on their investment.

    Also, their product is targetted at OEMs and computer manufacterers, which is why the technical details are in the press pack, and not the webcast.

    Looks like Transmeta's smartest move was to hire Linus, 'cause the whole of Slashdot is believing the (and feeding) the hype without knowing all the facts.

    I agree. Those moderators that feed the heretic-wannabes really should stop. Sorry if I sound a bit harsh, but I really can't stand these "I have some pissed-off opinion to be moderated up" posts escaping my score 2 threshold anymore.

    1. Re:anti-dogma rant != "+1 Interesting" by Anonymous Coward · · Score: 0

      >>You're not one of the few negative posters around slashdot, but you're certainly among the many negative posters that got moderated up - not because your comment was interesting, but you included the "I'm might get moderated down for this..." disclaimer.


      Amen. You can write "I'm going to be moderated as a troll for this, but hear me out: Linus has sex with goats. K3Wl Do0dZ!" And then you'd get +3, and Interesting.

    2. Re:anti-dogma rant != "+1 Interesting" by Anonymous Coward · · Score: 0

      ohmygod ... snot flew out of my nose when I read that last sentence.

  238. Wraping up by gmarceau · · Score: 1
    Intel and AMD can make fast bloated chips.
    powerARM makes a fast low power consumption chip
    Transmeta makes a fast low power consumption chip + they have enough good compiler people to write a state-of-the-art fast JITs for it.

    A figure the ARM people should get into compilers. In fact, all those hardware people should get into compilers. It is a wonderfull way to get over the backward compatibility problem

    -

    --
    This post was compiled with `% gec -O`. email me if you need the sources
  239. Congrats...welcome back to nineties. by knife · · Score: 1

    Wonderful. Transmeta has created the alpha and FX!32. Well, maybe they can market it better than digital could...

  240. how fast did quake run? - sm4115515402 by goon · · Score: 2
    how fast did quake run (fps)?
    specs of the machines that ran them

    • i didn't get to see the webcast so I would like to ask the above questions to anyone who was watching it The reason I ask is it's a (possible)example of how *fast* (quake framerate) the chip can run, in the absence of hard specs.

    --
    peterrenshaw ~ Another Scrappy Startup
    1. Re:how fast did quake run? - sm4115515402 by Anonymous Coward · · Score: 0

      Quake is the best running demo transmeta has. Gets the closest to 1:1 performance with P2 class machines. Demo should have been good - just don't try Excel etc and expect 1:1 Xtransmeta

  241. Is it SMPable? by Muggins+the+Mad · · Score: 2


    I've had a look through the site, but I can't see any mention of how well the Crusoe chips would work in parallel.

    Given the low cost and power consumption, they'd seem to be a logical choice for making cheap parallel machines.

    Does anyone know if they're able to run (closely) in parallel? (ie. sharing system memory and buses) ?

    1. Re:Is it SMPable? by Anonymous Coward · · Score: 0

      No it is not SMPable... onlu UP Xtransmeta...

  242. Re:Transmeta can design a whole new instruction se by gmarceau · · Score: 1
    yep, totaly. And the fact that they are not talking about that upsets me quite a bunch.

    Either :
    - they are actualy dumb and totaly blinded by their wish of being x86 compatible
    or : (maybe more likely)
    - their are not saying anything to not divert the attension from the x86-compatible hype.
    In which case their marketers are just as evil/more as any other.

    -

    --
    This post was compiled with `% gec -O`. email me if you need the sources
  243. Overclocking? by Darkforge · · Score: 1

    So, with all this heat/power consumption reduction in the Crusoe chip, does this mean we can overclock it more easily? What kind of speeds can we get if we're willing to get it as hot as, say, a P!!!?

    --

    When I moderate, I only use "-1, Overrated". That way, I never get meta-moderated!

  244. Linus could do it by Anonymous Coward · · Score: 0

    People would be up in arms, but he could do it. He has the publicity as the creator of Linux. In many people's eyes that makes anything Linux "his".

    1. Re:Linus could do it by Robert+S+Gormley · · Score: 2

      And then me and doubtless many others would haul him quite promptly to the nearest Supreme Court and batter him to death with a wet newspaper...

      --

      Open Source. Closed Minds. We are Slashdot.

  245. Yawn... I think so. by Anonymous Coward · · Score: 0

    I fail to see how this shakes things up much. The top 2 guys Intel and Motorola are very strong in these markets. This seems like a dud.

    1. Re:Yawn... I think so. by chris13 · · Score: 1
      Obviously, you're not looking very hard.

      Think about it, this chip, although not quite as fast as say, a P3, is capable of that nifty code-morphing thing. That means, as someone said earlier, I can theoretically run x86, PowerPC, arm, alpha, etc. software on a single machine.

      So what, you ask? Let's say they make the Crusoe emulate the PowerPC. Bam. Apple no longer has a monopoly on the mac arena. Ever wanted to play with an amiga but couldn't find one? Here's your chance. You could load these things into those webpads with Gameboy code-morphing. Your browser becomes a gameboy. Or even better, load it with SNES code-morphing. You're got a portable Super Nintendo. :)

      And those are just trivial things. The whole Crusoe thing brings up a lot of possibilities. People don't have to alienate an entire market by coding for a single platform. The consumer can just grab an update to the instruction set, and a copy of their favorite OS. This would be great for schools, especially. Windows *and* OS 9 all on the same machine. Double the learning, half the cost. :)

      --
      "How do I know it's really you?" "Uhh... because." "Yeah, okay."
  246. MMX by ecloud · · Score: 1

    It has a multimedia processor unit alongside the ALU and FP units. Presumably MMX instructions do retain some speed at least.

  247. The Motorola 88000 all over again? by Anonymous Coward · · Score: 0

    1> This sounds much like the 88000 that was supposed to replace the 68000 in the Mac, Amiga, Atari, & SUN stuff before Motorola went RISC. What happened to that project? 2> We need to start working on an OPEN SOURCE instruction set. Like all closed source projects, there have to be some suboptimal aspects to the x86 instruction set. Within the constraints of the Crusoe hardware, we may be able to target GCC towards a GPL instruction set. Of course we'll need an ESR instrution. Plus a GNU instruction that recurses forever. You get the idea. 3> Is there any support for SMP in crusoe? If they run on small amounts of power, and can have their virtual instruction sets optimized for MP, then they'd make nice very small footprint cluster machines. 4> With low power needs, and a small physical size, these would seem to be good chips for building scalable display systems. In other words, think of a foot square display with a crusoe processor, and a really thin edged case. Slide a few edge to edge, and you have a smart large display. (The bigger the display, the larger the CPU horsepower.) 5> Solar powered palmtops. Solar cells on one side, display on the other. When your capacitors get discharged, you turn it over and leave it in the direct sunsight? Kind of a etch-a-sketch for the future. 6> Forget the wind up radio, use those spring loaded generators to power your palmtop. Oktober

    1. Re:The Motorola 88000 all over again? by Guy+Harris · · Score: 2
      This sounds much like the 88000 that was supposed to replace the 68000 in the Mac, Amiga, Atari, & SUN stuff before Motorola went RISC.

      Motorola first "went RISC" with the 88000. I forget whether they pitched the 88000 to Sun, but Sun didn't bite - they rolled their own, SPARC.

      What happened to that project?

      It went into some machines - I think Omron had an 88K workstation, and Encore had 88K-based superminis - but, for better or worse, didn't catch on big enough to survive; Motorola eventually dropped it after joining up with IBM and Apple to do PowerPC.

      However, I don't see how this was like the 88000, which was a RISC chip, and which didn't use the RISC instruction set solely as

      1. a target for a binary-to-binary translator;
      2. the instruction set in which said translator ran;

      but provided it as the instruction set that assembler-language programmers and compilers generated.

    2. Re:The Motorola 88000 all over again? by stripes · · Score: 1

      You said:

      Motorola first "went RISC" with the 88000. I forget whether they pitched the 88000 to Sun, but Sun didn't bite - they rolled their own, SPARC.

      What actually happend was Andy Bechlstien (and others) at Sun read a lot of the academic papers about RISC (possabbly before it got that name), specifically the SAIL and a few other CPUs. They told Motorola that there was definitate comercial intrest from Sun.

      Motorola in effect blew them off (Sun was a largeish buyer of 680x0s, but not the same kind of volume as many of mot's other customers).

      Sun partnered with (I think) Fujtsu to build the first SPARC on a 20K gate-array (the original 68000 was a 68K gate full custom job, so the SPARC was really quite impressave). That ran around 15Mhz to 20Mhz (the 68020 ran 20Mhz to 33Mhz at the time), but it got far more work done per cycle. The result was a $10,000 "12MIPS" machine (Sun4/110) as opposed to the $10,000 "4MIPS" machine the 68020 powered (3/150, or the slightly slower 3/60, or lots of 3/numbers-here machines).

      Mot started a crash program to devlop the "88k", which DG eventually used for some ill-fated Unix workstations (The AvIlon i think). They may have started the program before Sun brought the SPARC out, because IBM managed to market a RISC baised machine before Sun's SPARC got out the door (About "6 MIPS", the IBM RT). That CPU had actually been around internal to IBM for a number of years, but never turned into a computer product (I think it came out of hte typewriter labs).

      The 88k never really got any signifigant market share. Even upstart MIPSco got more market share (in part because DIGITAL was cought by Sun -- Sun's boxes were TEN times faster for the same money until DIGITAL droped their internal titan project, and brought out some nice MIPS R2000 and R3000 baised machines, total design to market time of about 8 months, incredable work, and the R3000s were slightly faster then the SPARC, at least for a little bit).

      Then a whole bunch more stuff hapened:

      • eventually some big complicated Apple-IBM-Mot deal created the PowerPC which failed to crush the world, but has managed to capture decent market share.
      • DIGITAL droped the MIPSs in favor of their own quite nice Alpha, which despite being far ahead of every other CPU for many years (or at least bing allways in the top 3, and normally in the top one for FP perfomance, and in the top 5 for integer performance)...um...where was I, oh yes, the Alpha also failed to crush the world. Then again the world has repeatdly failed to crush the Alpha, so i guess they are even.
      • HP's PA showed up and battled the Alpha for a long time, frequently having the number one spot for weeks at a time. Even months sometimes. It seems to have withered a bit in the past year or two.
      • The death of RISC has benn predected several times. As has the death of CISC.
      • Intel brought out a pretty cool RISc CPU, the i860, which went in a OkiData Unix workstation, and then faded pretty much from existance
      • Lot of other stuff that I forgot, didn't know, or don't think is important
    3. Re:The Motorola 88000 all over again? by Guy+Harris · · Score: 2
      Sun partnered with (I think) Fujtsu to build the first SPARC on a 20K gate-array

      Yes, the first SPARCs were fabbed by Fujitsu; the first full-custom SPARC (7C601) was done by Cypress.

      because IBM managed to market a RISC baised machine before Sun's SPARC got out the door

      As I remember, the RT PC came out after SPARC - I remember people waiting in fear at Sun because they didn't know whether the RT was going to crush them like a bug.

      That CPU had actually been around internal to IBM for a number of years, but never turned into a computer product (I think it came out of hte typewriter labs)

      It was called ROMP, for Research and Office Microprocessor; I guess the "Office" part indicates that it was intended for office equipment such as word processors, etc..

      Mot started a crash program to devlop the "88k", which DG eventually used for some ill-fated Unix workstations (The AvIlon i think).

      Right - it was used in the DG Aviion machines, as well as the nes I mentioned.

      I seem to remember hearing (from a comp.arch posting by John Mashey, I think) that Motorola pitched the 88K to MIPS as well as to Sun.

      eventually some big complicated Apple-IBM-Mot deal created the PowerPC

      ...based on IBM's POWER (Performance Optimized With Enhanced RISC - yes, really; some marketoon must've gotten a bonus for that one) architecture, which was their attempt, after the ROMP and a second generation thereof; the first POWER chip (RIOS) did a fair bit better than the ROMP, performance-wise.

      HP's PA showed up and battled the Alpha for a long time

      I think the Precision Architecture antedated Alpha.

  248. New possibilities by tilly · · Score: 2

    Yes, there will be faster chips, but not with the battery. There is also room in rackmountable stuff.

    Plus I am looking forward to high-end SMP laptops. Here the power savings is not as important as speed. (Think demos.)

    No, Transmeta does have some good opportunities if they can execute well.

    Cheers,
    Ben

    --
    My usual seat in the cluetrain is at A HREF="http://pub4.ezboard.com/biwethey.ht
  249. This has serious possibilities ... by ikaros · · Score: 1

    I don't want this to sound like flamebait, but if you don't see immediately the possibilities beyond just having a waycool Linux-based PDA, then I hope you're not working in the technology field.

    Remember, if the Crusoe can emulate the x86 family, then emulating the PPC, the Alpha and the Motorolas -- or anything else, for that matter -- is just a matter of software.

    This has the potential to be the first true multiplatform system. This is what gives it the potential to be both an Intel-killer and a Microsoft killer. The user can run whatever they feel like -- most importantly, x86 compatible apps without having to have an x86.

    It's this lack of total portability that's prevented me from completely escaping the evil clutches of Redmond -- I still can't run Poser 4 and Jack Nicklaus 6 on Linux. Wine is a valiant effort, but it's not stable enough for that yet (unless some major breakthrough has been made since the last time I checked in). The Crusoe looks like it's offering me a possible way out.

    I'm also stunned by the prices -- a Crusoe-based box is actually within my reach. These guys are serious ... I have to wonder if Paul Allen has any idea the danger he's put his MS holdings into by backing this project.



    ikaros, who will almost definitely get one of these systems

    --
    You're only as young as the last time you changed your mind -- Timothy Leary
  250. TM is doomed by Anonymous Coward · · Score: 0

    Who does Transmeta think they are going to get to buy these processors? Intel and AMD own the market, so TM is awfully foolish to release something like this and dare to compete with the big boys. TM and Crusoe are doomed.

  251. Multiple ROMs, Multiple CPUs, OpenGL ... by Anonymous Coward · · Score: 0

    That's an interesting point. They Could also have Open GL, AltiVec, x86, and MMX all in the Morphing ROM code at the same time couldn't they? If the ROM isn't big enough Couldn't you just add another one? For example could you have a Morph Engine for your sound card instructions, video instructions and x86 instructions communicating with an arbitrary number of TM CPUs? That way adding a chip increases video and sound ability too... Then again I don't know anything about hardware ;-)

    1. Re:Multiple ROMs, Multiple CPUs, OpenGL ... by Esperandi · · Score: 1

      Good question... you don't really need this though. The way the processor works, the code morphing layer has a cache. Once it translates a certain OpenGL instructions fallout (a sinlge OpenGL instruction boils down to hundreds of little math instructions through the API that deal with matrices), it caches that. The next time that comes around, those instructions aren't re-translated, they're simply re-translated. If this happens a few times (and in real world applications it does constantly... unfortunately benchmark programs specifically do the opposite of this, trying hundreds of different things in a few seconds) the code is also optimized.

      So you don't need to add these things into the code morphing layer, they will automatically be folded into the cache which is sort of like a "superfast bag o tricks" the Crusoe uses. Its sorta like how they emulated the N64 on a PC. They simply do things differently, which results in much more speed out of much less hardware.

      Esperandi

  252. Yeah but... by chris.bitmead · · Score: 1

    it sounds like the chip does the translation one-off into native machine code. So after the initial translation it should presumably run at native speeds.

    1. Re:Yeah but... by pnagel · · Score: 1
      The point is to do optimization at compile time and at run time, since at run time a lot of other information becomes available.

      For the compiler to be able to prove that one branch is taken more than another is often impossible, but at run-time you can just statistically observe that and adjust accordingly.

  253. Yeah but... by chris.bitmead · · Score: 1

    if the chip's internal logic for optimisation was embedded in a compiler, the optimisation layer would be redundant and you could generate native code that could run un-translated.

  254. That doesn't make sense... by chris.bitmead · · Score: 1

    Since they've got the technology for instruction morphing, COMPATIBILITY DOESN'T MATTER ANY MORE. They could just write an emulator for the old VLIW instruction set and then go on to change the instruction set any way they please.

    Of course I can see that they wouldn't want this burden on day 1 of their release. On the other hand locking people into their VLIW instruction set (or any of the versions thereof) would probably be good for them.

  255. Fear of a Backwords Compatible World by Spasemunki · · Score: 1

    One reason for Transmeta not to encourage writing native code for the Crusoe instruction set is that it very quickly creates the same trap as the current x86 instruction set: Backwords Compatiability. No one wants to buy a computer that cannot run the software that they have now, and this kind of attitude propigates down the line, until we have an instruction set that is being held back by the fact that we're still carrying the capability to use software that noone has been interested in for 10 years. If Cruose starts having native code programs written for it, than the next time they want to make a "next big thing" jump, they have the ball and chain of carrying all the old instructions with them. No one wants to rewrite their compilers and programs to work on a radically new instruction set, but maybe a radically new instruction set is what will be needed to face whatever computing challanges come up in the next 2/5/10/n years. Someone mentioned earlier the "vulerability" that Intel was facing, because they want to move away from the somewhat bloated by thoroughly entrenched x86 instruction set. x86 has so much momentum that Intel is going to have some very big growing pains once they try and move away from it. Leaving the code morphing in place as an "insulator" between programs and the instruction set permits them to tinker with the chip itself as much as they need, while keeping the support of users and developers, who will be happy to avoid a complete system overhall and rewrite every 2 years. Plus, if the Code Morphing is as fast and quality as the hype, than the ding in performance will be minor, and barely perceptable to most users.

    1. Re:Fear of a Backwords Compatible World by Jason+Earl · · Score: 1

      If Cruose starts having native code programs written for it, than the next time they want to make a "next big thing" jump, they have the ball and chain of carrying all the old instructions with them.

      It seems to me that this unfortunate occurance has made Intel billions of dollars over the life of the x86 instruction set.

      If I were making a fancy dan new processor you could bet your socks that I would be interested in getting developers and users hooked on my instruction set. Otherwise I have to spend all my time worrying about my competitors out-performing my chip. After all, if the instruction set is taken out of the equation then it all comes down to power consumption and raw performance.

      Right now the Crusoe processor has to emulate the x86 instruction set because no one is interested in anything else. But the x86's days are numbered. Pretty soon it will be a whole new ball game. At first most people are going to be interested in running their old 32 bit software on their shiny new Crusoe or IA64 chips, but it won't take long for the battle lines of the new instruction set to be drawn.

      It makes me wonder what Transmeta really hired Linus for. Say what you will Linux will be a big part of the software world for some time to come, and it's low cost, small footprint, and high stability make it an excellent choice for smaller embedded applications (like the ones that the Crusoe will be used in).

      The fact that Linus is working for Transmeta means that they have a substantial amount of control over what happens in the Linux kernel...

      We live in interesting times.

    2. Re:Fear of a Backwords Compatible World by Anonymous Coward · · Score: 0

      The fact that Linus is working for Transmeta means that they have a substantial amount of control over what happens in the Linux kernel...

      Not really,... Linus holds linux totally seperate from his work at transmeta... A new feature might get added because of transmeta.. but it wont be because he works for them, it will be because the feature interests him.

      Transmeta has ZERO control over the kernel except for the fact that they are closer to linus and its easier for them to bug him when he is in the next office rather than somewhere farther away.

    3. Re:Fear of a Backwords Compatible World by Anonymous Coward · · Score: 0

      Has any producer, researcher or computer geek been able to make an new better instruction set than the one existing today? First when a new ISA is created then will their be possiblites to make a new generation of computers.

    4. Re:Fear of a Backwords Compatible World by Anonymous Coward · · Score: 0

      >The fact that Linux is working for Transmeta means that they have a substantial amount of control over what happens in the Linux kernel... Uhm, apparently you don't know much about Linus or Linux. Transmeta has no pull at all over the kernel. If they asked, and Linus refused, the worst thing that could happen is that Linus would lose his job. Obviously he's not too worried about capitalizing on Linux, or he would have done so quite some time ago. Linus took the job at Transmeta because it allowed him to work on Linux. Not because they were willing to give him the most money. Besides, any changes Transmeta wanted Linus to include to the kernel would have to remain open source, unless Linus plans on forking the kernel. Any changes Linus would make that Transmeta doesn't want Open Sourced would have to remain in-house, which means that the open Linux kernel development would remain the same as it has for the last 8 years or so. So, basically, if Transmeta wants Linus to take the kernel in a new direction, so what? Either it'll remain in-house, and OUR beloved kernel stays the same, or Transmeta has the changes open-sourced in the kernel. Either way, there's not really a down side... now, personally, I wouldn't be too surprised to see certain Crusoe-specific code pop up in the kernel, but chances are that would have happened eventually even if Linus had nothing to do with Transmeta. -DarkProphet

    5. Re:Fear of a Backwords Compatible World by Anonymous Coward · · Score: 0

      Why couldn't they have a Crusoe code morhping layer as a bridge between the older and newer architecture?

    6. Re:Fear of a Backwords Compatible World by ChadN · · Score: 1
      The fact that Linus is working for Transmeta means that they have a substantial amount of control over what happens in the Linux kernel...

      On the other hand, Linus can leave and find a great job ANYWHERE he wants (basically). So, it is not like Transmeta can really dictate anything about Linux.

      --
      "It's overkill, of course. But you can never have too much overkill." - Anonymous Slashdot Coward
  256. What's wrong with Transmeta? by Anonymous Coward · · Score: 0

    The biggest problem with the StrongARM, the draggonbalz, or any other crappy portable processor, is that they are *NOT* x86 compatible. Making an x86 compatible processor means that millions of screaming Linux weenies will have something to play with. Oktober

  257. nice dog and pony show by bangman · · Score: 1

    Having Linus playing games. I felt sorry for the dude. Looking like some kind of trained monkey. I can go to chinatown and watch a chicken play a piano too. Next time their demo would be cooler if it had a barnyard animal playing quake!

  258. Other possibilities by rob_from_ca · · Score: 3

    Now it seems to me that the Crusoe processors that are coming first are mobile just because the intial ideas of Transmeta lend themselves well to that application. Has anyone else read the whitepaper and seeing what I think I'm seeing? It looks like they've just taken the back end instruction unit of a more traditional CPU to start. But rather then add microcode and transistor based software to control getting software instructions to it, they've moved those functions in normal software. This allows the actual CPU's to be cheaper, smaller, and simpler (hence the lower power requirements, in addition to their clever management scheme). But is it just me, or does this also provide a fascinating layer of abstraction? Couldn't the code morphing software just as easily be programmed to dispatch instructions to multiple VLIW cores? Not to mention all the advantages of being able to write more complicated software that can make better choices about how to optimize and keep the core processor running. Very interesting idea I think - lots of possiblities. It seems like it's essentially turning the x86 instruction set into an API, so bigger and badder advances on the silicon side of things can be taken advantage of without having to go crazy optimizing existing applications. Abstraction is one of the most powerful tools we have, and this is a fascinating use of it. Of course, they don't mention much about the extra RAM that translation cache and code morphing software is going to require - I wonder what type of cost this abstraction comes at.

  259. "Not really, but yes it could" by Anonymous Coward · · Score: 0

    Ok I read the whitepaper now. The read/write reordering would be as usefull to compiler generated code as emulator generator code.

    Instrumenting the code seems to serve two purposes, to determin wich code to most aggresively optimize and for tracing branches. (to determine wether to do speculative execution of the most common or even both branch targets)

    The first purpose would not be much of an issue with a native compiler, everything would get fully translated. The second would still be an option for compiler generated code.

    Apart from changing the optimisation level of code and changing it depending on branch probabilities they dont seem to mention any other instances where the code would be changed during run time...

  260. Re:technical GPL violation? immaterial maybe, but. by Anonymous Coward · · Score: 0

    Mobile Linux ought to be good for all of us.

    Even if Intel is able to respond to Transmeta by improving StrongARM, Mobile Linux will run on it too.

    Linux has needed a consumer-level platform to grow, to correspond with success in the server and workstation area. Microsoft and Intel have not been able to come out with a good mobile platform (one without a hard disk) since the HP Omnibook. And 3Com needs some competition to its pokey CPUs.

    The mobile, handheld computer applications are likely to grow along with cellphones and increased wireless bandwidth. It will be good to have real web pages instead of just a few lines of text and no graphics. And we will be able to run our favorite Linux programs too, vi or emacs.

  261. The Power of Crusoe by Ranger+Nik · · Score: 2

    as i read through various sources, i could not help thinking "yeah, it is low power and translates code, but P3's have been translating code in hardware which must be faster".
    What IS the Power of Crusoe? Why is it so fast?
    /. made everything clear. here it is:

    1) Crusoe is so fast (almost equal P3) because the code morphing is done in software. software is slower per se, but it is a lot easier to use lots of memory for state caching, can be tuned easier, etc - all the reasons you would not want to have your average application program etched into silicone.

    2) the true power of Crusoe is that actual silicone development is completely independent of previous iterations. if the next-best-thing-after-VLIW gets invented, they can put it into silicone in no time flat, while all other processor developers can only watch from a distance. WOW. WOW. WOW.

    of course, there is also small die size (read: cheap!) and low power comsumption. but these advantages pale in comparison to 2, above.

  262. Code morphing for MAJC by Wesley+Felter · · Score: 1

    I don't see any reason you couldn't write code morphing software for MAJC (or IA-64 or whatever), although it might be tricky because of the exception ordering issues mentioned in Transmeta's white paper.

  263. Yeah, I have a thought on this by LinuxParanoid · · Score: 2


    If Transmeta can keep the competitiveness of the
    chip up from a speed/computer-power standpoint, then vendors of RISC instruction sets will do a cost analysis of whether its cheaper to design their own processors or to design software to program Transmeta's. Licensing terms and Transmeta's ability to drive volumes would also be pivotal. They do give up some control, true, but this may be preferable to an Intel-only migration, since with Transmeta they can retain their ISA lock-in and binary compatibility.

    --LP

  264. Pricing? by CaptainDrewle · · Score: 1

    Since these chips are smaller, etc.. Would that mean that they will be cheaper than Intel processors of similar speeds? It would kinda make sense in the case of the "web pad" where people might not want to spend more than a few hundred dollars.

    Does anyone know the estimated pricing of the processors?

  265. Intel and AMD also emulate the instruction set by JoeBuck · · Score: 2

    Nobody builds processors that directly run the ix86 instruction set any more. Both Intel and AMD put out processors that execute RISC-like instructions at the core, and then surround it with a lot of machinery to translate the user level instructions into these operations on the fly. There's no reason why Transmeta's approach can't be competitive to that, especially given caching of translated instructions.

    What we're seeing is the rebirth of microcode. This approach makes the most sense when executing a tightly coded instruction stream -- CISC instructions, or maybe bytecodes (there might be a performance advantage in doing JIT translation from Java bytecodes to VLIW instructions, but then again the combination of the two JIT translations, from bytecodes to ix86 and from ix86 to VLIW might be nearly as good and cheaper in the long run.

    1. Re:Intel and AMD also emulate the instruction set by Anthony+Kilna · · Score: 1

      I think a good point is being made here: it probably won't matter much that we have two layers of translation (Java bytecode -> x86, x86 -> TM VLIW) because of the built-in self optimizing nature of the TM processor. In fact, I imagine TM will be the best PC platform for running any sort of instruction set translation because of its "smart" self-teching method of translating. I'll bet something like the PlayStation emulator will rock on their hardware.

      --
      s/[BW]ill(y|iam)?( H\.?)?( G(ate|8)(s|z))?(,? ?v?(III|3)(\.\D)?)?/Girly-man/gi
  266. Re:the iBook doesn't have a fan!!! by Anonymous Coward · · Score: 0

    ..and the fan in my PowerBook G3/400 doesn't run constantly.. Chances are the next PowerBook will also be fanless...

  267. Intel/AMD don't execute x86 instructions, either by DragonHawk · · Score: 3

    ... they are still emulating the x86 instruction set. Even with all their fancy technology a 700mhz processor will only perform, at best equivalent, to a 500mhz Pentium.

    It is important to realize that the current Intel and AMD CPUs do not execute x86 instructions, either. They use hybrid RISC cores with front-end instruction decoders to break down the x86 CISC operations into smaller, RISC-style operations which can be more easily optimized.

    So I wouldn't be so quick to dismiss Transmeta's performance claims as impossible simply because they are using emulation. If this was traditional emulation, I would agree with you. But it isn't.

    In traditional processor emulation, you write a special program which sits between the host CPU and the "foreign code". This program reads the foreign code, figures out what it is doing, and does it. Essentially, you write a program to act like the foreign CPU. Not very efficient.

    Transmeta's "code morphing" techniques, on the other hand, do something a little more intelligent. They start by translation of x86 to native instructions. So you are running native code, with an up-front penalty for translation. Then they apply selective optimizations to tune the translated code to the native design as needed.

    Given time, that could result in much higher performance then traditional emulation. You might get very close to "native" performance by optimizing the code that matters in ways that apply to the native CPU.

    Or you might not. There is very little hard data available right now, so all anyone can do is speculate. We can't say "Yes", nor can we say "No".

    --

    dragonhawk@iname.microsoft.com
    I do not like Microsoft. Remove them from my email address.
  268. Taos by Anonymous Coward · · Score: 0
    Does anyone remember Taos? It was (is?) an operating system where binaries are compiled for a "Virtual Processor" (32 bit, 16 address regs, 16 data regs).
    The loader would recompile for the native machine, giving outstanding performance with complete hardware transparency.
    I believe at the time there were Amiga, Atari, Macintosh and 386 ports - all binary compatible.

    -Anonymous coward

    (And How can they call it VLIW when there's no 1-1 mapping between instructions (atoms) and execution units?)

    1. Re:Taos by Trent+Took · · Score: 1

      Where TAOS approached the problem of binary compatibility between platforms with the Virtual Processor concept the Crusoe solution looks like the exact opposite.

      The have accepted that x86 IS the market (at least volume wise) so have focused on that. I'm not supprised that they were cool on the idea of porting the CodeMorphing engine to PowerPC for the Mac. IMB/Motorola have a lot less problems with the power consumption of these chips as they are not tied to Wintel legacies.

      Having said that there is no reason why you can't run native - and everybodies favorite OS seems to be the easiest to port.

      --
      See old Atari's come to life stonx.sourceforge.net
  269. Virtualisability? by shirro · · Score: 1

    Not sure that is a real term but as I understand it x86 processors don't have the stuff to be fully virtualisable so people have to do really clever hacks like vmware so we can run multiple OS.

    Now that the instruction set is in software I wonder if Transmeta can make this a bit simpler to do?

  270. Want mine in candy apple flavor by Anonymous Coward · · Score: 0

    Flexibility Reliablity Funtionality and last Cheap 1 watt power consumption easily means solar power Battery what battery use a flashlight for light. Now that ya got an xtra bay what to do. But other devices would suck the life out of the battery Software for coding, hum that will be tweaked The chips not going for claims to speed but feasible battery life Sounds like this could be the beginning of the practical laptop market. Intel with there newly debut itanium-64 bit and Crusoe 128-bit why pay more for something outdated within months Cant be a hoax with IBM shelling them out before there on sllers shelves. Oh and Paul Allen multi -billionaire dumping money into so fast he forgot he is stilling living with his mom and dad. If Crusoe had not been for laptop Joshua from Via would have had little impact on desktop market but Joshua may gain remarkable share of the desktop market while Crusoe take on laptop. What is intel to do but buy a 51% stake and ride it for all it's worth.

  271. Privacy concerns of the Crusoe Model by Steeg · · Score: 1
    I am concerned with the business model and architecture Transmeta has chosen for the Crusoe Processor family. One reporter asked about viruses and the reply was "we" shut down all the pathways right away so there is no access to the underlying processor or morph code. At the same time Linus joked about what other processor could you get a fix from the web for an instruction gone bad.

    Without open sourcing the morph code Transmeta has created a black box. This processor not only has the potential but was specifically designed to track everything that is going on in the machine and react at a very high level. People flipped their lid when a number was imbedded in the Pentium III. Transmeta has the means to imbed, remove and potentially send anything they wish or are pressured to do.

    It is my belief that a major reason for Linux growth in the world is the ability for nations and individuals to not only take the word of the distributor but have the means to verify it themselves.

    I hope Transmeta can see this very real obstacle for what it is and opens up the morph software as soon as they possibly can.

    I feel Privacy and Freedom are integral parts of the same thing. I look forward to Transmeta changing their cash flow away from needing a proprietary lock on the Morph code. I hope they will see this as code death in the internet space.

    The processor won't be a personal processor for me until I can look at the source and compile it myself. Then I know what I have and it is mine. This is the only processor package I've seen that has the potential. I have in Linux my "own" OS and I always wanted my own processor. I don't think I am alone. :)

  272. Oh no!!! Amiga is going to kill my favourite OS :( by Anonymous Coward · · Score: 0

    Yeah Tao has always seemed to me to be very cool. Their homepage is at www.tao.co.uk your post made me check them out again... but horrors of horrors, there is a very recent news item saying they are going to cooperate with Amiga. Dont they realise its a kiss of death?

    Of course Taos has been kept back by their expensive restrictive licensing/development model anyway... so it wouldnt matter that much if they went under.

    Just goes to show how revolutionary Transmeta really is though.

  273. hey! by fleckster · · Score: 1

    I'd REALLY like to see THIS on a Beowulf cluster!!! ;-)

    Bring the karma-toilet on if you don't think this is funny! Heheheh!

    --
    ............ no.
    1. Re:hey! by TeddyR · · Score: 1

      hmm... one of the RARE beowulf comments that is actually on-topic...



      --

      --
      Time is on my side
  274. Re:The Athlon!!! - HUH? by Anonymous Coward · · Score: 0
    Really no moddern app uses it (x86 instruction set) anymore...


    every look at some assembly listings?
    stuff like:

    mov eax, 5
    mov ebx, [ecx+0005]
    cmp eax, ebx
    jz 79835024

    All that sorta stuff is just your basic i386 (286?) instructions, 95% of code (at least) is just your basic instructions (no mmx 3dnow or whatever)
  275. Technically Incorrect, I believe by Anonymous Coward · · Score: 0

    Technically, if an AS/400 "CISC heritage" compiler were run on AS/400 RISC hardware, the resulting program could in fact run on a CISC box. You are partially correct in that the MI is continuously enhanced with new OS/400 releases and the changes are only forward compatible. At this time, I think IBM still supports V3R2 compatability for the older CISC boxes on the current RISC hardware...

  276. Scalability of Crusoe by varjack · · Score: 1

    Have you read the Whitepaper on Crusoe? Nobody seemed to realise the real potential beyond the little guy. 1.It says (well, almost) that if you want to get a 4way parallel processor you only have to do minor adjustments in the translation software. And it still won't overheat(I hope). 2.The scalability is GREAT. You could stack 4FPU units, 2 integer units...etc(8 units per total) and use 256bit VLIWs to send data to it. Minor adjustments in the translation software are needed. 3.You can upgrade your hardware without any kind of change to the apps. Merced watch out:)). Think of an 8 way Crusoe being fed reordered translated instructions. By the software god. 4.This is why we should await a little more from Transmeta in the next year. just a little more...

    --
    It's easier shooting someone than beating him to death? Hell, it's easier shooting someone than buying the Times. Su
  277. The virtual virtual machine, alas TM only by Anonymous Coward · · Score: 0

    Yep, the really exciting part of Cruso is that enables a basically virtual CPU upon which you could experiment with better support for High Level Languages concepts, like Logic Programming, Lazy evaluating functional programming, Lisp, etc. Alas, the TM revealed much a dissapointment as this will all be available only to TM. From the user and developer perspective all this is is Yet Another x86 with half the power consumption, which doesn't amount to all that revolutionary once you realise that most of the other factors that consume power are basically unchanged.

  278. want EASYCPU by Anonymous Coward · · Score: 0

    always i don't have time read article. IMHO,huge streams are human oriented. so wish the genius considerate "easy to use/programming hardware/CPU". Personally, i think future cpu will be uncomplicated instruction structure. anyways great thanks for "new cpu" to transmeta with their brightness. TOT

  279. The Revolution Will Not be Televised by Ravagin · · Score: 1

    Hey, they never said The Revolution Will Not Be Webcast!
    ===
    -Ravagin

    --

    Karma: T-rexcellent.

  280. David Cassel's Update by cave76 · · Score: 1

    Later Linus added that "It looks alot better this week than it did last week," and that it "needs some work..."

    I may be mistaken, but wasn't Linus talking about the virtual keyboard and not Mobile Linux when he made this statment?
    Mike

    1. Re:David Cassel's Update by NovaX · · Score: 1

      It was. I noticed that mistake too. Considering this is slashdot, where they can't even be grammically correct, let alone be factual, what do you expect?

      I believe they also said the 500mhz P3 was equivelent to their 700mhz processor, not 667mhz. Of course the P3 mobiles are considerably slower than desktop versions, but still fast. I hope that now since they have the hard part dealt with, they can work on desktop/server chips, as they can move from a simplistic internal to a complex one, though without all the problems everyone else has. Considering they have experienced Sun and Motorola engineers, it should be easy enough...

      My only wonder is why people were so ignorant to why they won't publish the internal instructions. Heck, when writing object classes, the point is that implementation can be changed whenever, but the user can still make the same calls, and recieve the same result. Performance can be improved, features added, and bug squished. Just the programmer writing with that object doesn't need to know. If they tied themselves down, they'd be in the trap their breaking the microproccessor out of...

      --

      "Open Source?" - Press any key to continue
  281. how to pronouce "meta" by crayz · · Score: 2

    See, I've always said it like "may-ta", but in the conference they were clearly saying trans-met-a

    how do you guys pronounce it, and are both ways correct? and not just for transmeta, but for meta-anything.

    may-ta just sounds so much better to me.

    1. Re:how to pronouce "meta" by costas · · Score: 2

      Since it's a greek word --actually a word-stem, I can guarantee you is prounounced met-a (and in Greek, actually met-A).



      engineers never lie; we just approximate the truth.

    2. Re:how to pronouce "meta" by Anonymous Coward · · Score: 0

      Its definately "met-uh", like "Metamorph".

  282. Key Q: firmware updates available for end-users by Anonymous Coward · · Score: 0
    If yes, then you have a somewhat future protected investement. AND you can be _sure_ that many people will have fun reverse engineering it and providing access to the native VLIW cpu.

    If not, well... it's just another boring x86.

  283. Emulating x86 is a tour de force. But... by Brett+Glass · · Score: 1
    ...imagine how much better the CPU could run if the translation overhead were minimal, or if all of the translation were done in advance.

    Wouldn't it be funto develop a virtual processor -- one which never existed in hardware -- that was designed to map particularly well onto the current Crusoe CPUs? (It probably would not map badly to future ones.) It'd also be fun to precompile critical sections so that dynamic translation would not be necessary for those portions of the code -- the equivalent of custom microcode.

    How about it, Transmeta?

    --Brett Glass

  284. Slashdot Interview! by Uart · · Score: 1

    How about a slashdot interview with a)Linus, or b) one of the other transmeta guys about crusoe?

    --

    Opinionated Law Student Strikes Again!
  285. SIMD? by TheBashar · · Score: 2

    Can some more intelligent people than myself comment on how these Crusoe processors might or might not be able to handle the Intel and AMD SIMD instructions (ie MMX and 3DNow!). Could this morphing use those, or are they proprietary and need to be licenced? If they're not I guess most apps would take a performance hit from not being able to use them?

    1. Re:SIMD? by aok · · Score: 1

      Someone asked this question during the Q&A period. The code-morphing includes the MMX instructions. But not 3DNow! I forgot if he said they were working on it or didn't think it was worth it though.

  286. Two Questions by Uart · · Score: 1

    1) How much do they cost?
    2) When/where can i get one? or possibly two? (mmm... SMP....)

    So, wait, i thought Linus didn't want to back any particular distro, but it seems he just wrote one for them? Also, why does a processor company need to have a Linux distribution?

    Anyway, if the power consumption is anything like they said, then I can't wait to get one in laptop form. Do they have any OEMs' backing them yet?

    --

    Opinionated Law Student Strikes Again!
  287. Transmeta might be pulling our leg here by Squeezer · · Score: 1

    Transmeta says it uses one watt, but that doesn't mean much for low power consumption. Its an indicator, but its like Microsoft and Mindcraft, its more of a half truth...so...What happens to be the milliamp hour rating on the 400 and 700 mhz processors?

    --
    Does the name Pavlov ring a bell?
  288. YAY! the price!! by Anonymous Coward · · Score: 0
    I gota say, the Slashdot crowd is now much more well to do than before, no one talks about the price, which concerns me most.

    but check this out: it is less than 400 dollars!! now if it runs linux, if i can install the compilers on it, than it's fabulous! my money is right here, come take it.

  289. Morphing on Java bytecodes? by Anonymous Coward · · Score: 0

    Does anyone know if this thing can Morph java bytecodes??

    --

    1. Re:Morphing on Java bytecodes? by Anonymous Coward · · Score: 0

      Yes that was one of the demos. They according to the morning session had a demo where they are running java bytecode at the same time as x86 instructions

  290. I think it's a potential major revolution by RayChuang · · Score: 2

    Personally, I think the Transmeta CPU's are potentially the "tip of the iceberg" when it comes to CPU designs.

    Remember, Transmeta has only shown x86 instruction set emulation; it could conceivably run another CPU instruction set (e.g., PowerPC G3 or G4). After all, if it takes only a little programming for the TM3210 or TM5400 to run Java almost like "native code"....

    The very low power consumption of the Transmeta CPU means the elaborate and expensive CPU cooling needed for the mobile Pentium II/III CPU's are almost eliminated; this also means that battery life can be dramatically increased, which means that the "Internet appliance" shown at the demo today will become reality at a reasonable price.

    By the way, the Transmeta TM3210 could form the basis of the "Internet terminal" most everyone forecasts to be popular in the next few years. Because of its lower power consumption, such terminals could be quite small, and could be fitted with USB, IEEE-1394, RJ-11 analog telephone and RJ-45 10/100Base-T Ethernet connectors. In short, it'll be just one flat panel box, with motherboard and a small HD in the box, plus external keyboard and mouse pointer.

    It'll be interest to see if TM3210-based machines can compete against the Sony PlayStation2....

    --
    Raymond in Mountain View, CA
    1. Re:I think it's a potential major revolution by Quantum+Phase · · Score: 1

      I am not sure anything can compete with Playstaion II but I agree that this is the tip of the iceberg. It sets the hardware guys free to create insane designs that they couldn't consider before because the software layer insulates apps from the changes. Intel would be wise to adopt, clone, or license this tech.

  291. PC add-on card by BorgDrone · · Score: 1

    what about sticking one of these chips on a PCI add-on card for a desktop PC ?
    since these CPU's can be reprogrammed to execute virtually any instruction set they can be used as great hardware accelerators.
    e.g. start a java application, the computer reprograms the crusoe as a native java processor and you have a java accelerator card!
    or when you start a game , you reprogram the chip to do 3d instructions (@700MHz!!!) and there you have a 3d accelerator! .
    or make it do encryption!
    the possibilties are unlimited!
    and when no application is using the card, make it do work for distributed.net

    ---

  292. buzz vs hype by MattMann · · Score: 2
    the difference between hype and buzz. ("Buzz is when you're quiet and someone else talks about you.")

    he was quoting a bowdlerized version of the original:

    the difference between advertizing and PR: advertising is when you say how good you are in bed. PR is when your ex-es say it. PR works better.
  293. Liar! by Anonymous Coward · · Score: 0

    In fact, Ditzel was really building up momentum. I asked him what had happened in early 1998 -- when he was quoted as saying "We had a major change in direction a few months ago, and that has slowed us down a bit." His immediate answer drew applause, and probably the biggest laugh of the morning. "That was just something to throw off reporters.

    That's a lie. They found out the original chip couldn't compete with Intel's performance. This is when they decided to go after the mobile market. They're making it sound like they were aiming for the mobile market all along. That's just not true - they changed direction when they couldn't compete.

  294. This is all great, but... by Oz+Factor · · Score: 1

    I really love these chips, but it's much more x86-centric than I expected. Low power is all well and good, but that's not really what I see as the great strength of Crusoe...what about other Instruction Sets?

    PPC? Sparc? Alpha? Even more intriguing, what about IA-64?

    I know that they repeatedly said they could do this sort of thing, and during the Q&A said that Mac compatibility is theoretically possible, but will any of it happen? TM will have to write a whole set of code-morphing software for each ISA. Have they even done any work on these fronts? To me, THIS is the real potential of the Crusoe family.

    That said, this type of thing may require more work than just the code-morphing software. For instance, to achieve Mac compatibility, you would need to do a whole lot more than get PPC compatibility. They would have to comnvince Apple to design a whole motherboard/chipset to work with this processor, which means any possible Mac compatibility would be a LONG way off, even if they did manage to strike a deal with Apple. Is this in Apple's best interests? Certainly. A PowerBook with even more battery life would be wonderful. But will it happen? Somehow I doubt it. I assume the same probably holds true for other platforms as well.

    Any thoughts?

    1. Re:This is all great, but... by fReNeTiK · · Score: 1

      what about other Instruction Sets?

      PPC? Sparc? Alpha? Even more intriguing, what about IA-64?

      From what I saw on the webcast and the first reports on the newssites, I just can't imagine this feature not being planned. Altough they said the current chips were "optimized" for X86, I was unable to figure out if they meant the actual VLIW hardware or only the code morphing software. I think it has to be the latter, and if it is, the only thing they'd have to do is create a specific code morpher for any instruction set to be supported. I really like this idea, let's hope it is the case...

      Too bad they won't open up the translation stuff. That would enable third party developers to create translators for other chips. Oh well, it's their IP isn't it.

      One thing to consider is Transmetas strong alliance with IBM (i suspect there's more between them than just IBM building the chips, Transmeta said something about having access to all of IBMs relevant know-how). IBM has always supported the PowerPC chip... Makes me wonder ;)

      --
      I strongly believe that trying to be clever is detrimental to your health. -- Linus Torvalds
  295. Borrowing the Crusoe 'Dynamic recode' by technos · · Score: 2

    Obviously, from the extensive demo'ing Transmeta did today, their method of run-time recode and optimization works. But think of what else a dynamic microcode processor can do! The dominant x86 binaries run fine without recompile. When the dominant arch switches over to PPC/AMD64/IA64/TM256/etc, Transmeta adds a new flashed layer of microcode, and now our three year old 'IA32' Crusoe laptop will run Windows2004 for the SuperAlpha-III.

    There is nothing that ticks me off more than buying a $12,000 workstation that will be useless in four years when the manufacturer decides to switch processor arch and no longer offers support.

    --
    .sig: Now legally binding!
  296. Re:Yeah but..., yeah! by trey · · Score: 1

    So what? you still get a speedup from not having silicon implemented branching instruction sets. The software can do a better job and generate no heat in the process as well. Crusoe Assembly makes little sense. It would be like trying to write a book using ones and zeros.

    --

    he who has the fastest cart always has the best lie.
  297. DVD in Mobile Linux? by qqaz · · Score: 1

    Does this mean that Transmeta will be helping the Linux/DVD cause?

    --
    sup :cool:
  298. Transmeta misses a MASSIVE market opportunity by Morgaine · · Score: 2

    I think you've missed the main thing that's wrong with Transmeta: it's the lack of a key vision as to the possibilities of an exponential market explosion based around their product. They're occupying the valley instead of conquering the universe. Let me explain.

    What's the single most important lesson that the computing scene has learned in the last few years? Hey, that's an easy one: that the massively parallel development in the Free/Open Software community outpaces that in even the best funded closed commercial teams by a huge factor, easily two orders of magnitude. The productivity is without peer, verging on the ridiculous by traditional measures.

    Yet Transmeta is not opening up the development of Code Morphing Software to the community, apparently as a result of an incredibly regressive devotion to the antiquated concept of Intellectual Property and its laughable protection through obscurity or non-disclosure. So, instead of the next few quarters seeing a myriad of ports of CMS to dozens of architectures being attempted (and inevitably quite a few of them succeeding rapidly), we're going to be forever waiting on the closed Transmeta team to deliver. It's going to be the pits. It's going to be annoying. It's going to be slow.

    Even worse, open development would be highly likely to optimize the hell out of the x86 version of CMS in very short order, far more rapidly than TM could despite their current great lead. You just can't hold back a thousand highly enthusiastic parallel streams of development, even if there is a substantial attrition rate. In contrast, now it's going to be a straight race between TM and Intel, and in this head-to-head between closed teams, only a fool would discount Intel despite the far greater ease with which Crusoe can be optimized compared to Intel hardware.

    This is a terrible waste of a supreme opportunity. Transmeta could lay to waste all competition everywhere with Crusoe if they freed CMS from its appalling harness of proprietary development. Imagine all computing everywhere running on TM hardware, simply because Crusoe executes *all* instruction sets in current use. That's not an impossible vision, but it can't be achieved within a small closed group of developers.

    What a pity ... and 10 times the pity because the presence of Linus must have kept the benefits of community development pretty obvious within the company. A missed opportunity, possibly the biggest missed opportunity ever to occur in computing.

    [And no, their portability/flexibility argument for not opening CMS and the VLIW architecture does not affect what I've said above, because whereas the porting of CMS for a single ISA to a new VLIW architecture can be performed effectively by their single centralized team, there is no way that this scales to porting dozens of ISAs to it. That can be done effectively only in a distributed manner. (The portability benefits of non-disclosure could have been achieved quite easily through a simple system of TM design guidelines.)]

    I'm sad for what might have been.

    --
    "The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
    1. Re:Transmeta misses a MASSIVE market opportunity by Detritus · · Score: 2

      I disagree. IBM has successfully used this strategy with their AS/400 system. Hiding the hardware implementation of the architecture has given them the freedom to make radical hardware design modifications without breaking customer software. It has also allowed them to present a very high level "virtual machine" to application software.

      --
      Mea navis aericumbens anguillis abundat
    2. Re:Transmeta misses a MASSIVE market opportunity by Anonymous Coward · · Score: 0

      Re: AS/400. What works for businesses and their need for stability doesn't necessarily work for the average individual consummer who wants the latest, brightest, fastest, shiniest new toy.

    3. Re:Transmeta misses a MASSIVE market opportunity by Morgaine · · Score: 2

      Yeah, sure, it's a well-proven strategy with many accepted advantages, and Transmeta will have much success with it.

      But the universe could have been theirs instead of just the valley. The opportunity for an explosive ascendency was there and they failed to recognize it, despite having possibly the best icon for such a course in their midst. What a waste.

      --
      "The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
    4. Re:Transmeta misses a MASSIVE market opportunity by LinuxParanoid · · Score: 2

      You raise an interesting point. Open sourcing the CMS would make for some interesting market opportunities. But at the end of the day, I'm just not that the market for more instruction sets is that big. Do consumers or device designers want more instruction sets? As far as I can tell, the guys who like more instruction sets are the chip designers who want to lock in a community of software developers. And they are competitors of Transmeta; no point in them using Transmeta technology.

      I suspect Transmeta will try to get vendors of existing instruction sets (SPARC, MIPS, Alpha) to license the CMS and build their own instruction sets off of Transmeta's. If, for example, Sun ever decided it couldn't keep up the SPARC speeds with Intel and Transmeta had a better shot at that, Transmeta's chips would make a nice migration; you could run both x86 and SPARC (and Java.) I'm sure that this has occured to Ditzel et al.

      But then again, I don't buy the premise that Open Source is two orders of magnitude more productive than closed source. Lacking more rigorous analysis, NT began in October 1988, Linux began in 1991, and I don't think, good as it is, Linux is 100x better than NT. Over confidence in open source can be just as deadly as under confidence or incomprehension.

      --LP

    5. Re:Transmeta misses a MASSIVE market opportunity by Hard_Code · · Score: 2

      Who's to say they won't give out a meta-compiler for generating new ISA mappings, then take output from that compiler, produced by all those altruistic open-source people, and actually use it? They don't /need/ to open-source the CMS to gain a benefit of open-source development. All they have to do is provide a way that open-source developers can give them input that they can use to plug in an ISA. My impression is that the CMS is actually a layer /into which/ an ISA translation can be plugged. The CMS should not be dictating the mapping...some pluggable part of it should. An ISA mapping could be compiled/generated then handed to Transmeta who could then say "thanks a lot, we'll now /plug/ this into our CMS", or "great, here is a util which will allow you to add this ISA to your CMS". They don't need to open their CMS, just like Intel doesn't need to open it's CPU...it just needs to make sure the developer has the tools he/she needs (in the Intel case - a native compiler).

      Jazilla.org - the Java Mozilla

      --

      It's 10 PM. Do you know if you're un-American?
  299. MacOS vs Linux & Windows by Anonymous Coward · · Score: 0

    Just an idle thought...

    What will the MacOS do when it encounters a 2 button mouse?

    Even more interesting, how will a Mac user react when they find a 2 button mouse? Willl it cause a core dump in a Mac user? Decisions, decisions...

  300. Running a server from the Crusoe by euphorbus · · Score: 1

    I would really like to see a good lap top small scale web server is this the solution?

    --
    http://!nDolphin.com
  301. Re:battery life and underclocking by Anonymous Coward · · Score: 0

    More for Less, and better batery life. I say portable DVD and MP3 player is the go My old dauphin (386) had underclocking to save battery, and in em days fans and heatsinks were not needed. With .18 tech had 1.3 volt tech and copper, AMD can afford to underclock too, as die size=profitability. The CPU is now less that 10% of the cost of a laptop. But the darn graphics accellator and 13.1 LCD display now become the next concern. My next point is that perhaps one will be able to single step debug code, to expose obscured secrets

  302. Sounds like "SoftPC 2000" by Animats · · Score: 2
    Remember SoftPC, the old emulator that ran x86 code on Macs? The approach the Crusoe translator is using is very similar to that used by SoftPC: convert if you can, interpret if you must. They have some extra support hardware in their VLIW machine to help with exception backout and self-modifying code, two of the ugly cases for a object-code translator. That helps a lot; without it, you tend to have to generate overly conservative code.

    So it's a reasonable implementation of an old idea. The interesting question is whether it works well enough that it can make up for the lack of speculative execution in the machine. From the documentation, Crusoe is a classic VLIW machine; the CPU executes one big word worth of instructions at a time, in lockstep. Visit the Transmeta web site for details.

  303. Re:BULLSHIT. by Anonymous Coward · · Score: 0

    Post details. Heard this rumor before, but no details. Anyone ---

  304. Re:Intel/AMD don't execute x86 instructions, eithe by ecampbel · · Score: 2

    ... they are still emulating the x86 instruction set. Even with all their fancy technology a 700mhz processor will only perform, at best equivalent, to a 500mhz Pentium

    That was a quote from a Transmeta employee. I do realize that Intel and AMD processors don't execute x86 instructions. However their vector units (MMX or 3DNow) are executed natively. The MMX unit is highly optomized hardware that takes advantage of all the latest hardware techniques. Therefore, even if MMX emulation was easy (I don't think that it is), it would still be slower because it would be emulated in software. You don't see as much of a penalty while emulating x86 instructions because CISC instructions are relatively quick to translate, and both the Athlon and Pentium are also emulating the x86 instruction set.

    --

    Sig goes here
  305. Re:L2 Cache by Anonymous Coward · · Score: 0

    Read somewhere PIII has 7ms latency, Athlon more like 23, which is why PIII not dirt yet. IF Athlon can trim this number down - huge improvements in store. Intel has no more rabbits to pull out (bar moving to real copper fab).

  306. s/Gateway/eight-way/ by Anonymous Coward · · Score: 0

    Linus was misquoted (I think).. He was talking about his eight-way Pentium Xeon box when he said his system took forever to boot. (Not a Gateway)

  307. It is made with Alien Technology by Anonymous Coward · · Score: 0

    They are building The Time Machine with Crusoe

  308. Not new... by Anonymous Coward · · Score: 0
    "Code morphing" (binary translation, dynamic recompilation, whatever) has been around a long time.

    I don't suppose you've ever heard of FX32, Executor, UltraHLE, Nestra, YAE, Generator, sope, pex, or the dozens of other emulators that use this technique.

  309. Internet Connection for Mobile Linux by UncleRoger · · Score: 2
    A valid question...
    No mention was made regarding the connection to the Internet...that was just assumed to be there. But I have yet to hear about any affordable and sufficiently fast connection via mobile unit... How will they address this, or will they just leave it up to other companies to solve this general problem?

    Transmeta is a chip company. They have come up with a innovative new chip for use in a mobile platform. How that platform will connect to the internet, is up to the companies that implement the chip in their portables.

    The internet connection could be something as simple as an ethernet jack that you would plug an ordinary cat-5 cable into, or it could be an Apple-stype airport wireless LAN connection.

    Personally, I'm going to use a Ricochet wireless modem.

    ------------------------
    Help me switch to Linux!

    --
    Stupid people will be persecuted to the fullest extent allowed by law.
  310. The truth is out there... by pclinger · · Score: 1

    FINALLY!
    --

    --
    /. editors made it impossible to link to file:///c:/con/con in my sig. Please just type it in
  311. I don't quite get it. by jmd! · · Score: 1

    While I like the idea of trading off a bit of performance from today's way overpowered chips (for general workstation use, anyway) for compatibiliy, I'm not quite sure on what Crusoe provides. Can I run a win32 and linux binary similtaniously without rebooting? Because if I have to reboot, hell, I can do that with my current intel systems. Some comments mentioned performance would be terrible for RISC emulation...what other CIRC processors are out there that are even worth emulating? The only advantage I'm clear on is the power savings, which Intel can easily compete with before Transmeta ever gets off the ground. It's a nice idea, but with Intel having more of a monopoly then the Redmond crew, I don't see what we need to emulate.

    1. Re:I don't quite get it. by Guy+Harris · · Score: 2
      Can I run a win32 and linux binary similtaniously without rebooting?

      Sure, if it runs under WINE, or if you run VMware on the OS the machine booted under, and run the other OS under VMware, or something such as that. :-)

      What Crusoe provides is a processor that looks, from the standpoint of the OS and stuff running atop it, just like any other x86 processor, but that implements the x86 instruction set by dynamically translating code from x86 to the native instruction set of the particular chip. It's not providing, from what I've seen on the Transmeta Web site, any magical ability to run two OSes simultaneously, or to run apps from two OSes simultaneously, beyond what you can do on any other x86-compatible processor.

  312. Yes, Linux on the 700Mhz version by Guy+Harris · · Score: 2
    it was mentioned near the end that the vliw instructions AND the code morphing software on the two chips are different and NOT binary compatible.

    The instruction sets are different - and the binary-to-binary translation software would then be different because it has to translate to a different instruction set.

    However, as the Compatibility page on the Transmeta web site says:

    All Crusoe processors are:
    • Fully x86 compatible: they run x86 applications just like conventional x86 microprocessors.
    • PC compatible: Crusoe processors already include portions of the traditional PC support chipset, and they run all popular PC operating systems.

    so they're "binary-compatible" to the extent that they can both run x86 code.

    So possibly no Linux on the 700Mhz version?

    Umm, Linux is a "popular PC operating system", is it not? Given that, the above indicates that Crusoe processors, plural, can run Linux.

    For that matter, the Crusoe processor family page says quite explictly of the TM 5400 (that being the chip to which you're referring):

    The TM5400 is compatible with the complete range of x86-based operating systems. This includes all versions of Linux, as well as Microsoft's popular Windows 98, Windows NT, and Windows 2000 operating systems.
  313. Yeah, 8-way and very noisy. by aok · · Score: 1

    I heard it as an 8-way Pentium Xeon that took forever to boot and was really noisy.

  314. But what browser? by ceeam · · Score: 1

    .. do they use?

    You know - now we have quite a selection of hardware but still most of the browsers people use (MSIE and NN4) are crap. And others are unknown/don't have momentum/not fully featured. (With a possible exception of Opera on Wintel).

    And secondly - what's SO important about WebBrowsers?

  315. To the metal, what? by Quantum+Phase · · Score: 1

    The phrase, "To the metal..." doesn't quite have the same meaning after today. As a programmer I have always equated programming to the metal with the ultimate in speed but now there is a whole new spin to consider with this new layer of software around the core. As a big fan of both Linux and Java I feel pretty good about abstraction at this low level. It's like object oriented CPU design when such a thing shouldn't really be possible. The best part is that you get a performance gain. Very cool!

  316. Xybernaut (wearable PC maker) has NDA with by FleaPlus · · Score: 1

    Transmeta. This, of course, should mean some pretty spicy freshness for wearable PCs. I'm not certain, but I think they're currently using a 200 pentium in there; replacing that with a 700 Crusoe should speed it up a good bit and use less power. Might even cost less.

    And yes, I just purchased stock in Xybernaut (XYBR).

    Question though: does the display system of a Xybernaut system (which beams into your eye) use less power than a regular LCD?

  317. Re:Intel/AMD don't execute x86 instructions, eithe by Guy+Harris · · Score: 2
    Transmeta's "code morphing" techniques, on the other hand, do something a little more intelligent. They start by translation of x86 to native instructions. So you are running native code, with an up-front penalty for translation. Then they apply selective optimizations to tune the translated code to the native design as needed.

    The first part of this is, of course, hardly unique to Transmeta; binary-to-binary translation's been around for ages (dating back at least to the IBM System/38, compilers for which generated a very CISCy pseudo-instruction set that got translated to the native instruction set by code running on the machine; the AS/400's continue to do this, which made it easier for them to switch from the older S/38 and AS/400's with their 360-ish instruction set to the newer AS/400's with an extended PowerPC), and it was also used, I think, by HP to migrate users from the 16-bit stack-machine HP 3000's to the 32-bit PowerPC HP 3000's, and Digital's^H^H^H^H^H^H^H^H^HCompaq's FX!32 translates x86 Win32 programs to Alpha Win32 programs.

    A Smalltalk implementation for the 68K, by L. Peter Deutsch (yes, Mr. Ghostscript), translated Smalltalk bytecodes to native 68K instructions, and, of course, Java just-in-time compilers do that as well.

    I have the impression that FX!32, at least, would do post-execution optimization of the generated code based on profiling.

    What I think is new about Transmeta's stuff is

    1. the only instruction set they expose, even to the OS and the BIOS/PROM monitor, is x86;
    2. they have hardware features to make it easier for the binary-to-binary translation software to generate high-performance code.
  318. overclocking crusoe?!? by kiley · · Score: 1

    So if I read this stuff correctly, software upgradable processor. Open-source mobile linux. It should be easy enough to overclock one of these babies with a few lines of code. Then we can create a virus that will enter through this "elusive new internet connection" and fry someone's chip. Gotta love technology!

  319. SMP? by Anonymous Coward · · Score: 0

    I heard no mention of SMP. Is the hardware able to handshake with its siblings?

  320. How well will "tuned" code run? by Anonymous Coward · · Score: 0

    Lots of applications have critical areas tuned for cache and other features. Will these applications be severely impacted running on a morphed processor?

  321. Crusoe and Native Code by Anonymous Coward · · Score: 0

    I can't believe the number of people who talk out of their rear-ends when it comes to compiling Crusoe code natively. The "Smart" part of this "Smart" processor is the software! Skip it and your left with a processor that can't do out-of-order execution or any other "Smart" function. To all those people who want a native Linux kernel or apps please read the docs before you comment. Oh yea and you would need two ports one for each model since they differ on the backend!

  322. Multiple CPUS? by BrookHarty · · Score: 0

    Im moving to a dual or quad cpu systems soon. Will crusoe do multiple cpus? Its software based right? Soon as AMD comes out with thier quad system Im all over it. -Brook

  323. Re:Screw the emulation - 128 bit ? by Anonymous Coward · · Score: 0

    thanks, somebody had to explain that.

  324. Transmeta buzz by david.heyman · · Score: 1

    It would be interesting if someone, a reporter or /. hack, were to put together an overview of the various rumors that have floated around since Transmeta started up in '95.

  325. Crusoe has enormous potential yet to be exploited. by Deven · · Score: 1
    Kudos to Transmeta. They have brilliantly executed a cunning strategy, and are poised to capture the mobile computing market. Intel can't leverage their current technology to compete with this; it isn't like their competition with AMD at all. Transmeta probably has several years to establish themselves in their chosen market before a serious competitor (even Intel) has a chance to catch up with them. (Battery life is a topic near and dear to the heart of any laptop user!)

    But I've been thinking about some of the other potential Crusoe has. The possibilities are mind-boggling, and I can only hope Transmeta recognizes them and will exploit them soon...
    • Consider that with roughly 25% of the transistor count of a Pentium III, their Code Morphing software is able to give them roughly the performance of a 500-MHz Pentium III with a 700-MHz Crusoe chip. This chip executes VLIW "molecules", currently composed of up to 4 "atoms"; it has 4 functional units in the chip, and can do 4 operations per clock cycle.
    • Now consider what they could do with 100% of the transistor count of a Pentium III. Perhaps 16 functional units executing in parallel, with the effective performance of a 2-GHz Pentium III from a 700-MHz chip? (It may not have the low-power benefits anymore, but it would attract all the power users searching for the fastest chip.)
    • Even if the gains aren't quite so linear, it seems clear that more parallelism is possible, which could easily lead to speed jumps that Intel might not be able to match, given the extra complexity inherent in out-of-order execution on a superscalar chip like the Pentium III. (And would it really be a problem if that extra parallelism required 1024-bit VLIW instruction words?)
    • Worse for Intel is the potential for Transmeta to adapt the Code Morphing system to work across multiple Crusoe CPU chips for a single x86 instruction stream. This could lead to an enormous performance increase, possibly at a fairly low cost as well, throwing an amount of silicon at the processing that wouldn't otherwise be manageable on a single die with current fabrication processes.
    • Crusoe chips could be made to emulate multiple CPUs instead of a single one. This could be used to simulate an SMP arrangement, either on a single Crusoe CPU for testing, or on many Crusoe CPUs to combat the diminishing returns from trying to spread a single x86 stream across too many chips.
    • An idealized "native" instruction set (let's call it the "Crusoe instruction set") should be created for Crusoe. This should be a CISC instruction set, not RISC or VLIW. It should be a clean architecture, more like M68K than x86. Many registers should be available (256? 1024?), and opcodes should exist to support compilers and hand-coded assembly (such as low-level OS or BIOS code). There should also be opcodes and flags to access special hardware features of the Crusoe chip. It should be possible to map almost any RISC or CISC instruction (or flag) directly into the Crusoe instruction set. Like the PowerPC, it should be able to handle big-endian and little-endian data with equal ease.
    • The Crusoe instruction set could be used for new Crusoe-specific applications, without restricting future architectural changes in the underlying VLIW instruction set or the Code Morphing software itself.
    • Higher-level translation software (implemented in the Crusoe instruction set) could then directly translate almost any instruction set into the Crusoe instruction set, which the Code Morphing software would then execute as efficiently as possible on the actual hardware in use. This would allow Crusoe to emulate many different CPUs without requiring a rewrite per emulated CPU for each architectural change to the VLIW instruction set or the Code Morphing software.
    • Perhaps x86 instructions could be translated to the Crusoe instruction set, but that might be slower. For the x86 market, every ounce of speed still matters, so this may not be worthwhile.
    • A host OS could be written in the Crusoe instruction set, which would create full Virtual Machine (VM) environments, possibly with diverse requirements. (Project UDI might be helpful here.) Just imagine one Crusoe CPU (or a group; only the Code Morphing software would necessarily know) running virtual machines simultaneously for all of the following: Mac (68K), Mac (PowerPC), Windows NT (x86), Windows 95 (x86), Linux (x86) and Solaris (SPARC). (Of course, the host OS would want to create a Crusoe-instruction-set virtual CPU per virtual machine...)
    • Given a GCC backend to generate output in the Crusoe instruction set, a "native" Linux build would be trivial. (This might be more efficient than x86 Linux, although with a loss in binary compatibility.)

    On a side note, I hope Transmeta doesn't go public. Then they'll be beholden to the whims of investors who care only about short-term returns, even if long-term interests suffer. They've already got some serious funding behind them, and they have a good chance to make a killing in the market to recoup the money they've invested. What would they really gain from an IPO?
    --

    Deven

    "Simple things should be simple, and complex things should be possible." - Alan Kay

  326. Re:The Athlon!!!, ahem theres more to processing. by Anonymous Coward · · Score: 0

    Ahem, the 'traditional' is still used a lot in most applications. Even as we now have the 'cool' extra instruction sets, the file loading and other basic tasks still have to be done. And in those the MMX,3DNOW,etc don't help at all.

  327. Yaaaaaaaawwwwwwwwwwwwwwnnnnnnnnnnnnnnn........ by spullara · · Score: 1

    This is by far the lamest announcement that has ever been kept a secret. Coming out with YAIC (yet another intel clone) seems ludicrous. The only advantage they have to claim is reduced power usage! The power that the processor in laptops is consuming is becoming the power consumed by the rest of the system like the screen and hard drive. There doesn't seem to be any interesting work being done at Transmeta. Maybe I'll release a slower Intel processor next week -- I'll call it the P-II.

    --
    "If I can see farther it is because I am surrounded by dwarves." -- Murray Gell-Mann
    1. Re:Yaaaaaaaawwwwwwwwwwwwwwnnnnnnnnnnnnnnn........ by Anonymous Coward · · Score: 0

      Agreed. What you are seeing right now with all of the excitement is Slashdot naivete at it's worst. Whoop-dee-doo. So you have a processor that can emulate x86. Big deal.. there are too many clones out there to count. Oh.. it can emulate PowerPC/Alpha/MIPS too? So what? Who *really* cares about those processors anyway? In applications that matter, the *real* processors can do the job much better, thank you. Notice the fine print -- "... scales to 400 MHz" -- "... scales to 700 MHz".. meaning that there are no 400 MHz or 700 MHz parts out, and that the architecture for these chips runs out of steam sometime SOON. 128 bit? So what? What we're dealing with here is a new twist in the application of VLIW CPU architecture. I think Intel/HP has the cutting-edge of software-driven VLIW covered at this point...

  328. On Linus by ceeam · · Score: 1

    Now that they "made it" I hope Linus will have a lighter schedule probably and linux (kernel) will resume its normal evolution...

    :-)

  329. Success sceptics by drnomad · · Score: 1
    According to the ZD anchor desk there exists some sceptisism on the Transmeta success.

    OK, Linus is our hero, but you know that the best ideas usually don't have the best success. It's usually the worst ideas who have great success like: any Microsoft product, SQL, Intel x86 processors.

    I wish Transmeta the best of luck though, I reckon Crusoe could even penetrate the PC market in the end.

    --
    Bizar technology?
  330. Benchmarks by hoss10 · · Score: 1

    First of All, I apologise if this point has already been made (which it probably has!). I tried to post this last night but /. wouldn't work (had it /.ed itself - i was attempting to post during the announcement itself)

    I'm just curious about the performance specs. 400 MHz for one of the variants released today and 700 for the other says nothing. They could advertize 20,000Mhz for all I care.
    I mean, comparing Athlon and Pentium on MHz is stupid enough and this Crusoe is a completely new (and off-the-nut!) architecture.
    Even if the 400MHz refers to the virtual x86 performance and not the "native" Crusoe clock-speed I want to see real benchmarks now (Quake fps, Excel calculations etc.)
    I was quite pissed-off when the first few suits on kept emphasising that they were "going for the megahertz" - they deliberately didn't clarify but I suppose they just want a big IPO so extreme hype and FUD is needed ;-)

  331. It't all in the Microcode. by AftanGustur · · Score: 1



    since it is maintaining compatability with x86 instruction sets, it will always follow Intel's lead (and require Intel to continue leading) mainstream chip technology.


    Not quite, what they have done is to create a programmable processor. There is nothing that limits theire ability to add new fuctionality. And, unlike Intel or AMD, Transmeta can just release a upgrade to the 'morphing' software, where AMD or Intel would have to roll out a new line of processors.


    It will never run as fast as a native x86 chip would (because it must execute extra instructions). Of course, being smaller and independent to the hardware, these chips may be made significantly faster (clock-wise) than mainstream CISC/RISC chips and comparatively match performance. But not yet


    The 'morping software' is implememted as "lazy morphing" meaning, "translation-as-you-need-it".
    Although this means that the first time you run some partition of the code it will be a bit slower, overall is might actually speed up the overall execution. The reason is that modern processors have lots of instructions that are rarely used but have to be accounted for in the processors' microcode. And the microcode is basically just a huge "case" loop. Ergo: The Crusoe has the potential to actually be *faster* than it's competitors. Remember that they are changing the processors microcode on-the-fly.



    --
    Why pay for drugs when you can get Linux for free ?

    --
    echo '[q]sa[ln0=aln80~Psnlbx]16isb572CCB9AE9DB03273snlbxq' |dc
  332. How about morphing Duel, Quad, etc. Processors! by krad · · Score: 1

    How about this? To speculate what Trans-Meta is going to do for server processors! Have a multi-processor VLIW and "morph" it so it appears you only have to compile for one CPU no worries of clustering requirements, it all scales up.

  333. Crusoe == JIT Compiler + RISC Chip by Anonymous Coward · · Score: 0
    Can someone explain the difference between a "code morpher" and a JIT compiler?

    I reckon Crusoe is just a simple (therefore low-power) chip with a nifty JIT'er (in software). So, why invent the chip? Why not just use an existing low-power chip like the ARM/StrongARM?

    Looks like Crusoe will be best used as a java platform.

  334. Software processors + soft radios = ??? by gpvillamil · · Score: 0

    The Transmeta processors seem to be part of a general trend to define increasing amounts of functionality in software, rather than actual physical hardware. Another great example of this is MIT's software defined radio project (http://www.sds.lcs.mit.edu/SpectrumWare/home.html ) which basically attaches a fast DSP to an aerial, allowing a single device to emulate anything from multiple flavours of cellphone (GSM, CDMA, AMPS) to a walky-talky or a radio. Combine Crusoe with a soft radio, and you have an incredibly powerful mobile communicator, that can change its fundamental nature with a software download!

  335. They cant change the VLIW set at will by Anonymous Coward · · Score: 0

    You dont change a processor just like that... even if you have the freedom to change your ISA at will redesigning a processor is a massive undertaking. Not to mention start up costs for a .18u chip.

    Designing and implementing a new chip is much more work than creating a new compiler backend and porting the kernel IMO.

  336. Low Power + Low Temperature by GC · · Score: 3


    Wouldn't this processor be great for Overclocking enthusiasts or is there something obvious that would get in the way of this.

    If it runs at 48deg C at 700Mhz then can't we boost up the clock speed a little?

    May not be much good for mobile computing, where Transmeta seem to be focusing, and the CPUs won't be available to end-users (nor will the equipment), but isn't the potential there?

    To be honest I would like to see something run natively on VLIW instructions, surely that would rock!

  337. Re:Backwards compatibility isn't the important par by Simon+Brooke · · Score: 1
    Alternatively, the Lisp Machine people might find it a slick idea if Transmeta provided microcode to provide a Lisp-oriented instruction set that (notably) provides a really tightly microcoded set of garbage collector instructions.

    ohshit. iwantit iwantitnow.

    What killed those beautiful things like the Ivory is that they were made in such small numbers that they were fiercely expensive... This makes not just modern LispMs but also all sorts of other special purpose processing possible. Xerox were working on a really cool processor design in the mid-eighties which did selector-method resolution for object oriented languages as a single (very CISC) machine instruction. If it's possible to write your own instruction sets for this beastie, that could become a realistic possiblity.

    ohshit. wow. gimmeeeeeee

    --
    I'm old enough to remember when discussions on Slashdot were well informed.
  338. Its a step back IMO by Anonymous Coward · · Score: 0

    It promotes binary foreward compatibility... wich is fine for buessinuess. But what I personally want to see is source code portability. The amount of optimisation you can do for your architecture is always greater at the source code level. (that doesnt mean you cant still use tricks to do stuff at run time)

    Oh BTW others have already said it, but the Crusoe is very much x86 optimized.

    As for handheld stuff... if you dont care about x86 compatibility there are processors wich are more efficient. Why would you use Crusoe?

  339. Re: the universe by SEAL · · Score: 2

    The universe can be anyone's as long as they agree to distribute it under an Open Source license, eh?

    Cmon... give me a break. Simply put, it does not pay to tell your competitors all your secrets. Yes Open Source has it's benefits, but when you are running a business, it pays to sacrifice some of this flexibility for increased profit. Just ask Bill.

    This is even more of a concern for a company which has done nothing but R & D for five years (i.e. SPEND money).

    Best regards,

    SEAL

  340. Its hardware is x86 optimized by Anonymous Coward · · Score: 0

    Exceptions, floating point formats etc are hard to emulate if the underlying hardware doesnt map naturally to them. (x86->Alpha for instance)

  341. x86 instructions by Anonymous Coward · · Score: 0

    Everyone who thinks only mmx and 3d instructions are executed these days...bloody wrong! Take a look at the generated assembler code of a program you created, or install SoftIce and take a look which instructions Windoze pumps thru your hardware...

  342. GC - was Re:Backwards compatibility... by Anonymous Coward · · Score: 0
    a really tightly microcoded set of garbage collector instructions

    What exactly do you have in mind here? The problem with GC is not so much efficient instructions, but rather efficient strategy for maintaining the reachability graph.

    Hardware support for write barriers might help, but given that write's are rare (10% of instructions), and usually you can do the barrier on traditional hardware with a few instructions, you don't buy much. The difficult part is what comes after a triggered barrier.

    There is hardware that snoops the bus to maintain reachability stuff in parallel, but that is essentially independent of the CPU.

    Anyone care to speculate how to do hardware GC?

  343. Sure, x86 emulation... by ArchAngelQ · · Score: 1

    But, this thing runs Linux natively. Now, either this means they've gotten gcc working under it, or it'll need to be ported, and then, things will be able to be recompiled into natively on to a, and let me clear my throught and look intensely at all of you before I say this, 128 bit processor. This means vastly improved accuracy in all calculations, and this is especially important for floating point numbers, such as the ones extensively used by, say every 3d app known to exist (at least by me). Now, I'm just speculating at this point, but this gets me very excited. Has anyone here heard of a wonderful little box called a dreamcast? Guess what, it's got a 128 bit processor in it. Do you know what clock speed it runs at? 200mhz. That's right, it can push out it's massive frame rates running at 'only' 200mhz, admitedly with it's impresive 3d chips as backup. But what do you think it'll take to get one of the two big (or even top four) 3d chip manufacturers to make a card that'll run on this chips platform? Not much, I'd wadger, after a many thousands of crazed linux nuts start crying for them. Then consider that the low end chip that's coming out is running at 400mhz, and is priced about the same as a celeron. Now, I don't mean to get everyone thinking that we'd all be living in some fantasy world with these chips in everything, but I am pointing out that there is some missed potential. Anyone with more techinical knowlage of both 128 bit computing, and the Dreamcast's platform, want to shoot this to crud, or perhaps agree with me?

  344. money? by digitalunity · · Score: 1

    If you haven't noticed, they have been working non-stop for 4 1/2 years. With no IPO money. At the top, it says they spent a sum of money 'well north' of $100M. When sales pick up their debts, they will be a privately held, debt free fabless R&D company. How many of those do you see around?

    --
    You can't legislate goodness. Let each to his own destiny, by will of his freely made choices.
    1. Re:money? by um...+Lucas · · Score: 1

      They got their $100 million from people such as Paul Allen's Vulcan Ventures among others. None of those financiers are going to invest that much money unless the business plan includes mention of the company going public. If he'd used any other words besides "rounds of financing" then i may have doubted it, but that's CEO speak for raising money prior to the IPO.

      And yes, they won't have fabs. Is that good? That means that rather then themselves having to come up with the money to buy the fab (a material asset) they need to use that money instead to pay IBM (cost of sales). In the short term, it leaves them with more money in their pockets, but in the long term it means that there's an added expense that will get passed on (even though the price range is only 60-90 dollars.)

  345. Forget about this different instruction set stuff by Anonymous Coward · · Score: 0
    All those drooling about how one could run Mac applications together with x86 ones on one desktop: forget it.

    The whole code translation stuff will have to be loaded into the CPU, since loading such important stuff with 128Bit instruction set via external DRAM all the time will be much too slow. The size of internal memory used for this will be chosen in a way that it will not suffice for more than one processor emulation. So this would all have to be swapped out and other stuff swapped in on every context switch between tasks for different processors.

    This will be one bear of a context switch.

    Of course, if you did not multitask at all, things could work quite nicely.

    Unless there were a sizeable market for running different processors quasi-simultaneously, I would not expect TransMeta to come up with a processor with oversized translation program memory.

  346. Emulators? No, sorry. by ArchAngelQ · · Score: 1

    I'd just like to point out for the people who missed it, not rudely, but just to make you perhaps aware of it, that the software emulation layer called code-morphing works at the acceptable level that it does because the chip was built to support the instructions most used by the software layer, i.e. the direct equivilents of the intel instructions. So while other code morphing layers certainly could be done, they performance hit would be quite a bit more noticable. And that's assuming anyone else can figure out how the whole code morphing software works (I somehow don't think that transmeta is going to open source this particular little jem).

  347. some other thought by renoX · · Score: 1

    One of the strongest selling point of Transmeta is that their processor is low-power and 80x86 compatible.

    Well the compatibility is very important for closed source software but unimportant for open-source type where you can easily recompile, so a StrongArm could be a real competitor here,
    not only they are cheaper I think but also more power efficient and already here now. On the other hand ARM CPU and the like often have no Floating-Point Unit which can be a bad point for some application.

    1. Re:some other thought by GregWebb · · Score: 2

      Key phrase: the compatibility is very important for closed source software but unimportant for open-source type where you can easily recompile. You really expect the average user to recompile their software? Not going to happen, sorry.

      I agree, StrongARM would be a good one if they could find a way of getting it into this sort of market. They don't seem to know how to yet, though.

      Greg

      --

      Greg

      (Inside a nuclear plant)
      Aaaarrrggh! Run! The canary has mutated!

  348. Memory footprint is going to be a dog by Anonymous Coward · · Score: 0
    If they're caching 'morphed' code in memory, which it sounds like they are doing, then memory requirements are going to be much higher than normal chips, not to mention bus traffic.

    I don't see how 'morphed' code can exist only in the CPU, since it's software generated. Help, my brain is exploding!

  349. Wireless Internet infrastructure already exists by evilandi · · Score: 1

    Re: Crusoe's aim at wireless Internet appliances

    Linux Paranoid wrote: Wireless internet is cool, but I find it hard to be optimistic about the per-month pricing over the next 3 years at reasonable bandwidth rates attracting serious (5+ million) consumers. Guys putting up towers and satellites are the bottleneck here, as is the degree of competition.

    We have 98% digital cellular coverage of the entire landmass (that's landmass, not population). GPRS (128kb/sec+) cellular bandwidth goes live nationwide this year. IMT (2mb/sec+) cellular bandwidth goes live nationwide in two years. Those two services are software upgrades to the existing hardware. No-one needs to errect any more masts or launch any more satellites.

    And that's not to mention Digital Terrestrial Television which is right now pumping 50 channels of MPEG TV (widescreen, DVD quality) into my living room (no cable, no dish, just a normal TV aerial). You heard me - right now - for US$10 a month. Cartoon Network humming away as I type. They are already trialling Internet services over Digital Terrestrial as we speak - although admittedly the bandwidth is downlink only.

    I don't see a problem here. The infrastructure for wireless Internet already exists.

    But then I live in the UK.

    --

    --
    Andrew Oakley - www.aoakley.com
    1. Re:Wireless Internet infrastructure already exists by Anonymous Coward · · Score: 0

      In hope for more precision:

      You are right, but GPRS and IMT (alias UMTS alias ...forgot that one) is _not_ a software only upgrade for mobile operators.
      HSCSD (57kb/s by channel bundling in GSM) is/was
      a mere software upgrade for mobile operators.

  350. question about emulation by stge · · Score: 1

    Can someone explain how this emulation thing works ? I read that the chip can run Apple or Win software. But doesn't Apple software, say, need libraries that are part of the MacOs ? And doesn't the MacOs rely on the Apple architecture, not just the powerpc chip ? It seems to me that you would need more than just a chip with code morphing to run software from several platforms on the same architecture. Am I wrong ? Can someone more knowledgeable than me explain ?

  351. Re:I guess variable CPU speed Linux problem solved by WinTired · · Score: 1

    • nope. from what i gather this just goes to sleep and back up again at full speed with noops/hlt for power saving. variable speed asynchronous machines cant be handled by linux due to its bogomips setting..

    I think you got it wrong. They claim their LongRun Technology actually reduces both clock speed and processor voltage on the fly, rather than taking this on/off approach.


    -------------------------

    --

    -------------------------
    "People ask FAQs all the time". - David Allen

  352. DVD code is tight... how about a kernel compile by Anonymous Coward · · Score: 0

    Games and DVD stuff should run well enough. Although I would like to see some game benchmarks too.

    The real kicker will be stuff like kernel compiles emacs etc.

    Between development games and entertainment I have about covered all my needs for high speed processing. Dont need much speed to read my email or browse... I dont think this thing is suited for handhelds or pads.

  353. The Ultimate X86 Debugging Platform? by Anonymous Coward · · Score: 1

    I want a Transmeta chip for one reason: With a little help from Transmeta, it will be the IDEAL platform for performing low-level debugging of X86 applications.

    Why? Except for debugging hardware interfaces and a very rare device driver (which rely on the actual low-level hardware interface), there is no need to develop X86 software on a "true" X86 chip. All that is needed is something that ACTS like an X86 chip, and when performing low-level debugging on complex software, the ability to examine the internal state of the processor at any given time, and to use a debugger in real-time, are both very important.

    What if Transmeta built the guts of a debugger into the Code Morphing software? That would allow the developer to easily set the equivalent of hardware breakpoints without having to go through the high overhead of software interrupt tricks or the expense of in-circuit emulator hardware. Both of which also require all on-chip caching to be disabled, further decreasing the speed of the debugging process.

    It is my guess that such a debugger is ALREADY part of the Transmeta Code Morphing software, for I see no other way for them to implement and debug the X86 ISA itself.

    Even if the "first pass" Transmeta performance rate is just 10% of an Intel or AMD chip at the same clock rate, under a low-level debugger, that rate would hardly be affected, while the AMD or Intel perfromance would drop by a factor of 100x to 1000x due to the overhead imposed by the debugging environment.

    In my book, that puts low-level debugging on a Transmeta chip 10x to 100x faster than on true X86 hardware.

    Big deal? Well, what if you want to run some processes (say, those that manipulate hardware) at full speed while you debug one or more other tasks? For true low-level debugging, Intel and AMD caches must be disabled, where the Transmeta caches would not have to be.

    I have worked on many real-time systems where the state-of-the-art in low-level debugging environments was terribly inadequate. I have had to resort to full system simulations to get some timing problems resolved, an extreme cost to bear for what is a latent "design flaw" in the processor architecture.

    Try debugging a settop box where the MPEG decoder is 85% in hardware and 15% in software. The two have to work together at full speed, and even with good software, the system can't run at all when the cache is disabled. Debugging must be done at full speed. Given that, how does one detect and react to a "suprious" event that could be due to hardware, software, or the combination? The debugger needs not only to be fast, but it must also be "smart" enough to know a lot about its environment, if it is to be of much help.

    There are many similar examples, for example in robotics, where real masses, velocities and accellerations must be accounted for when a reaction is being comtemplated, not to mention the delay associated with computing that reaction. Again, fast debugging at a very low-level is extremely helpful.

    Transmeta's got it. I need it.

    -BobC

  354. Crusoe is bad for everyone by Anonymous Coward · · Score: 0
    Despite all the claims about 'revolutionary' stuff, it doesn't seem that clever at all. The main change seems to be in terminology: 'molecules' simply translate to the EPIC instruction groupings in Merced; 'code morphing' is the same as the x86 instruction decoding that happens in Intel processors, just done at a different level. They go on about the cost in transistors of hardware-based x86 decoding, and everyone agrees with them, but what they propose is a half-arsed, middle-of-the-road, 'third way' which combines the downsides of both alternatives. Both Transmeta and AMD seem to be determined to keep the vile x86 instruction set alive; ironically it is Intel who want to kill it. The whole RISC philosophy of 'let the compiler optimise' has won, and the Merced architecture takes it to its logical conclusion. Crusoe is a step back, in my view: instead of having a choice of optimising compilers, or even, gasp, writing your own optimised assembler, you're tied into Transmeta's dubious software translation engine, which has to optimise locally at run-time, meaning it will never produce even half-way optimal code. What they've come up with, it seems, after all that money and time, is yet another low-cost, low power consumption RISC processor, and a bit of emulation software. So what?

    Worse, from an open source point of view, is the fact that they won't be releasing the source to their translation engine. They seem to appease open-source advocates by a) employing Linus Torvalds and b) calling it their 'intellectual property'. Well that's OK then. We can all be held to ransom by a company peddling their closed-source solution to everything. Doesn't that sound just a bit familiar? With Merced the compiler, and hence ultimately the programmer, gets to choose the packaging of instructions. With Crusoe Transmeta does it for you, badly. If you really care about open-source, you should be supporting Merced.

  355. Re:Intel/AMD don't execute x86 instructions, eithe by dopeghost · · Score: 1

    There I am thinking how lame .sigs are getting these days but thats just classic (I really feel for you)

    --
    This UID is 7651 digits too high to subjectively infer IQ from.
  356. Re:Transmeta can design a whole new instruction se by Karellen · · Score: 1

    Someone moderate that up some more.

    If the x86 instruction set has as much cruft in it as I've heard some people say, due to keeping backwards compatibility from the 286 while adding new stuff on for each generation since, then writing a nice, new, shiny, cruft-free instruction set for the Crusoe (having learned what's good and what's not from the last x generations of processors) could be A Good Thing (tm).

    It might be faster than x86 emulation (with fewer instructions to translate, the instructions in the translation cache might stay there longer with fewer misses or something) and compilers for it could probably be ported in a relatively short space of time if it was clean enough.

    Hmmmm....a nice new 64-bit OS running on cheap hardware? Sounds good to me :)

    --
    Why doesn't the gene pool have a life guard?
  357. Multiple processors? by MrKipling · · Score: 2

    Seeing as the power consumption is so low, couldn't someone put two or more Crusoe chips in a laptop, increasing the performance, while still having a superior battery life to anything that's currently available? Also, the code morphing software could perhaps be adapted to make all applications run multi-threaded, regardless of how lame your OS is. These boys have really cooked up something special with this chip.

  358. Other instruction sets by Anonymous Coward · · Score: 0

    How about Java ByteCode? Instead of a JVM which does ByteCode->x86->Crusoe VLIW, why not go straight to Crusoe VLIW?

  359. Crusoe can never outperform an Athlon! by exa · · Score: 1

    Think of it! Athlon got all the translation hardwired, and all optimized specifically for x86+MMX stuff. There's no way a software optimizer could be faster. Really.

    But it could be cooler anyway :)

    --
    --exa--
    1. Re:Crusoe can never outperform an Athlon! by Anonymous Coward · · Score: 0

      Think of it! Athlon got all the translation hardwired, and all optimized specifically for x86+MMX stuff. There's no way a software optimizer could be faster. Really. Hardware can only know the past. A compiler can know the future: it can look ahead to see what the program will do after the instruction it is currently generating and make different decisions based on that. A software optimizer can be faster.

    2. Re:Crusoe can never outperform an Athlon! by exa · · Score: 1

      :I
      As someone whou took a couple of courses on compiler design/implementation and theory of computation, this just doesn't sound convincing.

      How is your compiler supposed to look ahead? please.. It can just do some lazy translation and cache that, and it seems just that way as I read from transmeta's site. It's just an interpreter based emulator with memoization. Whatever, of course the number of translation rules you can write are FINITE. And recall that these are lousy low-level translations, so it's probably not doing global optimization based on data dependency analysis. That chip won't be "smart" unless there's some strong AI running on it :) Hey, I like it though, but I suspect the internals are a bit "tuned" for the x86...


      --
      --exa--
  360. Not exactly earthshattering. by Anonymous Coward · · Score: 0

    So that's it, eh?

    The object of all this hype from slashdotters over the years?

    An x86-compatible CPU. Wow, big smegging deal!

    So the power consumption (under low-use conditions only, if you do the calculations) is lower, but that's it.

    Sorry, folks, but that ain't the Revolution you were all expecting. That's pure evolution, nothing more.

    Yes, they *claim* they can do emulation of other CPUs. But they don't offer it yet, and there's no guarantee they will.

    Interesting how nobody is bitching about the inherrant closed-source nature of the system too. Interesting...

    Wake me up if Transmeta actually come out with something new, not just Yet Another x86 Clone, yeah?

  361. Some thoughts on slashdot on transmeta. by toppk · · Score: 1

    First, in the update, this guy quotes linus as saying his "gateway" takes a while to boot. He said "8 way" as in 8 CPUs. Second, I though the slashdot questions at the conference were dumb: effectivly the question was "Can you run different instruction sets on the virtual machines" but the virtual machine itself was provided by their software. The question was not based on the information they released, but on earlier rumors. They could have gotten a better answer if they asked "what are the potentials for running different virtual cpus simulaneously" Some questions I'd liked to have seen were: 1) You say the software in the cpu is locked right in the beginning of the bios loading, no chance for viruses. but you also say that you will be able to download fixes off the web. Explain that! 2) How long does it take for writing x86 emulation, and why (x86 complexity, clean room stuff?) 3) How would the transmeta CPU running a simpler (designed for lowpower) CPU instruction set, like dragonball, strongarm compare. Question 3 is especially good, as especially with linux, x86 isn't all that. All they seem to have is the low power x86 chip, but that doesn't mean that it's the low power chip.

  362. OXYGEN? by griffjon · · Score: 2

    So, what does transmeta mean for the Oxygen project at MIT? I heard Deterzous' talk at RSA keynote on the handheld cellphone/palm-pilot/pager/etc.etc/ that links up to a stationary system (desktop)--I'd bet that the Crusoe will be the backbone behind Oxygen (I'd be surprised if there isn't some collusion already) and if not the Oxygen system, some decent portable connectivity device.

    --
    Returned Peace Corps IT Volunteer
  363. I got to get me one of these by RPoet · · Score: 1

    cool stuff.

    --
    "Oppression and harassment is a small price to pay to live in the land of the free." -- Montgomery Burns.
  364. BeOS in this matter. by be-fan · · Score: 1

    Well, since no one has made a comment on this as of yet, I guess I will. Transmeta has domeonstrated BeOS running on Curusoe. Considering Stinger (Be's embedded product) is almost nearly BeOS, it should be a chinch to put Stinger on Curusoe. That means two things. A) Be gets a huge boost considering stuff like the Compaq webpad (which uses Stinger) can be made much more powerful, and B) It helps embedded products in general since Be is probably the best suited general purpose OS for embedded products. Speaking of OSes I'm not too particularly impressed with the whole embedded linux thing. In my experience, the good Linux GUIs (ie. Windows maker, E, KDM) need a fast graphics card and lots of RAM to work well. Not to mention all the hardware X needs. Can something like the compaq webpad be made that uses X as the GUI? Even the most basic linux X system needs a few hundred meg and that simply isn't available on a webpad type device that uses flash RAM cards. (Or even microdrives) Not to mention all the cost for RAM. QNX would also be a good platform (better than Be maybe) but the lisencing costs would be a bitch. Another problem is the fact that the users of these embedded products are about half as smart as users of normal computers.(not everyone, just most of them.) Were talking iMac user level here. (not mac users, iMac users) I don't see them being too happy trying to configure all the options in GNOME. So reaching the end of my rant it is time to make some forcasts.
    A) Be makes a killing of Curusoe by putting Stinger on every webpad out there.
    B) All the Linux heads go out and buy a webpad then install linux on it.

    --
    A deep unwavering belief is a sure sign you're missing something...
  365. Re:16-bit instructions by CodeShark · · Score: 1
    However, the PDF for the TM3120 says:
    • When used in conjunction with Transmeta's x86 Code Morphing software, the Crusoe Processor provides x86-compatible software execution without requiring code recompilation.

      Systems based on this solution are capable of executing all standard x86-compatible operating systems and applications, including Microsoft Windows 95, Windows 98, Windows NT, and Linux.

    No Listing for Win3.1.1 or DOS--perhaps because those OS's aren't really out there for most of us anymore. So IMHO -- and I assume you'd agree --Transmeta really needs to clarify this issue better.
    --
    ...Open Source isn't the only answer -- but it's almost always a better value than the alternatives...
  366. Thank you for the clue! by DarkClown · · Score: 1

    Went to check mac-on-linux out with an OS9 cd last night on my linuxppc box, it complained about the lack of a system folder (the OS9 cd isn't bootable), so I'll check it out this evening with an 8.5 CD. It sounds great! It's a wonder I haven't heard of this before - there was much talk about something called sheepshaver a while back, but haven't heard much about it lately...

    1. Re:Thank you for the clue! by puetzk · · Score: 1

      I think you'll need 8.6, or a real (not newworld) rom, which only works from a small selection of machines. 9.0 should work though... the CD's not bootable?

      --
      The Matrix is going down for reboot now! Stopping reality: OK. The system is halted.
    2. Re:Thank you for the clue! by DarkClown · · Score: 1

      I think you'll need 8.6, or a real (not newworld) rom, which only works from a small selection of machines.
      Aha. I got some new set of errors last night off 8.5 (forget exactly what - in mac os at the moment) - so 8.6, eh? Shoot, I just downloaded the upgrade for that off apples site, I suppose I'll need the cd - maybe someone I know has it.
      9.0 should work though... the CD's not bootable?
      Nope, no system folder to boot from, BUT -this is from a GM that a friend that works for apple brought over. I'm going to go ahead and buy it - still running 8.6 on my macos side and should upgrade anyway. Thanks again- I'll get it going, by cracky! The only bummer about this is that at home I'm locked down to 1 ip address dues to the pipe50 router being thus restricted, but I suppose I could use IP Masq on the linux side to get the mac-on-linux deal on the network. That could be interesting.

  367. Code once, run in any HW environment, also Crusoe. by Anonymous Coward · · Score: 0

    Write a program in C, C++ or in JAVA, compile once and then run it on PowerPC, x86 or on Crusoe, using ElateRTOS.

    Or, if you wish, run it on Linux, Windows or...

    read more: http://www.tao-group.com/2/tao/index.html

  368. Oh my, aren't we naive? by Anonymous Coward · · Score: 0

    Oh my, aren't we naive?

  369. sh*t happens, now fixed: Re:Mobile Linux? by Anonymous Coward · · Score: 0

    >Also, this processor could be pretty cool for
    >running the tao virtual machine that amiga are
    >using for their network environment - remember,
    >amiga have already said they're going to run
    >stuff on top of linux in palmtops.

    And note that even if it(the taoAmiga -thing) sounds like the Gateway AmigaOE/AmigaObjects, it is a totally different beast...

    This time they deliver .... hopefully... 8|

  370. Re:16-bit instructions by Guy+Harris · · Score: 2
    No Listing for Win3.1.1 or DOS

    Unless Win 3.11 or DOS are not "standard x86-compatible operating systems", the note on the TM3120:

    Systems based on this solution are capable of executing all standard x86-compatible operating system

    sure seems to imply that the TM3120 can run Windows 3.11 and DOS code.

    So IMHO -- and I assume you'd agree --Transmeta really needs to clarify this issue better.

    You assume incorrectly - the people who seem to think that they haven't stated pretty clearly that both chips are completely x86-compatible seem to me to be trying, for some unknown reason, to read something into Transmeta's statements that isn't there, and I don't think it's particularly worth Transmeta's time and energy to worry about people who choose to read "the complete range of x86-based operating systems" or "Fully x86 compatible" in non-obvious ways.

  371. Re:Intel/AMD don't execute x86 instructions, eithe by Stradivarius · · Score: 2

    There's more to consider here...

    1.What Transmeta is doing isn't truly emulation, at least as it is traditionally known. In, say, a Pentium III, the chip uses hardware to decode x86 into a series of internal RISC-like operations. This decoder hardware is a huge beast, taking up a tremendous percentage of the overall silicon and is responsible for most of the power consumption issues. In Crusoe, however, Transmeta moved most of the decode functions to a low-level software layer. Which, being software, has the *potential* to inhibit performance, though saving power. However, they suddenly don't have this huge hardware decoder to deal with, and so can devote that silicon to extra execution units (integer, FP, etc). Which will help increase performance significantly. Whether this is {enough/not enough/more than enough} to compensate is an interesting question, which I don't really think any of us have enough info to truly answer just yet. Also keep in mind that the decode hardware a la PIII can have a negative effect on the sorts of clock rates you can achieve, so moving it to software may help with that as well. Complex decode hardware also forces a larger branch penalty in case of a branch misprediction.

    2. Not only can they devote extra silicon to execution rather than simply decoding, but doing the software translation to native VLIW adds further performance benefits. The first of which is this: it allows Crusoe to perform optimizations on the incoming code which would be impossible with a purely hardware chip.

    One of the biggest of these is dead code elimination. x86 compilers commonly save a variable out to memory, then load it back in shortly thereafter - this is because of the very limited number of official, or "architected" x86 registers. The compiler thus ensures that it always has the correct register values. Much of the time, though, the load is unnecessary because the register wasn't modified (the compiler was just being conservative). Transmeta's Code Morphing software can detect these situations and actually remove unneeded loads, something that no purely hardware solution can match. And loads tend to be a big performance inhibitor; due to the long memory latencies, as well as the fact that loads typically start long dependency chains, where subsequent instructions require the load to finish before they can start.

    There are several other compiler-like tricks that the Transmeta software can (and does) perform in realtime (and with runtime data that compilers don't possess) that can greatly speed up the performance of code.

    3. Also, consider this: traditional x86 chips have had to decode instructions every time they execute. This is a pretty big overhead. Crusoe actually caches intructions in decoded, optimized form in a hardware "translation cache". This means that while yes, you do a software decode once, which could be slower than a hardware decode, you only have to do it that one time, versus every time with hardware decode. Thus the cost of software decode is amortized over many many cycles, and might actually be more efficient than hardware. (think about it: 99.99% of the time you're going to be executing native, optimized VLIW instructions coming from the translation cache. Only the remaining small bit of time do you actually have to decode).

    4. There is also something to be said for the fact that this chip is targeted at the mobile arena. It's not meant to be the desktop/server powerhouse. The needs of mobile consumers are different. Mobile uses typically use only a small fraction of the processor's capabilities, so even *if* the Transmeta chip was slightly less strong on benchmark performance vs a PIII, it's overwhelming superiority in power consumption still makes it a big win, since you can't notice the slight performance difference anyway. Probably the most taxing thing you might do with the laptop is play DVDs or something similar, which the benchmarks show Crusoe handles just as well as a PIII - and at a fraction of the power.


    Oh, and just as a side note: CISC instructions are most certainly not "relatively quick to translate". One big reason behind Crusoe's existence is that CISC decode/translation is a difficult, long-latency operation, that uses a lot of hardware. Mostly because of the variable-length instructions, which forces the processor to use a largely serial decode process. Which eliminates the primary advantage - parallel processing - of hardware. Said decode hardware could be put to better uses. MMX/3DNow instructions are probably easier to translate, being a newer, cleaner extension to the instruction set. Execution may very well be more difficult than standard x86, but the translation itself shouldn't be that bad.

  372. TM is doomed by Anonymous Coward · · Score: 0

    Intel and AMD will happily (and easily) crush the foolish team at Transmeta. Its only a matter of time.

  373. Re:BULLSHIT. by Score+Whore · · Score: 1

    Look through these. There are several documents that indicate which processors have had updates released. Additionally there links that point to additional information re. what's changed, etc.

    From the looks of things, it's a PII, Celeron, PIII feature and not in the PPro.

  374. Damn straight (no text) by Stradivarius · · Score: 1

    I said there was no text :-)

  375. Re:Is the microprocessor so important? - yes by McKing · · Score: 1
    For example, the StrongARM achieves that low power rating (in part) because
    it leaves out a bunch of processing units typical of a desktop CPU - eg it is integer math only, no floating point unit. My
    suspicion is that Transmeta must be doing something similar to achieve this.


    If you saw the presentation yesterday, you would know that nothing was "taken out", per se. The whole point of the Crusoe was to start from scratch, design a simple, elegant, low power, and fast VLIW CPU core, and then implement software that allows the CPU to be x86 compatible. It has floating point (which the StrongARM lacks), but it does not have a lot of things that you find on a "traditional" CPU like x86 and Sparc. All of the branching, out-of-order execution, and similar logic is implemented in the software translation, so that by the time the instructions reach actual silicon, they are already as optimized and in the the optimal order as they can be. Most CPUs have HUGE areas of the silicon devoted to such things, when causes them to draw more power and generate more heat.

    Also, another advantage shown in the presentation is their "Long Run" power management system (which I supect came about as a side-effect of the code-morphing). Long Run continuously monitors the running code, and throttles both the speed _and_the_voltage_ of the CPU when the full performance of the CPU is not necessary. The example they showed (while running a DVD) was that by combining both, they were able to use only 25% of the power compared to a PIII. Cool.

    --
    If only "common" sense was actually that common...
  376. you wouldn't want to by Stradivarius · · Score: 1

    The Great Idea (tm) that Transmeta had was that by doing the instruction decode in software, they decouple the requirements of legacy instruction sets from the hardware implementation. This allows them to use the latest and greatest in hardware design techniques, and even change the internal instruction set to get more performance. They then just tweak the Code Morphing software to recognize the new internal instructions, as you mentioned.

    The big reason why not to code directly for Transmeta's internal instructions is that you'd have no market share. x86 is around yet almost solely because of the "golden handcuffs" of backwards compatibility.

    There's also the fact that if you were to code directly for the VLIW internal instructions, you still need to use the code morpher (that's just the way the chip works), so you probably wouldn't derive any benefit. And even if you could bypass the morpher, then you would be depriving yourself of the optimization techniques the morpher uses to improve performance.

    Taking this further, Transmeta *could* release a "static" external instruction set that the Code Morphing software translates into whatever the current native set is. Then we don't need other architectures *at all*. All we need is to use this static external instruction set. We don't even have to worry about being compatible with future releases of Transmeta, since the Code Morphing software takes care of that.

    They have...it's called x86 :-)

  377. Re:pMSNBC misspells Linux. Coincidence? by Anonymous Coward · · Score: 0

    Find it spelled linix, win a prize!

  378. Re:Intel/AMD don't execute x86 instructions, eithe by WGR · · Score: 1

    Essentially the difference between the Transmeta and traditional microcode is that the Transmeta uses a "compiler" for the x86 language while trational microcode a la Athlon/Pentium III uses a massive hardware base interpreter.

    What you lose on the initial compile, you make up on the execution and since you see all the resulting code, you can optimize it.

    What we have is the revolutionary idea of treating x86 instructions as a programming language, then applying compiler techniques to generating microcode out of it.
    As mentioned, MMX instructions are probably hard to compile in the same way as vector operations in higher level languages are. But it still can be done.

  379. Re:BULLSHIT. by Score+Whore · · Score: 1

    Also, see pages 8-31 through 8-50 (pages 303-332 in the pdf) of Intel Architecture Software Developers Manual, Volume 3: System Programming. It provides some details about the process of updating the microcode.

  380. PalmOS on crusoe? by Grey+Mouser · · Score: 1

    Since the PalmOS is the leading handheld os shouldn't transmeta be trying to port Palmos to Transmeta chips? Or are they going after a different market? Or are they trying to replace PalmOS with portable linux? I'm not sure what class of machines they are trying to position their chips for?

  381. Re:L2 Cache by higuita · · Score: 1

    >Read somewhere PIII has 7ms latency, Athlon more
    >like 23, which is why PIII not dirt yet

    this is because the P3 have his L2 cache on chip and the K7 dont...
    the P2 neither (that have about the same latency as K7)
    when the K7 move the L2 cache to inside the chip, the latency will drop alot...
    the new latency will be a little more than the P3 because AMD will want to be able to push the clock higher and too lower latency would stop that...

    the L2 cache in the P3 is at full speed, K7 and P2 have it at half speed of the cpu (2/5 to K7 800)
    if K7 is still faster now than the P3, his new version will be alot faster

    --
    Higuita
  382. Re:THANKS by Anonymous Coward · · Score: 0

    Yup 2048 bytes, updated by WRMSR instruction. look like functions only, and you could recover. Possibly enough to disable/subsitute CPUID if line of thought carries to p3. Or add lots of wait cycles - or prevent software copying in hither to unknown ways.

  383. link to arstechnica "in-depth look" by treebeard77 · · Score: 1

    Ars's resident CPU-meister, Hannibal, has taken an in-depth look

  384. Re:THANKS by Score+Whore · · Score: 1

    From a superficial reading you can send more than one update to the processor, just 2k at a time.

  385. Chuck Moore's ShBoom Chip by linuxrus · · Score: 1

    ya'll remeber good old chuckie cheese's chip? Well the benefactor ifin you don't remeber of this fine chip was Patriot Scientific They just announced the .35 ...seem ta me in synch with Torvald's Crusoe... I almost fell off my chair in the lab today when i saw this stock quote on PTSC today http://quote.yahoo.com/q?s=PTSC&d=v1

  386. Programming by Anonymous Coward · · Score: 0

    Its more the compiler then the programming which is important. People code the same, but if they use the right compilers they optimize for this and this instruction set - which also means, it will run on other chips not using those instruction sets, albeit slower. So this is what we call Software optimization for XXX (eg 3dnow and SSE etc.) so weird disussion you guys are having.