Slashdot Mirror
Crusoe and Benchmarks
duffbeer703 wrote to us with a ZD story regarding Transmeta's Crusoe speed and benchmarking. As we've heard the benchmarks haven't overwhemled people - but are we measuring things the wrong way? Of course emulation is slower then native chipsets - that's a given - but are the other elements of Crusoe enough to make up for it?
Exactly... if I want a mini-laptop like one of those slick VAIO's that's based on the crusoe, or a cool webpad, or something similar, I don't care how well it will run Quake, 3DStudio, or how many Linux kernels it will compile per second. I'd want to be able to run a browser, a word processor, an email client, and maybe a DVD player. My old K6-III/400 handles that beautifully. Price, heat dissipation, size, and power consumption are all one should worry about in this case.
"Evil beware: I'm armed to the teeth and packing a hampster!"
Lex orandi, lex credendi.
Sigh. It's a troll, but I'll bite...
You may not run the same program twice in a row. (I do. Often. But let's ignore that for now.) But... Within a single program of any complexity, there will be code that gets executed repeatedly. It might be the menu-redraw routines, or the file load/file save routines, or the search routine, or the "EMACS has consumed all available memory" error dialog... whatever; there is code that will be executed repeatedly.
Crusoe will most likely perform better than the benchmark shows in these real-life instances. Benchmarks are an artifical environment... how many benchmarks incorprated floating point calculations before FPUs became common? Given that Crusoe is the first chip of it's kind, I'm not surprised that it does poorly on current benchmarks, which were written assuming that performance on the benchmark would not change significantly from one run to another. If Crusoe (or other code-morphing chips) gain in importance, we'll probably see new benchmarks designed to test the new capabilities of these processors.
"Great men are not always wise: neither do the aged understand judgement." Job 32:9
We need a good set of common use benchmarks. Things that run the spellchecker in Word while trying to play MP3s. Then the tests need to be done, on these mundane tasks, with non-virgin crusoe chips. Chips that have been run through these routines and have adjusted themselves to those tasks.
WARNING: there is a trojan on your
The goal of a benchmark shouldnt be a best case scenario. Maybe I'm wrong, but when I see a benchmark, I want to be able to think 'this is the least I will get out of product XYZ'. I dont want to see benchmarks which have had every extra .0001 tweaked out of it. If I see a benchmark exceptionally high, only to get the product home, and see it low, what has the product gained, other than my dissatisfaction?
---
Video meliora proboque deteriora sequor - Ovidius
Hell, I'll probably find one pushing my Howard Miller clock some day. Why did they need a CPU to do this?
Transmeta's edge is small, fast-enuff, low power consumption and flexibility. I swear too many articles pop up on /. questioning its viability and not enough people spend time in at Best Buy, Fry's, Circuit City, et al, watching masses of perfectly happy buyers fish out the plastic for something they perceive will make their life better.
As you rightly point out, this is not the number cruncher and stupid people like me (who try to do photo editting on a 300mhz win98 choked laptop) would best remember the AMDs, PIIIs, Alphas, et al are the CPUs for the serious crunch work.
Yeah, I'm sure I'd like a 1GHz 64bit laptop stuffed with DDR SDRAM and a RAID5 Array of UW-SCSI drives, but these boogers really do suck to do much work on and this isn't going to make my cel calls any clearer when I order a thai take-out "Thzz wzn gzzmpd zhhhzh bzzf wzh zhoo rizz, yzz?" "No! wzh zoozfzz, z rizz!"
Better off lounging in one of them Cyber chairs, hands on keyboard and pasting Natalie Portman into photos of me with my backyard Space-O-Rocket built for two.
--
A feeling of having made the same mistake before: Deja Foobar
We're still not measuring the correct quantities: consumer value and whether the damn thing will have to be recalled within the first week of hitting the market because of some stupid FOOF error like Intel's have been known to do.
You make a correct point, but you don't take it far enough: what's important is not whether Crusoe is clock-for-clock equivalent to Intel's and AMD's offerings. The question is whether a device can be put together using Crusoe at a reasonable price that taken as a whole, fulfills consumers' needs.
The issue is not so much that Crusoe cannot compete in markets currently dominated by Intel and AMD. The question is whether Intel and AMD are the appropriate choices for those markets to begin with. For most consumer devices, they're overkill, both in computing power and electrical power dissapated. Crusoe has a bright future in these markets, and if it were publicly traded, I'd consider picking up a couple shares.
-- Anne Marie
It's the economy, s....d!
Seriously, the following things matter:
1. longer CPU life for same wattage (power times time) means CPU can sleep better
2. lower power CPU sleep mode means longer non-use time
3. lower power CPU means more RAM, which means less hard disk, more caching
4. both of the above in a proper system mean a 4 hour webpad now runs for 6 hours.
5. lower battery usage means lower weight
As you point out, manufacturers need to work on screen usage, since you can compensate for hard disk with increased RAM and cache. But now this becomes the biggie.
There is no single answer, all laptops are multi part systems. But this allows you to work on the other two.
[note: I'm biased, I'm down for 500 shares of TMTA on IPO through the lead broker, and I don't own Intel or AMD]
--- Will in Seattle - What are you doing to fight the War?
I have an older Toshiba Satellite Pro at 120MHz. I have Slackware Linux and Windows 95 dual booting, and I use it for mobile CD playing, MP3 playing, and basic word processor/spreadsheet stuff for note taking, problem solving, etc. I'm not using the hell out of it for performance, doing stuff like 3d gaming, tons of source code compilation (though in Linux I do end up doing some), and complex modelling... It doesn't matter what the processor is the the greatest extent as how long I can keep using it.
I can see this Transmeta based laptop being something I would buy if I could afford it. It's faster than the mobile computer that I have now, it's got a good screen, it runs for many, many hours on just one battery, it won't give me 3rd degree burns to my lap.
How much processor power does it take to load a web browser or a telnet client or such? USE is more important than simple speed...
IBM had PL/1, with syntax worse than JOSS,
And everywhere the language went, it was a total loss...
Why does everyone worry so much about speed? Why not worry about "enough to get the job done"? I've got a PPro200 system on my desk, because I dont *need* anything faster. My wife's favorite system is a Toshiba Libretto 100CT - a Pentium 166MMX-based micronotebook. I mean, HOW MUCH cpu do you need for Netscape, SSH, and an occasional MP3? CPUs are getting like cars; people think they have to upgrade JUST BECAUSE a company has released a "bigger faster better" version - not because they actually NEED the increased speed or performance. Gamers will argue with me, of course, but I'd say the majority of computer users can get by with a Pentium 166-class machine for the majority of their daily work. The "must upgrade to faster CPU" thing is just propagated by Intel and the other CPU manufacturers to drum up business.
...is the one you run yourself, doing the things you do every day. Everything else is just hype.
--
Disinfect the GNU General Public Virus!
However, this world is still run by Microsoft and ceteris paribus Crusoe will never be able to break the unofficial Intel-MS marriage. Transmeta should gamble 100% on Linux notebooks and perhaps on embedded systems too.
Linux doesn't need to devore every few months an increase in processor power and can also focus on stability and technical neatness besides blunt power. Windoze may have a shorter time to market although one can discuss that, but it is also made fast-fast with major bugs waiting to burst (or to patch, depending which one happens first - usually the bug which forces to patch)
I think there is a market for a notebook which as primary system runs Linux and just in the second place Windoze 9x/NT/2000. That would be a lean mean portable Geek Machine!!! After all most of us work for the corps and must be able to run from time to time an Wintel app.
Transmeta, do you hear me?!?!? I am waiting for your open notebook platform, open both in hardware as well as in software of course!
The nice thing about Windows is: it does not just crash; it displays a nice little dialog box and let's you press 'OK'
So, what you are saying is that Crusoe concept is interesting? So is FX!32, Napster, Iridium... Being interesting does not make a product successful in a market place. What I hear about Crusoe is that it gives 10% battery increase for 30% drop in performance in an average notebook. I think I'll pass...
This whole business with emulation is very interesting, but I fail to see where this will fit into the current marketplace. When I need to run an x86 OS, I'll buy an x86 chip. When I need IA-64, I'll buy an IA-64 chip. The people who are buying these things for future compatabilty are the ones that baffle me. If the Crusoe can't even keep up with current IA-32 architectures, how is it going to keep up with Merced and its successors? Why doesn't Transmeta just focus their efforts into producing a native x86 chip? or a native IA-64 chip? Who the hell needs one chip to do both? Certainly not the average road warrior that runs Win98/WinNT/Linux. This is the demographic that the Crusoe is targeted at.
OK Hemos, it's "overwhelmed", not "overwhemled". One misspelled word in a three sentence burst wasn't so bad. However using "slower then native" instead of the correct "slower than native" was pathetic. Even though Slashdot will never have journalistic integrity, it could at least have grammatical integrity.
flame on...
There are three ways to lie: lies, damned lies and statistics. Benchmarks are, have been, and always will be about politics.
Every time I see a benchmark or other metric in the computer world, I have to constantly ask myself, "Is that Linux worse than NT article comparing a Linux box on a 386 vs NT on a 8 processor SMP Alpha?", excuse the exaggerating to make a point.
So what if the benchmarks suck. Is getting the highest numbers the aim of the game? If so, then the person who wins hands-down is going to be the guy with the 5000W power supply that has to be plugged into a 220V outlet, with a liquid-nitrogen cooled non-conducting fluid bath containing a 20 PowerPC SMP Beowulf cluster with attached tachyon cache. The thing'll weigh 250 lbs, be totally impractical for practical use and burst into flames within five hours, but by God, it gets some serious BogoMips before heat death.
A processor that runs really cool, with low power consumption and the capacity to adapt to whatever's thrown at it is pretty cool in my book.
And given that most people do email and write documents, really a 8088 would do. I think we should get over the whole comparing antler or genital size thing that men do. RRR RRR RRR Thog have 200 MhZ! Ug have 240 MhZ! GRAR! Ug get girl!
Sheesh.
--- Jump!! Fire!! Bullet time!! - Lego version of the Matrix
I am not Eric Raymond, but I have done it occasionally - more often, however, on a train, and more often compiling not kernels, but other systems (most likely my equational theorem prover E . And higher speeds also come in handy for LaTeX-ing large documents.
I generally agree that it makes sense to trade speed for longer battery life. However, given that I can work for about 5 hours on my new notebook (P3 600 with SpeedStep, 64 MB Memory under Linux, the current technological compromise is quite ok for me.
Stephan
The question that begs to be asked here is, "How many users NEED to search and replace a 250 page document? How many Excel users even know what a Black-Schols analysis is?" Honestly, that's like telling the average user that they need a Cray because it runs weather simulations so much faster than their Macintosh.
hmm... the processor isn't designed for running benchmarks... it is made for running real apps. and it's kind of slow. AND... according to the newest german c't it's still slow at the third run
"It's such a fine line between stupid and clever" -- David St. Hubbins, Spinal Tap
I must admit that I haven't looked at the Transmeta design very closely, but there is one thing that seems to have gotten relatively little attention yet...
What happens if the code morphing cache is too small for the workset used by the benchmark or application, and how likely is this going to be, given the 16 MB cache used, I believe, by the Sony device?
In other words, can it be expected that the entire translated code of your average application fits into the code morphing cache, so that optimizations from multiple independent runs can really take effect? After all, this is what most reviewers seem to assume.
Also, does paging in and out of code through the operating system's VM manager influence the effect of code morphing in any way?
Hear hear!
You mean you don't understand the marketing, or you don't understand the bragging rights? ;)
Getting serious: I don't compile kernels on an airplane, but I do serious scientific and engineering work on 'em -- often with a dozen or so heavy-hitting apps running at once -- and sometimes I even attempt as much CAD as I can manage, without having my tablet available.
Some people need the power: my notebook has to do most of what my desktop (a dual PIII 800) does, while I'm on the road. So I have a Dell Inspiron with a PIII 650 (SpeedStep is usually OFF!), 320MB RAM and a 1400x1050 screen... and it's barely enough to get me by.
All that said, I agree with you that 90% of the notebook users don't need more than a 300 MHz processor for what they do. At least, not this year...
---
---
Politics is about making compromises. Religion isn't. --Michael Horton
---
---
Politics is about making compromises. Religion isn't. --Michael Horton
if a copy of Linux was ported to transmeta's native CPU (Very Long Instruction) would the performance be better than under "morphing"?
AFAICT the Crusoe is far more interesting for its design and the possibilities for emulation that for its raw processing power. If I wanted to have a fast IA32 machine I'd go buy one. But if I wanted a processor that had the potential possibilty of letting me design my very own processor to simulate atop it, and at near-hardware speeds, then I'd buy a Crusoe. Battery life is nice too, but I'm much more swayed by the possibility that I could even conceivably use the processor as a nice emulation engine for some other existing architecture, or one of my own design.
I also own an Inspiron (7500, PIII 500 Mhz, 256 MB of RAM), but this is used as desktop due to small size and it's cheaper that most of medium notebooks in the market. For the airplane, I still use my old Toshiba (P-MMX, 200 MHz) with a new battery, which gives me 5 to 6 hours without any problem.
I am convinced that Intel can produce new processor with the same capacity as old Pentiums and Celerons with a very low battery consumption, the problem is that the braindamaged "market" is eager for faster processor...
--ricardo
sgis ddo ekil t'nod i
And once more:
This processor is any processor that morphing instructions can be written for.
How long till the ultra low power solid state hdds come into being?
And wouldnt you rather use a headset display such as I-Glasses then a large (anyone can see) display when you are working in your resume? or looking at your Natalie Portman jpeg collection ?Try not to think of tomorrows computer systems by todays standards
Ok, so how does that relate to real world computing then? How often does an application do the same thing over and over again?
Yes, this is a good thing for the Crusoe, but just how many people sit there and run their favorite program twice in a row with the exact same data both times? Seems kinda redundant.
What am I missing here?
Personally, for cost-go x86, for performace-go alpha, for battery life/heat-go cruose or G4.
However, I see good things in the cruose's future. A friend has a PCI board with a Celeron 350 or so which plugs into the PCI bus, and takes over, perhaps curoses could do that because of power (eg cruose might run where a P3 or Athlon doesn't). (Yes, I know that is an 'old' way of doing things, but if they had a card like that, how may of us have old P75s and the like sitting around?)
Looks pretty clear to me. Crusoe, while being fast enough for the average joe, isn't as fast as folks had hoped. But because Torvalds works for Transmeta, every last Slashdotter is hoping and praying for Crusoe to come out on top. Linus working on something that isn't successfull?! Slashdot won't let that happen, or at least let its readers think that.
I would only trust benchmarks wcreated by an unbiased third party. Would you believe MS Benchmark 2000 when it showed Win2K was faster than everything else.
134340: I am not a number. I am a free planet!
Dims screen to save power every 5 seconds.
Spins down HD every 10-15 seconds of no disk activity.
Runs much slower on battery, in low power mode.
Is basically dead weight and bulk when it's dead.
I'd like to try something that runs moderately faster with considerably more battery life. Extra battery packs are just plain annoying and expensive.
--
A feeling of having made the same mistake before: Deja Foobar
I still don't understand this rush for ultra faster processors in notebooks, while more than 90% of the people use them only for text processing and power point presentations. I've never seen a hacker compiling a kernel in a plane. Humm... may be Eric Raymond?
--ricardo
sgis ddo ekil t'nod i
It wasn't meant to be a troll, I apologize. I honestly wondered how much of an effect this had. You talk about actions repeating, and I thought of stuff like that too. But what's the 'memory' on the chip? How far back does it remember to optomize?
I know that programs do the same things over and over and over, but what I wanted to ask was if this actually made a large impact in relation to the Crusoe itself.
The Register also has an interesting editorial from Wednesday about Transmeta's benchmarks.
--
Have fun: Join D.N.A. (National Dyslexics Association)
Similarly, I suspect (because it seems obvious) that the processor keeps track of how often translated code is used so it can discard the less-used translations when it starts to run out of space. It's like virtual memory, but instead of getting paged out to disk the less-needed translations are discarded since they can be redone if they're needed again. It sounds bad to repeat work, but the impact of re-translating is likely to be fairly small, because the code that needs it doesn't get called as often as the code that's kept.
One other thing that might be interesting is what sort of changes might be made to the code morphing software to improve the optimization. Perhaps it doesn't do a good least-recently-used discard of optimized code, but a processor firmware update adds that capability. What kind of impact will that have on system speed? I don't know if updates like that would be free, but would you pay $20 for a 10-15% improvement processor speed on your system?
It's interesting that Transmeta has four areas of potential improvement where Intel and AMD have only three. The first is improving processor speed, the second is improving the cache size and/or speed, the third is architecture improvements, and the third is better code morphing software (which is a form of architecture improvement). Intel has its BIOS update to let it address fixes in the same way that Transmeta could, but that's not going to make a difference in the overall performance of a processor.
I suspect that internally Transmeta is continuing to test processors to find out things like how much cache memory is needed to keep frequently-used translations from being discarded. I'm sure we'll never publicly see those, but eventually it'll probably be clear from the point at which Transmeta stops increasing the amount - when the amount of local cache stops growing it means they've either hit the point where most of what gets discarded wasn't going to be used again anyway or they've hit the point where indexing into the cache becomes a problem.
-- fencepost
fencepost
just a little off
Something I see over and over again is that emulation has to be slower than the original, yadda yadda blah blah.
m o-1.html for more information.
:-)
This is not true.
There are a lot of advantages to being able to modify the code at run time. Most of them involve the optimiser being able to check its initial guess about how to optimise the relevant code. Think of it as a sort of runtime profiling followed by regeneration of code segments.
One example is HP's Dynamo, a PA-RISC CPU emulator that runs code faster on a PA-RISC than the same code running natively. Try:
http://arstechnica.com/reviews/1q00/dynamo/dyna
Another example would be Dr Michael Franz's work on Dynamic Optimisation. Try: http://www.ics.uci.edu/~franz/DynamicOpt.html for more information.
Right. Now stop making assumptions.
---
Trying to compete with the price/performance ratio of the x86 architecture is a lost cause. The crusoe architecture might have a bright future if it could run Java byte code instead of x86 code.
I'm sitting here using an old pentium pro 200 with 256 megs of ram. It does everything I basicly need it to do. Sure it won't do four VMWare sessions, but it can run netscape 4.75, a dozen eterms and a ssh session for each term, dia, gimp, xmms, apache-ssl, and X11 all at the same time. That's all I really need. I mean really, do I care that this thing can only do quake at 200 fps? No.
/. , reading Nealz Newz, and IRC'ing becuase grep can't find 'life' anywhere in /proc/sys/marnuke/. What I do need in a laptop is something that can last for hours even days without losing power, not a damn supercomputer.
I do the same thing day in day out on this machine. Every now then, I might compile a kernel or some other package on this machine, but I have a cluster of dual 600's to build packages on. Of course I wouldn't do compiles on my laptop, and for the love of god not on this machine. Why would I?
When I'm not on my desktop, what will I be doing? Most writing a letter, checking email, ssh'ing into a server, posting on
If I have a choice between four days uptime on a laptop and being able to compile a kernel in a minute, I'll take the four days uptime.
MarNuke
I just hope this is being fair, and not whining. After all, emulation is slower, but Linus does work for these guys...
-----------------------------
1,2,3,4 Moderation has to Go!
By allowing the instruction set presented to the world to be set by software the chip has a lot of advantages in terms of customizability and turnaround time for fixing bugs. Software is also more compact that the corresponding hardware so the chip is overall less complex and presumably more reliable, once debugged.
There is another whole category of applications that may become possible thru use of the native instructions, or better, by custom design of a logical instruction set for a given use.
Four revisions ago, the transmeta chip took half an hour to boot Windows. Consider where they'll be four revisions from now before discounting them on speed alone.
This information and more can be found in a good Technology Review article I found on Linux Today here.
Okay fine, let's run the benchmark program 3 times in a row. Are we getting similar results?
Besides, benchmarks should be run at least 3 times and the result should be the average of the runs, assuming there's no large discrepancy.
While notebook computers may be the initial application for the Transmeta chips, there should eventually be a sizable market for x86 compatible chips that run cool enough that you can embed them in any of a variety of consumer devices where a cooling fan is a big drawback. As a simple example, an advanced set-top box sits next to the television and the noise from a fan interfers with the audio from the television. As such devices run more software that is ported from the desktop world, there will be advantages to having the x86 instruction set available.
For some of the situations that I look at, I especially like the previous suggestion about putting four processors (and implicitly, the needed SMP chipset support) on a single die.
They might end up with a CPU that burns 4 watts, is a little bit warm, and runs at about 280% the equivalent Pentium III.
Maybe they could put 4 cores onto a single chip, but this would certainly reduce yields (and thus increase costs). So you'd end up paying somewhat more than 4 times as much for this multi-core chip than for a single chip.
And this system wouldn't give you 280% of the performance of a PIII of the same clock: The things most software does isn't perfectly parallelizable, so there's a diminishing return that comes with each additional CPU.
All I use my latpop for is browsing code and typing. I am currently using a 266 AMD, it works find but the battery life is horriable. I have a 900 AMD for compiling my work on, so all they really need to do is, for most users, give decent speed at running word, so even if their 600 is only comparable to a intel 400 or even a 300 it will still do what most consumers use laptops for. Besides it sucks flying from Californa to North Carolina and have to cary 3 extra batteries just to use my laptop while I am on the plane.
.sigs suck, thus nothing here.
Not that I think that's bad, or that their chip isn't useful for the purposes they're advertising. They've chosen their strenghts to market on, and you can't tell me that a chip performing so much in software _isn't_ being constantly developed. I really think they have the benefit of being able to try lots of tricks with hardware and software for a quicker turnaround, should they choose to take advantage of the "virtual" nature of their chip.
It's not a given at all... That was the whole point. The point was that you can fit more run-time optimization smarts into emulation sw than into microcode or whatever. I think they exactly that over at HP - an selfhosted emulated system running faster than the native one or whatever. Correct me if I'm wrong
Despite the fact that I'm glad /. is finally back up today, I'm saddened to see this as the first story. This is news? All along they've been saying that Crusoe would give incredible battery life at the expense of speed.
what should be really interesting to see is how the runs when linux is coded natively for the chip, which should = less/0 code morphing, this I see is the main advantage of having Linus(tm) :) on the team, what's gonna happen to Linux in the future.. the future is HURD
I agree with you, but most of the laptops are used n a corporative environment, and don't use their batteries. My company is giving a laptop to every employee, in order to sutitute every PC in the office. They want the performance of a powerful PC in a laptop, they don't want 10 hours of life in a battery
But what REALLY isn't being taken into account is that the benchmark suites run everything ONCE... but the Transmeta chip gets faster with subsequent runs...
Check this article on Wired for more info.
According to something else I read (sorry no link on this one) performance improves an average of 30% on subsequent runs... so take the benchmarks with a grain of salt.
.technomancer
.technomancer
They're just questioning the usefulness of a processor that adjusts it's speed. "Is the decrease in speed worth the battery life?" It's the same mindless drivel as usual. If you can run [insert office app here] for that much longer when you need it, should the speed make a difference? I doubt the speed will drop to a rate where you type faster than the screen refresh can keep up. Just a thought...
Crusoe simply doesn't perform as well. In fact, it performs much more poorly. It does consume less power, but doesn't lower the power consumption of the hard drive, memory and display, resulting in less bang for the buck without a huge gain in battery life.
That's what the press is saying, and it's true. The Crusoe has an interesting design, and will clearly be a useful chip, but it is not competative against Intel and AMD. Perhaps when they get the speed up and let it run code frm several ISAs at the same time (say, PowerPC and x86 at once), it'll be cool.
I think Transmeta gets a lot of hype-factor from employing Linus.
________________________________________
Napster-to-go says "Fill and refill your compatible MP3 player", which is a lie. It's not MP3. It's WMA with DRM.
Ah, this poster is obviously a gamer or of similar ilk who doesn't care about the complexities involved in design, research, and experimentation. At least he'll have a good career in management.
People with this sort of attitude towards computing systems, ie "I want it fast and 31337 and hyped to the max!" scare me. I only hope that they grow out of teenagerdom and actually learn something about what sort of work actually goes into producing computer hardware and software.
This is the same sort of response from an uneducated art or music critic: "I don't like it because it's not my favorite {style,color,sound,*}." This sort of attitude is usually found amongst the consumer society, which is why automotive manufacturers can get away with producing such horrible products but still sell them to humans who foam at the mouth when the competitor to their favorite brand is mentioned.
The above is much too wordy. I think I just meant to say "*plonk*".
for a portable.
No one complains when their Palm won't run Quake Arena. (well they do, but they shouldn't...or should they???) This chip is not meant to be a full fleged asskicker. It's not competing with the Athlon or PIII. It's in a market all by itself.
If this chip is not noticably slower when I surf the web or read email or such from my laptop, and does noticably extend battery life, then that's enough for me.
Which one would you pick if you had the choise of taking a Palm Pilot with the 33MHz Dragonball in it or the same machine with a 500MHz Crusoe in it for the same CPU power consumption? I'd bet that Crusoe will out-perform any PDA cpu anytime with the same power consuption.
The products Crusoe is really ment for are entirely different from traditional laptops. The Crusoe chip is technically near ideal for WebPAD and PDA-like machines that do not need excessive amounts of computing power, but more like reasonable amounts of it with as low power consumption as possible.
The equipment Crusoe is ideal for is, say a WebPad which has enough of CPU power to play MP3's & MPEG2 (DVD) and perhaps runs a web browser & an office suite at the same time and is using solid state memory for storage, it's housed in an A4 size enclosure with an 13" 16:9 TFT screen and accesses the Net via WLAN.
Here we can see the scenario where the power saving features of Crusoe (and the rest of the design naturally has to support similar unified power saving features) can make a big difference.
In equipment like this the difference of operating 3 hours / battery or operating 8 hour / battery makes all the difference. I'd kill if my Toshiba Tecra laptop would stay alive on one battery more than 2h45m. Say 8-12h ago would be sweet indeed.
Raymond
There is a fundamental difference between the HP's dynamo and a "normal" emulator: it emulates the SAME processor as the initial target.
I think that it is a *BIG* difference which may help a lot in this case!
Transmeta's CPU was conceived to emulate first a 80x86 CPU, this can avoid some ISA "mismatch" where every gain which occurs thanks to dynamic optimisation is lost by the necessary translation from one ISA to another..
and what of the temperature difference ...
overtime a chip from intel or AMD will slow
down because of the extreme temperature..
while crusoe whizz'ess along..
"The world is coming to an end. Please log orff."
So, according to these benchmarks, at the same core frequency, a Crusoe is 30% slower than a Pentium III, right? Well, I think it's pretty good. I was expecting this code morphing stuff and this very low consumption architecture to hurt performance by a factor 2 or 3, and things are obviously much better than that. I'm especially pleased with the fact that the FPU is not ridiculous (which is definitely not the case for StrongARM and most ARM CPUs).
As far as I'm concerned, the speed of a Crusoe is waaayyyy enough for my needs, especially on a laptop, and I think it is true for most users. Besides, apparently a 1Ghz version is expected to ship by the end of the year, so there really is very few to complain about.
Thing is, it's not the data that gets optimized. It's the way it runs the code that gets optimized.
In the real world, executible code tends to stay the same (excepting viruses and self-modifying code), and its major bottlenecks don't change much.
SPEC works for everyone. If Crusoe doesn't perform well on SPEC, why do we need to design a new SPEC?
And yeah, it runs longer on a laptop's batteries. So do Speedstep chips. I just received a new laptop (a screamer from Dell) that runs beautifully in both Windows and Linux. It uses a Speedstep chip, and the battery power I save is noticable (I've run the battery sometimes up to 4.5 hours).
- I don't care if they globalize against free speech. All my best free thoughts are done in my head.
A low-power CPU doesn't give you much more battery life at all, in a full-featured notebook. That's really all that needs to be said to put the final nail in Transmeta's coffin.
What I would like to see is a company like IBM design a notebook completely from scratch, using nothing but custom-designed components designed to use as little juice as possible. I mean seriously, shouldn't we be able to get at least 5 or 6 hours of HARD use out of a notebook before it needs to be recharged? 3 hours, which seems to be what most notebooks manage these days, just isn't enough.
Of course, since I read, clicked, and then trolled this article, I suppose that I have absolutely no reason to complain. Not that that stops anyone else, so I don't feel too much like a hypocrite for doing it.
How can we continue to believe in a just universe and freedom to eat crackers if we have no ale?
Why is it that I keep seeing this same story re-hashed over, and over, and over again.. this has to be at least the third time the *same* story has been addressed here. By the way you guys hear anything about the playstation 2 ;)
JOhn
Campaign for Liberty
Power saving on things like the CPU fan that are no longer needed help, too -- besides making for a sexier machine.
I would use a crusoe laptop/webpad for serial consoles on servers, browsing the net with some stripped down mozilla, or otherwise just using a terminal for daily tasks. I don't need a huge PIII system, cuz I won't use the thing for games, photo editting, or ray tracing. That's what my desktop is for. But I do want either a huge 15"+ XGA TFT display or a nice sized webpad with no fan that runs cool enough to hold comfortably.
I would also hope that transmeta is not going to be another stuck up proprietary IP company like Intel, MS, Sun, Apple and the like. If they want to be a proprietary IP company, that's fine. Just don't use Linus as some marketting ploy. That's like a slap in the face to anyone who supports the ideology of the GPL. I know I won't be buying a Crusoe until Transmeta releases some information about their code morphing technology.
But I must admit they created a system that will live up to its intentions. To be an efficient, fairly fast and usable webpad/linux PC. As long as they come with wireless net connections or at the very least a PCMCIA port, and cost me no more than $1000, they will help change the way I use computers to store my information. Palm Pilots just don't cut it with their 8MB of ram. Crusoe is just a step up, not the portable server you were expecting. And NO, they don't perform like a P3 500 or 700, they're different. Wish they'd post REAL benchmark numbers on their site and stop causing all this bad PR by trying to live up to the high end systems.
You want performance, go with a Pentium. You want something usable for several hours at home while you are watching the TV, that uses your home net to access all your personal information and the web, that is fast enough to get your work done, potentially upgradable to support other OSs, light weight and cool running, get a Crusoe... but only after Transmeta releases their IP on code morphing. We can at least hold out for that.
As I understand it, this chip:
-Takes x86 instructions and decodes them to its own ISA and executes it.
-Runs very cool.
-Is a pretty simple chip.
-Runs at about 70% of the equiv. Pentium III.
What's to stop Transmeta from tweaking the software a bit and slapping 4 of these cores on a single chip. Heat? Gate density?
They might end up with a CPU that burns 4 watts, is a little bit warm, and runs at about 280% the equivalent Pentium III.
OK: Sounds reasonable. What does it take to dump this "special portion of RAM" to the HD (possibly in a separate filesystem, especially if your emulator is running Win), which the emulator would associate with the i86 (or whatever) binary file? The emulator would probably need the date/time stamp of the binary to make sure it hadn't been modified since the last run... better do a checksum, too, just in case... Certain anti-viral properties suggest themselves here. Then, when the program is executed again, the i86 binary need not be loaded, and the translation doesn't need to be done at all.
Further on-the-fly optimizations for loops, etc. could be performed. This could also be handed off to a dedicated deep optimizer that runs the machine at full speed while plugged in to the wall, as the owner sleeps. Set up the benchmark software this way, and allow the optimizer to run, then see how it does.
A somewhat unrelated thought is that a machine running an emulator can do debugging without requiring any special support in the i86 "object code" itself. A compiler can build a separate map file for the emulator's use in setting breakpoints, and as an added bonus, use the same info to 'optimize the optimization'. Once again, the map file need not be visible to the OS running on the virtual machine, just so that the real chip can find it and make good use of it.
[100% ISO 646 Compliant]
SVM, ERGO MONSTRO.
If nothing else, the Crusoe will put pressure on AMD and Intel to move in the direction I value most...low power consumption, heat generation and weight. Intel's already coming out with low-power PIIIs, possibly spurred on by the Crusoe. Linus' involvement matters not a whit to me, although I'm glad he's got a nice job.
I remember doing similar benchmarks when I worked at Sun. We were introducing some PC emulation, and normal benchmarks didn't give the whole story. So, we created our own. One example was to do a search and replace on a 250 page Word document. Another was running a Black-Schols analysis in Excel. Do any readers have a new Sony Picture Book they can spare?
****
"I'd never want to join a club that would have me as a member" - G. Marx
And yes, I typo'd in the subject.
Too bad Intel has low power faster chips already that dont emulate code. How about the xscale, or the low power P3 due to come out soon? I dont see why they are getting so much hype for a slower and more power demanding chip.
As we've heard the benchmarks haven't overwhemled people - but are we measuring things the wrong way?
If Intel loses on benchmarks - Benchmarks are the One True Way(tm) to measure computer performance and they were collected properly in this measurement
If Intel wins by a little on the benchmarks - the procedures were flawed; the results were probably funded by Intel.
If Intel slaughters The Company Which Linus Works For(tm) on benchmarks - then, as Hemos says, "we are measuring things the wrong way"; in other words all benchmarks are inherently flawed and incorrect, and no useful information can be garnered from them.
The Register did an article about all these misleading benchmarks being done on Transmeta systems:
....
Transmeta speed debate - damned lies and benchmarks?
Transmeta notebooks have been available for a week now, and although it's been a fascinating week for benchmark mavens, the casual punter hasn't been much enlightened.
Why? Well three serious IT publications have run benchmarks for Sony's new Crusoe laptop in the past week, published them... and disavowed the results immediately.
Sorta, kinda.
The problem is that Transmeta's new chip confounds the traditional one-pass benchmarks that their propeller head authors have spent years perfecting. Since the Transmeta architecture allows for dynamic and smart execution of code in software, the first pass of any test sample is guaranteed to be if not crap, then not quite optimal. With the instruction set and performance heuristics in software - a compromise that allows the chip itself to throttle back its frequency and voltage - judging a Crusoe on first-pass benchmarks is like judging a marrow competition based on the seeds. It isn't going to give you the full picture.
and it got modded up!
you're welcome. i understood that you were referencing the earlier 'm$ cracked' article orthagonally to the transmeta corporation, which is, of course, part 0wnZ3d by Paul Allen. A beautiful and deep joke, obviously wasted on the rest of the slashbot moderation wrecking krew.
We can't forget that the goal of Transmeta wasn't to create amazingly high speed processors compared to today's standard, but to make comparable (and slightly better) processors for mobile Internet-connected computers (or web slates, laptops, etc.). It's ashame that people are "dissing" the company, so to speak, because the benchmarks (very old-school traditional ones, mind you) don't amaze them. Give Transmeta time, and I think you'll be incredibly pleased at what they produce. After all, working with code-morphing technology is just a bit challenging ;-D
______________________________
Eric Krout
If you celebrate Xmas, befriend me (538
If the damned thing can increase my battery life by a lot while running programs at even P2 or even meer Pentium speeds then it's doing its job.
It lets you go farther with your laptop!
First, I understand the argument that benchmarks aren't necessarily relevant. I can only type so fast in Microsoft Word, so provided it runs at a decent speed, I'm happy. For the user who just wants a cheap laptop, these might provide adequated performance coupled with a long battery life; that's just great.
But I'm a nerd, and I want to see the benchmarks. First, I remember reading *another* article about this from Slashdot, where they claimed that the Crusoe really shined at memory bandwidth, perhaps because it had so much integrated on the chip. That's really important nowadays, and people tend to ignore it.
Also, if anyone has one of these, could you please post results for the BYTEMarks? The Unix port includes numbers for memory bandwidth as well as integer and floating point, and the benchmark statistically repeats the tests until it gets a steady-state result for performance, which means that the Crusoe would get to optimize them after the first few passes through, and therefore this test should be more than fair to the Crusoe.
Also, the Crusoe should be compared against an equivalent (specced or priced) *laptop* configuration; remember that.
---
pb Reply or e-mail; don't vaguely moderate.
pb Reply or e-mail; don't vaguely moderate.
There is an article on toms' hardware on Transmeta. The point is that 'yes, all current benchmarks are not realistic in a sense they don't reflect the actual usage patterns. They also mention they would be working on methodology to provide more realistic benchmarking setup.
Slower THAN!
Higher THAN!
Faster THAN!
What is up with the misuse of that phrase?
davejenkins.com |
Folks, I know I don't need to ask to be corrected if I am wrong here, but there seem to be some important issues being missed where Crusoe is concerned. Here goes.
One. The real issue that I have heard grumbling about centers less on speed differences as they do on the failure of the chip in laptops to really extend battery life as much as was anticipated.
I have heard gains are only 1.5x typical consumption. That is a long way from the 2.5 to 3x times that had been touted earlier.
Two. There are two lines of the chip, one explicitly aimed at "laptopish" devices, and the other for lighter weight appliances. Therefore it is perfectly appropriate to expect the chip to do what it is supposed to do in a laptop. It isn't being scaled up beyond its specs as some comments here seem to suggest in defense of its performance.
Three. The key issue in performance is due to what I understand to be a failure to deal appropriately with changes of state.
I was told that the algorithms have trobule dealing with interrupts. As a result, instead of smart updating and caching, there is a presumption that all of state has changed. This is apparently where the performance hits occur I have heard this is strictly a Windows issue. But for the laptop market as a whole, that is clearly a big issue
At any rate, the real problem no matter what the cause is that you effectively end up with performance that is about as good as you would get in ultra-conservative mode on current systems. Not awful, but not great either.
And there I think is the real key. If an end user feels delays to be unacceptable or performance net net is about the same, they just won't find a real world difference in their interaction with the device. Coupled with less than expected gains in battery life, and you basically end up in pretty much a stalemate with Intel products, and possibly worse off against the new low-power chips they are rolling out.
Just seems like the hype wave got ahead of deliverable real world performance.
You may have a point, but the benchmarks simply are not taking into account battery life -- which is very important to the non-quake crowd. What they need is someway to benchmark the performance over the duration of battery life. Start the tests going with a full charge and see not only how fast, but how long it runs for. Perhaps someone could workout how many D.net blocks or S@h WUs can be completed on a single charge with some of these transmeta portables...
Usability, however, has never really featured as a big item on Slashdot. For example, in coverage of Gnome and KDE, most of the articles seem to be more about the politics, personalities involved, and so on than the actual usability of the environments.
The overall usefulness of computer products is difficult to quantify, and is one of the reasons why companies like Apple have had a tough time. Issues of user-friendliness are thought of as the whinings of computer-illiterates.
Interestingly enough, now we arrive at the Transmeta speed reviews, and it appears that many of the same people who decry anything "pretty" are offering up excuses as to why their favorite chip doesn't run as fast as the others. "But it's more *useful*" they complain.
Hmm... usefulness rather than pure numbers as a benchmark. What an interesting concept.
Read the EFF's Fair Use FAQ
They make processors. You can't fault them because displays and hard drives eat batteries. They're taking a first step in CPU energy-efficiency, not laptop energy-efficiency. If display and hard drive manufacturers innovate to make their products more efficient, you would have the full package and the Crusoe could be integrated to its full advantage. But until that happens, the Crusoe can only be a first step. A good first step.
I think what's really important about the crusoe technology is not the improved battery life but the fact that code-morphing is so cool.
By ridding themselves of the hardware baggage of CISC designs, I expect that Transmeta chips will ultimately be faster and cheaper than their CISC contemporaries - as well as cooler and less power hungry.
amen.
Instead of having Transmeta go off and build a new style of (no doubt self-serving) benchmark practice, how about emulating the SPEC benchmarks with their "train" and "ref" benchmark runs.
For those of you who aren't SPECheads, "train" benchmark runs are used to profile and optimize your code for the final, measured "ref" benchmark set (although I think this only occurs for reporting of peak numbers, as there's a lot of opportunity for benchmark-specific compiler skullduggery here).
This is a well-known practice. It models well the fact that while there is a fair bit of common coverage between different runs of a program, it's not perfect. It affords opportunities for the "other camp" (IA-32) to do their own optimizations.
Granted, there's a difference between standard profile-driven optimization and on-the-fly optimization, but there's a lot less difference than you'd think in many applications.
A bit off my topic, here, but by the way: locality is typically much worse in real applications than it is in SPEC-style benchmarks, even the more realistic ones. Don't be too optimistic about all of that 90%/10% stuff you learned in undergrad CS; sometimes it applies, but a lot of bigger apps are not nearly as tidy as xlisp and m88ksim - or, god help us, the long-departed eqntott ("I think I'm an 8-line microbenchmark").
I see the point made all over this discussion that a Crusoe laptol should be compared to a similar intel-based one, and that the decrease in computing power is a sacrifice for the increase in battery life. IMHO the natural thing would be to have a comparison of two comparable laptops, one Crusoe and one intel, with performance AND power consumptions figures. And it would be quite visible wether the Crusoe compromise is worth it or not.
After all, you don't expect your laptop to have a 19" screen? Four speaker Dolby (tm) surround sound? In a laptop, you want it to be light, easy to carry, fast *enough* and to have *enough* battery life for your needs.
As has been said in previous articles, many people would be THRILLED with a Pentium 200 based laptop, if only it would have a reasonable battery life. The Crusoe chips appear to give FAR better performance than that, with lower power usage ( gotta love intigrated memory controllers! ).
Who REALLY needs a P-III Coppermine 800 in their laptop? And if you do, don't bitch about battery life, you're getting other benefits.
The fact that no one understands you doesn't mean you're an artist.
Ok. I'm sorry, but I think it is almost humorous to watch /. react when positive things come from 'evil' places or negative results come from 'good' places. I'm really not trying to troll here, but I'm beginning to think of silicon valley business as /. sports. Root for the home team!
Transmeta has posted poor results, so now we want to change the way we test, etc. They have sacrificed speed for battery life. I think that is acceptable, I want to see longer lasting laptops (even at the expense of speed). I wonder if Linus didn't work for Transmeta if it would get such a defensive response.
-Moondog