Transmeta Introduces The Efficeon

brentlaminack writes "Information Week and others are reporting on Transmeta's new Efficeon chip. 1.1 GHz, 7 Watts, 1MB cache, 130 nanometer technology. A marked improvement over their previous generation. Let's hope they can capitalize on this before Intel starts filling the same niche. Looks like a nice product, Linus and Co." Update: 10/15 00:22 GMT by T : woobieman29 writes "Looks like this is a good day for high-efficiency processors. Hot on the heels of Transmetas announcement of the Efficeon, VIA Technologies has announced the release of it's latest low-power processor, the NanoBGA EDEN-N. Capable of running at 533MHZ (4 watts), 800MHZ (6 watts), and 1GHZ (7 watts) this appears to be a very good fit for Thin Client and other embedded devices. One really interesting feature is the on-chip Padlock security suite incorporating AES encryption."

Comments

Let's change the name

[l810c]

Efficeon sounds kinda cheesy.

Here's some alternatives :

Ableon
Activeon
Adapteon
Apteon
Clevereon
Defteon
Efficaceon
Handeon
Potenteon
Powerfuleon
Shrewdeon
Tougheon

Re:Let's change the name

[MadCow42]

"Efficeon" (or "E" Model) came after the notoriously crappy "D" model: the Defficaceon.

Re:Let's change the name

[Tumbleweed]

Noone likes to talk about the first model, the Afisheon.

It smelled funny.

Re:Let's change the name

[LesPaul75]

I've heard guys who work at TMTA call it "defficeon." I think it's actually a really nice chip... it's just typical engineers' contempt for their marketing department. I gotta agree -- "Efficeon" is pretty dumb. Almost as bad as "Pentium 4 EXTREME."

Re:Let's change the name

[Anonymous Coward]

I use the newest chip for troll systems....
PuttinUon

Re:Let's change the name

[ExMember]

Originally it was called the C-on which is kind of an inside jab at Microsoft's supposedly multiplatform C#. Sometimes when it didn't work it was called the C-off, but C-on was the name that stuck until late in the stages of debugging the fab process. At that point every reference the engineers made to it was as the F-in' C-on. Marketing, in a spell of creativitylessness, took that and ran. And thus the Efficeon was born.

Re:Let's change the name

[Sonicated]

With Slashdot claiming earlier today that "The FSF [are] Linux's Hit Men" do you _really_ think they are going to risk naming it the the Apteon?

Re:Let's change the name

[Feztaa]

How about hardeon?

Re:Let's change the name

[Feztaa]

How about hardeon?

"Yeah, I've got Woody running on my new Hardeon!"

Re:Let's change the name

[Eccles]

"Efficeon" is pretty dumb.

A unique trademarked word is good for googling, so that's a good reason to make up a word. They picked a name with a link to efficient, since that is the strength of the chip. It's similar in concept to Acura, with its link to "accurate."

Re:Let's change the name

[salimma]

"Ableon", OTOH, sounds like Ablution. Beware of the censors !

Re:Let's change the name

[Anonymous Coward]

What? No Decepticon?

Re:Let's change the name

[jo42]

Efficeon sounds like something that would come out yer butt...

Re:Let's change the name

[whereiswaldo]

Endowedeon?

Only from the Mind of Dilbert ...

[CrankyFool]

Efficeon?

Wow, I thought Pentium was a bad name

[Anonymous Coward]

They've really outdone themselves with this one. Good job guys.

Employment for Linus

[Anonymous Coward]

Linus doesn't work for Transmeta anymore.

can't resist

[Pompatus]

But does it run ..... Nevermind.

fp, yo

[Anonymous Coward]

This first post is done using Transmeta processors, the world's fastest and kinkiest as everyone knows.

Why does Slashdot keep covering Transmeta? Since Linus doesn't work there anymore, isn't it just one of the Intel wannabe's, like VIA and Zilog and Microchip and stuff?

Re:fp, yo

[Anonymous Coward]

Yes, it is. It ranks dead last in 32-bit CPU production, but third in hype. http://www.eweek.com/article2/0,4149,1338456,00.as p

Re:fp, yo

[toddestan]

I would think that a 1.1ghz CPU that uses 7W of power is newsworthy. It seems that VIA is the only x86 CPU maker out there that seems to make these kind of processors.

I have a hunch that in a few years people are going to become uninterested in faster computers to do things like office apps, email, and web browsing and instead will demand smaller, cheaper, and quieter computers. At that time Intel and AMD may find that many people are not interested in the Pentium VI Supa-Extreme Edition or the Athlon 128XXP++++, but rather in VIA's powerful enough but energy efficient chips.

Re:fp, yo

[jir0]

Actually, Transmeta does deliver pretty okay products. Yes, I know, you'll say with processors, Intel pretty much beat them up after releasing the low-power Centrinos. But they were actually one of the first guys to go for in power-efficient processing. They did mobile computing a lot of good. The very first laptops that really caught my attention were the low-power Crusoe-run Twinhead notebooks. 6 hours on 300 MHz was pretty amazing a few years back. They still do nice jobs now and then (see a nice article here). I still drool over the Fujitsu Lifebook P-series, most of which run on the Crusoe. I for one go for portability as the first priority for laptops. (I can always just ssh into my main box if I needed anything other than emacs.)

They also have a number of impressive cluster servers. Again, having low power consumption is making their high-performance servers look good, even among today's Blades.

Any innovations are still welcome. At the very least, it's nice to know there are projects who keep Intel working on new ways to be better. :)

Now about that silly name...

Re:fp, yo

[Nynaeve]

My friend in the cube next to me has a Fujitsu Lifebook with the Transmeta chip. It has a wide-screen and lasts 9 hours with the extended battery. I do a lot of drooling myself, but at $1200 it's beyond my reach. :(

BTW, he tells me that ACPI doesn't work on it even with the latest linux 2.6-test kernel. You'd think that since Linus worked there they'd take a little extra effort to make sure its #1 feature -- low power consumption -- works properly in Linux!

Re:fp, yo

[Hoser McMoose]

Intel's ULV Pentium M chip runs at up to 1.0GHz and consumes 7W maximum. Their ULV Mobile Celeron also consumes only 7W when running at up to 800MHz. What's more, I haven't seen any info that says if the '7W' that Transmeta is quoting is it's maximum power consumption, a thermal design spec, or it's "typical" power consumption. Transmeta has a tendency to only talk about typical power consumption, while the 7W numbers listed above for Intel are the chips Thermal Design Power (TDP, basically the maximum you'll ever see with the exception of code specifically designed to consume maximum power). TDP is ALWAYS at least a few watts higher than 'typical' power consumption.

AMD may also have a low-voltage AthlonXP-M that is in the same power consumption range, but unfortunately AMD does a piss-poor job of documenting their mobile processors (read: there is absolutely no documentation publicly available). As you mentioned, VIA is also producing chips in the same power consumption range.

Long story short: Transmeta is going to have to either deliver on performance (like Intel does) or on price (like VIA). Right now they are talking about selling the chips for $100 a pop, which is quite a bit more than what VIA sells for. They are also talking about only a 50-80% improvement in performance over the Crusoe, which isn't going to do much for them in the way of performance. At 1.1GHz, they might be competitive with the 800MHz ULV Celeron, but I'm not holding my breath. The Crusoe had terrible performance.

Re:fp, yo

[Anonymous Coward]

Especially the quieter! I have a laptop as my only computer and I love the fact that it is very quiet especially compared to our last PC. Once you have a quiet computer i don't think you will go back to a noisy one, ever!

Re:fp, yo

[evilviper]

I would think that a 1.1ghz CPU that uses 7W of power is newsworthy.

ALERT! ALERT! MHz Myth in full force!!!

The fact is, this isn't news worth. Via C3 processors perform like AMD/Intel chips of half the speed.

This 7W 1.1GHz processor is about as impressive as a 7W, (at most) 550MHz AMD/Intel processor. Not all that impressive.

If anyone wants a low-heat, low-power system, there is one good way to do it... Find an ATX mobo that accepts Intel/AMD Mobile processors, which outperform C3s by an incredible deal, while still being very low-power.

Re:fp, yo

[MikeFM]

I have a mini-itx system with Via C3 processor and for most uses I'd not switch back to the power hungry heat blasting CPU's from Intel and AMD. Already a lot of people (family, friends, clients) I run into are interested in the Via CPU's and mini/nano itx mobo because they are power effecient, cooler, quiet, and have a very small form factor. For people who use their computer for business, the Internet, or even small servers these are exactly what they want. For most people a 1Ghz CPU is plenty. The only people that really want the higher speed CPU's are hardcore gamers and people doing heavy data crunching.

Also Via, I think, is being highly intelligent to take their low power CPU's and add-in special functionality that lets them do things like encryption and mpeg decoding at high speed with specialization. The only thing they are still really lacking is a powerful 3D graphics core that can also run within the low power requirements. If they could cover that base then I think they'd quickly become a major power in x86 CPU's. When they add that they'll also begin a wild spree in tiny homebrew game consoles (set top boxes being an existing common use). Even already I think there is a huge market that could be addressed by such systems as consoles.. for less CPU intensive games which are under represented in todays market. I think a set top box that could play CD's, DVD's, and a good selection of card, board, and classic arcade style games could be a major hit. I know far more people that play Solataire than I know that play Unreal. :)

Re:fp, yo

[akuma(x86)]

If there really is a profitable large market for these lower watt CPUs, then either Intel or AMD will just voltage scale down their offerings to be power competitive. It's really not that hard to build a 7 watt x86 CPU. Just do the math with voltage scaling and you'll see that a Banias can be brought down to 7 watts quite easily.

Re:fp, yo

[bhtooefr]

That's a 1.1GHz TRANSMETA. Even worse. At least a C3 1GHz can pull somewhere near 600-700 Pentium-series equivalence. There is a nice Micro-ATX mobo that takes Pentium M cpus, though - I might just buy one (it also has AGP and three PCI slots, and can be upgraded as much as a standard PC - just uses a Pentium M).

Re:fp, yo

[evilviper]

At least a C3 1GHz can pull somewhere near 600-700 Pentium-series equivalence.

I would say it's the very low end of that range, but fair enough...

There is a nice Micro-ATX mobo that takes Pentium M cpus, though

Personally, I prefer full-sized systems, but what the hell, why not... Would you happen to have the brand/model number of this mobo?

Re:fp, yo

[bhtooefr]

At least it's Micro-ATX, and not Mini-ITX (if you don't like Intel graphics)! It's the Radisys LS855, and their product page for it is http://www.radisys.com/oem_products/ds-page.cfm?pr oductdatasheetsid=1158.

And no one cares in 3... 2... 1.

[wankledot]

Seriously though. Has the Crusoe any impact at all? I can't think of a single device that anyone I know has bought, or wants to buy, that uses one. This chip will likely be the same. Also, Linus left Transmeta a few months back.

Re:And no one cares in 3... 2... 1.

[Anonymous Coward]

I know someone who bought one. Fantastic chip. Slow, sure, but beat the pants off everything (including the Pentium M) for battery life.

Re:And no one cares in 3... 2... 1.

[cgranade]

Also, Linus left Transmeta a few months back.
Probably when the chip was finished and ready to go into production and marketing.

Re:And no one cares in 3... 2... 1.

[kahei]

I use a Sony C1 mini-notebook with a Crusoe and it's excellent... fast enough to run whatever I'd normally run on a notebook but lots of battery life. And of course a pleasant warm 'non intel' feeling :>

Re:And no one cares in 3... 2... 1.

[woobieman29]

Hewlett-Packard has a very nice lineup of Thin Client products using the Crusoe processor. Crusoe is near perfect for this market niche: Low power and low heat are the way to go (you don't want to have a fan on a TC - moving parts just introduce another breaking point). The only problem is that the VIA processors compete very well at a lower price point. If Transmeta can keep the price down, the Efficeon should be great for high-end Thin Clients, as well as subnotebooks.

Re:And no one cares in 3... 2... 1.

[EddWo]

Compaq TC1000 Tablet PC. 1Ghz Crusoe. 4 Hours battery life.

Slightly slow with XP eye candy turned on. Great for taking notes, reading /. www.ncrg.aston.ac.uk/~jamescj/TC1000/photos.html

Re:And no one cares in 3... 2... 1.

[Anonymous Coward]

A co-worker of mine has one of these, and while it originally suffered from some heat problems (the chip would get too hot, system would start to chug, then eventually hang) after a quick send-back-to-compaq, it runs smooth now. Not sure what they did to fix it. I suspect it had something to do with the nvidia gpu getting hot though, and not the crusoe :)

But indeed, the battery life is impressive.

i'll wait for the next one

[prockcore]

which will be called the Masterbaceon, which is to be followed by the Ejaculaceon.

Re:i'll wait for the next one

[Anonymous Coward]

I love you.

Re:i'll wait for the next one

[Anonymous Coward]

which will be called the Masterbaceon, which is to be followed by the Ejaculaceon.

No, the Masturbaceon actually isn't next. It actually won't come out until after the Erectceon.

Uh.... question...

[miketang16]

Looks like a nice product, Linus and Co.

Correct me if I'm wrong, but didn't Linus leave Transmeta to work fulltime on the kernel, at OSDL? Cuz I know his email address definately changed from @transmeta to @osdl.

Re:Uh.... question...

[Anonymous Coward]

So, since he's moved to a different e-mail address, has he responded to your love letters?

Re:Uh.... question...

[Anonymous Coward]

Correct me if I'm wrong
You must be new here...people life for this shit.

Re:Uh.... question...

[cant_get_a_good_nick]

Whats with the "redundant" mods? He did take a leave of absence, yet anyone who corrected the post gets modded redundant. Do the mods you think the posters are part of some vast Microsoft/Intel conspiracy to trash Transmeta, so we need to hide where Linux is at? sheesh.

Re:Uh.... question...

[t0qer]

I'm just replying because I agree with you. The redundant mod was unfair (paticularly because I didn't know Linus had left transmeta)

Wake up mods!

Re:Uh.... question...

[Anonymous Coward]

I as a poster am insulted that anywone would think
that I am a part of a vast Microsoft/Intel conspiracy.
I am paid by IBM to post against SCO. I am loyal
to my employers.

Linus + Transmeta

[rf0]

Linus left Transmeta earlier this year

Rus

Re:Linus at Transmetta?

[wankledot]

No, he works for Open Source Development Labs Inc.

j/k

[Anonymous Coward]

Geek Tom: I suggest getting the new Transmeta CPU, it's cheap, and fast enough for our needs.

Un-educated Boss: OK *opens hardware selling website* is it under Intel or AMD sections?

Speed

[NetJunkie]

I hope it's faster than the current chips. I have a Compaq TabletPC with the current 1GHz Crusoe and while functional, it isn't that fast. The Pentium low power chips are faster. Even doing normal daily business tasks I couldn't see using one as my main PC.

Re:Speed

[evilviper]

I have a Compaq TabletPC with the current 1GHz Crusoe and while functional, it isn't that fast.

I was looking for someone to comment on the Crusoe chips... I know I had the same experience with the Via C3 chips, and expected the same from Crusoe...

While a C3 processor may say it's 1GHz or so, it performs like it's not quite a 500MHz Intel/AMD chip. Funny thing is, the incredibly low power, just puts it on par with the 500MHz AMD/Intel chips that it performs similarly to.

What I wonder is, why don't Intel/AMD bring back their old, slow, low-power chips to compete for what is obviously a viable market?

Re:Speed

[MikeFM]

Hrm.. I have a C3 933Mhz system and it competes well against my 1Ghz Celeron system. What kind of uses are you seeing where it is falling short? I do run Linux on mine and haven't compared them when running Windows so maybe that could be involved? Do you have the same type and amount of ram in both systems? Configured along similar lines?

Also my system uses a mini-itx mobo which from what I've read compliments the C3 processors to make up for some defeciencies of the CPU. Are you using such a mobo or a non-specialized one?

Re:Speed

[Nazmun]

Can you run some benches on both and report your fpu and integer performance? You can say they feel about the same but benchmarks i've seen report otherwise. Not that I don't automatically find you unbelievable. I'd just like some proof... I'm sure others in slashdot would find it interesting.

It's important to note that a celeron performs about the same speed as 400 mhz less pentium because of the cache in most instances.

Re:Speed

[evilviper]

What kind of uses are you seeing where it is falling short?

Absolutely, positively everything.

Videos that played fine even on older systems, couldn't keep up with th C3. Ghostscript processing, which used to take the exact same ammount of time, every time, took FAR longer on the C3. Mozilla/Netscape taking much much longer to startup than usual. System boot-up times significantly longer. X start/restart times noticably longer.

I do run Linux on mine and haven't compared them when running Windows so maybe that could be involved?

No, I didn't run Windows.

Do you have the same type and amount of ram in both systems? Configured along similar lines?

Not only did I have the same type and ammount of RAM, I had the exact same RAM chips. I took the RAM, Videocard, and Hard Drive from a working system, and transfered them to the C3 system.

Also my system uses a mini-itx mobo which from what I've read compliments the C3 processors to make up for some defeciencies of the CPU.

No, I was not using a mini-itx, but there really isn't much a motherboard could do to improve performance.

Re:Speed

[mahler3]

I have a Compaq TabletPC with the current 1GHz Crusoe and while functional, it isn't that fast. [...] Even doing normal daily business tasks I couldn't see using one as my main PC.

Perhaps not, but the Crusoe 5800 series wasn't really meant for that. It was meant for low power PC applications, where battery life is more important than performance. Last spring, I treated myself to a new Fujitsu Lifebook P1120, a very small and thin notebook with an 800 MHz Crusoe and an 8.9" touchscreen. It's a sweet machine, for what it's for. Yes, it's godawful slow when bringing up OpenOffice.org, but so is my main PC. See my review.

I haven't loaded Linux on it yet, but these people have. This page is worth a read, just for the travel-related stuff that they've crammed into their laptop. Anyone know where one can get a "Don't panic!" decal?

Re:Speed

[MikeFM]

Hrm, I'll see if I can run down somebody I know with a C3 on another mobo and see about doing a comparison. I remember reading that the mini-itx mobo was playing a major part in compensating for the CPU but I wouldn't think it'd be that much.

Have you tried turning off swap space on both systems and seeing how they compare then?

Re:Speed

[evilviper]

Have you tried turning off swap space on both systems and seeing how they compare then?

No, I really don't think either of the systems were even touching swap space (256MB of memory).

And here's a classic from another thread both of us are on (might as well consolidate the conversation):

Even fairly hefty tasks like unassisted dvd playback usually take the CPU usage to less than 90%

I have to say, DVD playback shouldn't take a tiny fraction of that! You may say that the C3 feels fast, but your own facts indicate that it's not quick by any means. Since DVD playback is smooth, even on a 400MHz system, you've just about proven my point...

Re:Speed

[MikeFM]

DVD playback on the equiv Celeron system uses about the same in my experience. What more do you want from it? With hardware mpeg decoding then sure you can play a DVD on like a 300Mhz CPU easily enough. All in software though and it takes a much more powerful CPU.

Re:Speed

[evilviper]

DVD playback on the equiv Celeron system uses about the same in my experience.

Well, somehow, you have a very serious problem with your Celeron.

With hardware mpeg decoding then sure you can play a DVD on like a 300Mhz CPU easily enough. All in software though and it takes a much more powerful CPU.

No, there is no mpeg hardware required at all. DVDs will play well on a 400MHz system... Divx files will play acceptably.

On my 1.2GHz Celeron, my CPU utilization is more like 20% when playing back videos like DVDs and Divx files (I'd give you more specific numbers, but I don't have my system nearby).

If you don't believe me, trust Google. You aren't going to find anyone that says you need a 1GHz system to play DVDs... They don't even go as high as 400MHz.

Re:Speed

[MikeFM]

Umm yeh right. I have several 300-500Mhz systems from AMD and Intel and none of them can play a DVD decently. Not without dropping shit loads of frames.

Re:Speed

[evilviper]

I can assure you, that I have been able to play DVDs, and Divx files on slower systems... Why you are unable to, I cannot begin to guess... Maybe your system don't support DMA on the DVD drives? Maybe you had the same DVD drive in all of them, that requires a lot of processing for some reason??? Maybe you are very low on memory? Maybe you have very slow videocards on those systems that couldn't keep up?

I don't know. But I can tell you that I have had no such problems, either with Linux or with Windows (98/2000), and I'm certain you can search the web and find many many posts from people that are playing their DVDs on systems with very low requirements.

Re:Speed

[MikeFM]

Interesting. I actually tested the speed of a faster C3 processor with more RAM on the system then my own and it's dramaticlly slower. It isn't using a mini-itx mobo so I guess maybe that IS the difference. Or maybe it's something to do with the video/audio drivers.

what niche?

[edrugtrader]

Let's hope they can capitalize on this before Intel starts filling the same niche

and what exactlly niche is that?? drastically overpowered portables and underpowered desktop devices?

Re:what niche?

[Henk Poley]

and what exactlly niche is that?? drastically overpowered portables and underpowered desktop devices?

In the 1970s they said people didn't need PCs. Who are you to say that "overpowered portables" are useless?

Re:Linus at Transmetta?

[Muerte2]

Last time I checked Linus worked for OSDL

Geez, who's writing these replies?

Re:Linus at Transmetta?

[ComputerSlicer23]

I'm assuming your intentionally posting mis-information. Linus has never been willing to work for a Linux distributor. Right now, he is on a leave of absence from Transmeta, and working for the an industry consortium. The name escapes me at the moment. Ahh, a bit of googling turns up he left for OSDL (Open Source Development Lab?). He announced that on the Jun 17th of this year.

http://kerneltrap.org/node/view/669

Kirby

Re:Linus at Transmetta?

[Anonymous Coward]

> Geez, who's writing these replies?

Trolls.

Re:Linus at Transmetta?

[Anonymous Coward]

You might want to check your facts before posting...

Linus does NOT WORK FOR REDHAT.

He works for OSDL http://www.osdl.org

Geez, who posts thes replies?

Re:Linus at Transmetta?

[Anonymous Coward]

low slashdot number...can you buy those?

Names

[BWJones]

Efficeon, Pentuim, Escalade, Celeron, Infiniband, Duron, etc...etc...etc...

Why not simply name a product for what it is instead of spending all those dollars to come up with lame names? Although I suppose that is the American way. Come up with some snazzy flash to sell a product based upon image rather than functionality.

Personally I like product names that mean something like Apple Powermac G5 2.0 Ghz. That is descriptive, says who built it, something about what is inside and how powerful just like Porsche 911 Turbo 3.6. Again, who made it, what it is, and some idea of how powerful.

Re:Names

[kryptkpr]

Intel Pentium 4 HT 3.06 Ghz

I rest my case...

Re:Names

[Osty]

Personally I like product names that mean something like Apple Powermac G5 2.0 Ghz. That is descriptive, says who built it, something about what is inside and how powerful just like Porsche 911 Turbo 3.6. Again, who made it, what it is, and some idea of how powerful.

Taking this off topic, but perhaps you meant to use a better example, like BMW 330Ci -- tells you that BMW makes it, it's an Inline-6 configuration engine (3-series), with a 3.0L displacement, in Coupe form. Porsche 911 Turbo doesn't tell you much (I've never seen Porsche cars referenced by their engine displacements). Is it a 930, a 993, a 996, or what? What year? air-cooled or water-cooled? You say 3.6L, but that's not part of the name designation. And more importantly, Porsche has moved towards the "flashy name" trend anyway, with the Boxster and now the Cayenne (nevermind that they both have internal codenames like 986 for the Boxster and I can't recall what the Cayenne's number is, something like 961). One could even call the 911 a "flashy name", since it originated as just a codename (for the 901, which couldn't be released as the 901 thanks to Peugeot's trademark on "x0y" numbers for car models) but now is nothing more than the name for Porsche's ass-engined car.

"Porsche Boxster" tells you just as much as "Intel Pentium III". Is it a 97-99 with the 2.5L engine? a 00-02 with the 2.7L engine? an 03-04 with the 2.7L that makes a bit more hp? Or did you mean to say Boxster S (Pentium III Xeon)? And again, is that a 00-02 with the 250hp 3.2L, or the 03-4 that adds more hp?

Re:Names

[Troll_Kamikaze]

Why not simply name a product for what it is instead of spending all those dollars to come up with lame names?

Because somehow I don't think "Shitteon" or "Also-ron" would move many chips.

Re:Names

[gl4ss]

you do realise that both of those examples have names that don't in reality tell ANYTHING about the products performance more than intresting numbers which could have been just as well pulled out of any hat(or substituted with some funky name).

sure, 911 turbo 3.6 tells that's it's a 911(but what does that tell you, without knowing beforehand what it stands for porsches?) and that it has a turbo and 3.6 engine(now, that doesn't tell much on it's own either since you still have to know the difference to a non-turbo version from somewhere else, it doesn't tell jack about it's power output or handling either).

apple g5 2.0ghz? oh please? that doesn't tell anymore than "efficieon 334" or "dubarun 888" either would tell of a product(sure you know that it's 2.0ghz g5 but that really doesn't take you anywhere on it's own, pure clockspeed, like displacement in engines, tells very little of the actual performance. sure it might give you some ballpark but that's just to let you know that it's not from the 80's. actually it could be some radio product from the 80's....).

i really much rather know some easy name for a nifty chip than some unusable transistor count number or whatnot(or a name that's half a page long).

anyways, you could have chosen military(non u.s, i'm afraid) firearms as better example, where you do (sometimes) actually know what kind of weapon you're referring to (it's caliber, is it an assault rifle, machinegun, anti tank rocket, or what).

Re:Names

[Kenshin]

I want a PowerMack with a Pimpeon 5 processor.

The ladies wouldn't be able to resist!

Re:Names

[nelsonal]

BMW isn't exactly a paragon of naming consistancy, thier coupes used to be iS (for sport I think) and the newer 323i's now have a 2.5l displacment depending on the year. I'd guess BMW decided that people wouldn't pay up for only .3 liters if they labeled them properly. M-B has been closer (they may have both done some funny things with forced induction engines especially in Europe) but their letter combos names don't make as much sense to me. It is certainly better than Porsche and most of the others, but you still have to look closely sometimes. Used Hondas are the worst, since the most common model had interchangable engines with vastly different characteristics.

Re:Names

[Gorignak]

Oh please. Everyone knows that a linux geek would have to refer to the "Porche 911 Turbo 3.6" as something completely abbreviated and totally enigmatic about what it's used for like "P911t3_6" and would be actually be named something like "Gnucar", "Kar", "LiCar", "FastGnu", or "TurboGnu".

Re:Linus at Transmetta?

[Anonymous Coward]

Linus doesn't work at Transmetta people, he's been at REDHAT since last year!

Alan Cocks works for REDHAT not Linus you moron.

Stupideon

[Anonymous Coward]

Moreon
Dumbeon
Dolteon
Dulleon
Dorkeon
Doofus seon
Dipshiteon
Dweebeon
Slashdoteditoreon

Green destiny

[goombah99]

I once had a look at the giant Transmeta Cluster at Los Alamos called green destiny.

the most impressive thing about it is how small it is.
over 280 blades + disk server in a single rack.

then you realize its sitting in an uncooled ordinary room shared by people. its not putting out hardly any heat the building air cant keep up with. its plugged into a normal building power strip, and its not making much noise.

then you see the benchmarks. this thing runs faster than the equivalent pentium on scientific codes. How is this possible you wonder if its doing this code morphing. the answer is that the transmeta JIT code morph results in code that executes faster on the transmeta than the original pentium code. On scientific code with lots of long tight loops the overhead of the code morph goes away and it runs faster. (the opposite is true for GUI desktop apps where it is constantly jumping around and not spending time in small sections of code.).

finally they show you the uptime. forever. no dead units. (on our other pentium cluster form the same manufacturuere we replace as mauch as blade a day)

these things are way better price performance ratio than pentiums when you factor in the total lack of building infrastructure, and maintainence. low heat keeps them stable.

Re:Green destiny

[4of12]

What I like about Green Destiny is that DOE is actually doing something about getting the most MFLOPS/Watt.

For far too long supercomputing has been a business of paying incredible money for only logarithmic gains in performance.

Re:Green destiny

[More Karma Than God]

Amen. For many applications the bottleneck in a PC cluster is the interconnection bandwidth, not the power of the processors. It seems that a lot of cluster builders want to generate heat instead of results.

Re:Green destiny

[salimma]

Interesting, is it not? Similar benefits as server-side Java, or ASP.NET - the overhead of JITing is insignificant once you start optimizing common code paths. Incidentally, Mono's XSLT implementation is supposedly already faster than libxslt written in C.

Which brings up an interesting question: official business and scientific benchmark figures. Chain as many Transmeta blade servers as needed to equal a given UltraSPARC/Xeon/Itanium2 server's power requirement, run load-balanced servers and benchmark.

Probably more useful to run web servers than database servers though.

Re:Green destiny

[forlornhope]

I wonder what the RMax value on a per chip basis. I was looking at transmeta's website and they say that it has a 256 bit VLIW. But that probably has nothing to do with my question, which is, how many double precision floating point calculations can one of these efficeon procs can do per clock cycle. The rest of the chip looks like it would be wonderful for a cluster. Also Im kinda wondering how many of these chips it would take to meet the RMax value of VTech's new G5 cluster.

Re:Green destiny

[Anonymous Coward]

Err, the problem is that server-side Java doesn't bring you any clear benefits, it doesn't reduce the need for anything (like decent processors do power).

Well...

[Anonymous Coward]

How about a beowulf cluster of...

Gimme cheap, low heat, fast

[Anonymous Coward]

If it were cheap enough, I'd buy a replacement for my ATX system with its intake fan, PS fan, exhaust fan, CPU fan and video card fan and 50C under load temperatures... and this is only an old non-overclocked Athlon 1.3Ghz.

Yes VIA is in this space... but let's get costs down and performance up!

And why not give us CPU throttling on the desktop.. summer mode (cool) winter mode (warm) video/audio mode (quieter) etc.

Via is competing hard with the new C3.

[Anonymous Coward]

http://www.via.com.tw/en/Digital%20Library/PR03101 4EdenN.jsp

Check out the link above; Via is pushing their C3 hard; also 7W consumption @ 1Ghz. Extremely small form factor (15mmx15mm) and built in AES engine (claims of 12Gbps). Plus they're being open source friendly, with OpenBSD already providing support for the AES portion in their crypto API framework.

Transmeta already licensed the x86-64/amd64 framework, I was kind of hoping maybe they'd announce something more intriguing than this.

Re:Linus at Transmetta?

[More Karma Than God]

Yes you can. Several months ago there was an Ebay Auction for a four digit /. user number.

Needles to say (as I have a newbieish six-digit number) I didn't buy it.

trasmeta processor plus mobo

[twistedcubic]

Is it possible for an individual to buy a Transmeta processor plus a motherboard on which it can live?

Re:trasmeta processor plus mobo

[pla]

Is it possible for an individual to buy a Transmeta processor plus a motherboard on which it can live?

Not that I know of, unfortunately. I've looked (though only quick checks here and there, nothing very thorough), and found that you can either get laptops, SBCs, and blade-based servers, but nothing desktop-like.

Really a pity, too... Although not as powerful as a typical desktop CPU, those of us running things like fileservers and/or internet gateways on our home LANs could benefit greatly from such a beast - The single most common cause of failure of machines I've owned comes from fans dying and taking something out with them (usually the power supply or CPU, both of which tend to take other parts out along with them when they overheat). Having a moderately powerful system with purely passive cooling would completely eliminate that problem.

If you find such a motherboard, please, let me know.

Re:trasmeta processor plus mobo

[Anonymous Coward]

If you actually check their web site (http://transmeta.com/developers/devkits.html) you can that Developers kits are forthcoming for TM8000.
You can already get them today for the TM5800 (although too expensive IMHO).

Re:trasmeta processor plus mobo

[martinde]

For the application you're talking about, check out the VIA C3M266 motherboard - $65 at newegg, with integrated everything including USB 2.0, and the VIA C3 1GHz. (I got mine for about $28 at MicroCenter locally here in Cincinnati.) That's what I'm using on a couple of these machines - low power, decent performance, great uptimes. Also it's a uATX form factor so you can buy a nice small case.

I like the idea of the TransMeta, but I doubt it can compete pricewise.

Re:trasmeta processor plus mobo

[Anonymous Coward]

Ya know what might be tight. Shink them down, slap on a little ram, an eprom, and a couple of gigabit ethernet ports and have a firewall in a card. It couldn't really cache, but that would certainly be cool.

Re:trasmeta processor plus mobo

[Mattsson]

I'd rather use the VIA EPIA-V mini-ITX C3-533MHz.
They've got more powerful versions of this, but those require a cpu fan.
This one only need a passive heatsink. =)
If you add a ide->compactflash converter and use one of their fanless 55w psu's you've got a machine with *no* moving parts.

Re:trasmeta processor plus mobo

[evilviper]

decent performance

I can't stand that. By what measure do they perform decently? They perform like several-year-old AMD/Intel chips at half their MHz...

Re:trasmeta processor plus mobo

[MikeFM]

By the measure that they can get the job done without hardly using up the available CPU horsepower? I've done a fair amount of testing of the C3 processors to find out how much they can deliver. Even fairly hefty tasks like unassisted dvd playback usually take the CPU usage to less than 90% even with a bunch of other stuff already going on (desktop, webserver, proxy server, mail client, etc). A Celeron CPU does about the same in my experience. Normal usage with the above mentioned desktop, webserver, etc only use something like 5-10% of the CPU. That'd seem to me to indicate that the performance is decent enough for most people.

Re:trasmeta processor plus mobo

[martinde]

By MIPs/watt and price/performance. Since this article is about the Efficieon (sp?) they seem like relevant metrics. My guess is the Transmeta would win in the MIPs/watt, but not in price/performance. (But this is an uninformed guess since I have never seen a Transmeta processor for sale retail.)

BTW, I've used a 75MHz Pentium for the application we're talking about - firewalling/proxying. It's not a CPU intense job, and the C3 does fine for it. I'm also using it for NFS service inside my home, and it does fine for that as well. (Would I serve hundreds of machines with one? - of course I would not, pick the right tool for the right job.) I went with this solution because the machine is up 24/7 and I didn't want a space heater for this application - I've got space heaters elsewhere, but I tend to turn them off when not in use. The machine is much faster and much cooler than the dual PPro it replaced.

Re:trasmeta processor plus mobo

[Jagasian]

Why go for something so slow? The VIA EPIA ME6000 is also fanless, but it is powerful enough to play DVDs, MP3s, and MPEG 4's.

I would personally prefer to wait for the Micro-itx 1ghz Edens with the full speed floating point units. Fanless AND powerful enough to do all my multimedia and serve as an emulation based console gaming system. Also the computer would be smaller than 5 CD jewel cases stacked ontop of eachother.

Oh, and I am typing this from my EPIA Eden 5000, running Redhat 9. No moving parts except the harddrive, which is an extremely silent Barracuda.

Re:Linus at Transmetta?

[geekoid]

do you know what it went for?

Linus didn't leave, he took leave of absence

[squashed]

see article.

Re:Linus at Transmetta?

[LesPaul75]

Yep, he's no longer with TMTA. Also, he had absolutely zero to do with the development of Astro (the defficeon).

how come nobody says...

[Anonymous Coward]

"imagine a [drool] Beowulf cluster of these suckers" anymore.

Re:how come nobody says...

[Anonymous Coward]

cause nobody needs to. Everytime something like this comes up, every last one of us is thinking IABCOT.

Re:Linus at Transmetta?

[More Karma Than God]

I don't recall the exact figure but it was less than $100.00

They didn't list the actual username either.

Intel?

[lwells-au]

If I were Transmeta, I wouldn't be so worried about Intel filling a niche as they seem to have their hands full with AMD at the moment. Rather, I would worry about Via and their Eden line of CPUs. This came in the mail yesterday (excuse the huge post):

For Immediate Release

VIA Unveils New NanoBGA VIA Eden-N Processor, World's Smallest & Lowest Power Native x86 Processor with Industry's Most Advanced Embedded Security Features

VIA PadLock(TM) Security Suite paves the way towards pervasive security for connected PCs and smart digital devices

MICROPROCESSOR FORUM, San Jose, CA, 14 October 2003 - VIA Technologies, Inc., a leading innovator and developer of silicon chip technologies and PC platform solutions, today unveiled the new VIA Eden-N processor at the 16th annual Microprocessor Forum conference in San Jose.

Based on a new streamlined Nehemiah core architecture the VIA Eden-N processor delivers enhanced digital media performance features and the industry's most advanced embedded hardware security technologies in a tiny nanoBGA package measuring just 15mm x 15mm to provide the world's smallest, lowest power, and most secure native x86 processor.

"With the launch of the new VIA Eden-N processor, we are continuing to enhance our leadership in developing innovative low power processor solutions that address the growing security needs of corporate, government, and home users," commented Glenn Henry, President, Centaur Technology. "With its fast, efficient, and cost effective hardware encryption capabilities, the VIA Eden-N processor is paving the way towards pervasive security for all manner of PCs and smart digital devices by enabling robust protection for sensitive corporate and personal data at the system level and across fixed and wireless networks."

World's Smallest & Lowest Power Native x86 Processor

Combining the streamlined Nehemiah core architecture with a new nanoBGA package that measures only 15mm x 15mm, the VIA Eden-N processor is approximately the size of a penny and less than half the size of the Intel(R) Pentium(R) M processor, and with its enhanced thermal characteristics achieves a Thermal Design Power of only 6 watts at 800MHz and 7 watts at 1GHz.

The processor's ultra low power consumption is of paramount importance for small and quiet connected devices because it means that they can run without any cooling fans, significantly enhancing reliability during 24/7 operation. Its small size dramatically reduces motherboard real estate requirements, facilitating the development of smaller, more highly integrated footprints such as the forthcoming 12cm x 12cm VIA Nano-ITX mainboard form factor.

World's Most Advanced Native x86 Processor Hardware Security Features

To address the growing need among corporate, government, and home users to protect sensitive data on their PCs and over networks, the VIA Eden-N processor adds advanced new on-die security features to the existing PadLock Security Suite on the Nehemiah core, including the new PadLock Advanced Cryptography Engine (ACE) and two hardware based RNGs. PadLock ACE provides world-class performance for the US government approved Advanced Encryption Standard (AES), performing cryptographic functions for securing e-mails, personal files, online transactions, and networks, including the latest high-bandwidth 802.11g wireless networks. Padlock ACE encrypts at rates of up to 12.5 Gigabits per second (Gbps) with a 1GHz VIA Eden-N processor, more than eight times faster than the best software AES implementation from a power hungry 3GHz Intel(R) Pentium(R) 4 processor based system that encrypts at a rate of a mere 1.5 Gbps.

"The AES accelerator core in the VIA Eden-N processor provides developers with an easy-to-use and substantially faster substitute for current software AES implementations," said Benjamin Jun, Vice President of Cryptography Research, Inc. "Its high performance and simple programming model should increase deployment of network and disk e

Slashdot editors are on crack (what else is new)

[illumin8]

Today we have two stories that about new processors that are about to be released. The Transmeta processor, while an incremental improvement, is nothing to really get excited about. The Clearspeed chip is simply vaporware.

Yet the one real story that is actually interesting "News for Nerds" was rejected by the Slashdot editors.

Sun Microsystems today announced it's roadmap for Throughput Computing. Remember how Sun has been talking about putting multiple cores on a single chip? Well, systems will be shipping in early 2004 that offer twice the performance of current top-of-the-line Ultrasparc IIIi chips. By late 2004, they will offer three times the performance. Coming in 2005, the second generation of this technology will offer 15 times the performance of current Ultrasparc IIIi technology. The roadmap extends to generation 3 (no date yet), which will offer 30 times the current performance.

This is way beyond Moore's Law and actually news that I want to read insightful Slashdot comments on.

With the anti-Sun bias the Slashdot editors show I guess I shouldn't be surprised...

[sarcasm]
Vaporware and anything having to do with Linus Torvalds' old employer are ever so more important than something that will radically change the computing landscape over the next few years.
[/sarcasm]

Re:Linus at Transmetta?

[Anonymous Coward]

I'm assuming your intentionally posting mis-information.

Errr, yeah, that seems like a completely reasonable conclusion to jump to...

OPEN QUESTION TO /. COMMUNITY:

[Anonymous Coward]

Please don't mark this as a troll. This is a totally serious question, and I would like to understand the answer.

Why do /. posters (story submitters especially) take every possible opportunity to suck Linus's shiny red nob (even when it's not a remotely plausible suckage)?

He doesn't work at Transmeta anymore, as any self respecting geek would know, and even when he did, he was just one of many employees. Given this, why oh why would you call Transmeta 'Linus and co.'? It makes no sense at all, and I would like to understand this. Is it hero-worship gone awry? You love Linus so much, that you find a way to give him propz at every opportunity, whether they're warranted or not?? WTF!!?

Re:OPEN QUESTION TO /. COMMUNITY:

[Anonymous Coward]

Dear Valued Cumguzzler,

Delicious, wholesome semen contains 40% of the daily amount of Vitamin E and 20% of a day's needed Vitamin A. Swallow semen for a guaranteed health rejuvenator!

Troll Pharmaceuticals
Christmas Island
trollpharmse.cx

Ability to trademark

[chmilar]

The naming game is done to allow the name to be registered as a trademark, and have the trademark enforcable.

Intel came up with "Pentium" because the chip cloners were calling their chips 80x86-compatible. Intel had not trademarked the numbers, and it is possible that such a generic trademark might not stand up in court, anyway. Now, if AMD wants to claim Pentium compatibility, they have to acknowledge that Pentium is a trademark owned by Intel.

These "made up" names are easy to trademark and enforce, since no one will "accidentally" use the word Pentium or Efficeon.

Apple might not be able to enforce a trademark on plain "G5" (after all, there are plenty of things referred to as G5 or 5G), but "Powermac G5" is not a problem.

Re:Ability to trademark

[TPFH]

That's what I heard. Although, the way I heard it was that it is clearly not possible to trademark a number at all. It probably has to do with not being able to trademark commonly used words.

Maybe it is like you can trademark Fish[tm] brand Fish, and since number are so obvious with computers that that.... nah. It's probably just absurd to trademark any number in any context.

TROLL?! MOD PARENT UP!

[Anonymous Coward]

Troll?! Geez!

Re:TROLL?! MOD PARENT UP!

[Anonymous Coward]

The mispelling of Alan Cox's name is rather trollish, yes, as is the general tone.

Efficiency Apartment

[Esion Modnar]

Makes me think of "efficiency apartment" which is not a good association.

Re:Are you UNCIRCUMCISED?

[Anonymous Coward]

Haha you got your cock mutilated! LOLOLOLOL.

Re:Linus at Transmetta?

[Anonymous Coward]

> Errr, yeah, that seems like a completely
> reasonable conclusion to jump to...


Geez! What's your conclusion then?

Re:Why don't you all buy Motorola, you GNU hippies

[Anonymous Coward]

Nice troll! :-) Motorola spun off their chip biz.

Re:Slashdot editors are on crack (what else is new

[sheddd]

Sun is doing some interesting processor work... really different way of going about the processor race...

But will they exist in 2004? They're in fairly deep shit revenue wise now, and I think their Solaris strategy is a losing one (i.e. I think they should dump solaris and start selling services and linux add-ons to make linux administration super easy).

How about

[Wolfier]

Abalone...

Re:How about

[wayne606]

Ahh, baloney...

What about the polymorphic CPU

[Anonymous Coward]

Whatever happened to their CPU that could natively run machine code for several other CPUs?

That's the problem with vaporware, Transmeta will always be associated with the stink of failure now. It doesn't have have Linus to lend it support among the slashdot crowd.

Re:What about the polymorphic CPU

[TheRaven64]

All of the Crusoe series are VLIW chips that morph x86 code into its native format and then run it. There is no theoretical reason why they couldn't ship a version of their code morphing software that would take PowerPC* or SPARC code and run it. One thing I would like to see would be a Java bytecode version, so that you didn't have to convert from bytecode -> x86 -> Crusoe instruction sets to run a Java app. The reason they don't, I suspect is lack of demand (from people willing to put a large enough order in to justify writting and optimising the code) for other architectures. This in itself is quite surprising, since I can see a potential market for them among UNIX users who wish to switch to x86 but have some old binary only code for Alpha (or whatever) that they need to be able to run. Having an SMP crusoe box that could switch one (or more) of the CPUs to running Alpha/SPARC/MIPS/POWER/Whatever code would be a good migration path for these people (if the Crusoe supported SMP, which I don't believe it currently does, although I could well be wrong).

* The reason it's hard to emulate a PowerPC on a conventional x86 chip is the near total lack of usable registers in the x86 architecture. I don't know how many registers the Crusoe series have, so I don't actually know if this is possible.

Re:Slashdot editors are on crack (what else is new

[Anonymous Coward]

Talk about vaporware...

yes, he does!

[Anonymous Coward]

Wow, all the trolls writing "Linus isn't with Transmeta" are getting modded up?!?

He just took a leave of absence:

'"The other big news -- well for me personally, anyway -- is that I've decided to take a leave of absence after 6+ years at Transmeta to actually work full-time on the kernel," Torvalds told list members'

Re:Slashdot editors are on crack (what else is new

[Anonymous Coward]

B-But..Linus works for Transmeta!!

Linus & Co

[chabotc]

Small correction on "Looks like a nice product, Linus and Co". Linus has moved to OSDL, so for now it's just & Co.

Re:Slashdot editors are on crack (what else is new

[Anonymous Coward]

I have a feeling Sun will end up with the likes of DEC.. They will design this amazing processor. Light years ahead of anything else. But the company will sink into the ground only to be acquired by a competitor who will indefinitely shelf acquired technology and assets.

Such a shame.

Re:Slashdot editors are on crack (what else is new

[adrianbaugh]

[sarcasm] Vaporware and anything having to do with Linus Torvalds' old employer are ever so more important than something that will radically change the computing landscape over the next few years. [/sarcasm]

You're new round here, aren't you? Seriously though, thanks for the link. Most interesting.

Re:Slashdot editors are on crack (what else is new

[cmcguffin]

> The Clearspeed chip is simply vaporware.

> Coming in 2005, the second generation of [Sun's] technology will offer 15 times the performance of current Ultrasparc IIIi technology

According to Sun's press release, they will release hardware in "2005/2006" that is "expected" to increase throughput by 15 times for "Web, application serving, simple databases".

> Vaporware

"You keep using that word. I do not think it means what you think it means."

Linus

[h4x0r-3l337]

"Looks like a nice product, Linus and Co."

Linus left Transmeta about 4 months ago.

Re:Slashdot editors are on crack (what else is new

[illumin8]

With $5 billion in the bank they will be here for a very long time.

Imagine

[Anonymous Coward]

A beow*whack*

Is the cache pre or post code morph

[metalhed77]

I'm curious if anyone knows this (I know nothing about chips). But do transmeta processors use the cache to cache code before or after it has been code morphed? I saw the large cache and assumed that must be the reason as it would seem that increasing the amount of code that need not be recompiled would probably help out the transmeta chips quite a bit

Re:Is the cache pre or post code morph

[i am fishhead]

I think the bios sets aside 16 MB or so of memory at boot time as a cache for translated code, so you're not just limited to cache alone ... that'd be insane. I'm guessing that's why they do the 16 MB thing ...

Re:Is the cache pre or post code morph

[JamesP]

I believe it's before (x86 code), since the "morphing" changes w every cicle. (i.e. it morphs it ever so efficiently)

Re:Slashdot editors are on crack (what else is new

[macshit]

The in-depth reviews of the new ultrasparc(s) I've seen have been rather less ... flattering. A doubling of speed is not quite as impressive if you were way behind to begin with.

As for `Coming in 2005 ... will offer 15 times the performance' -- what was that you were saying about vaporware?

Re:Slashdot editors are on crack (what else is new

[D.A. Zollinger]

Couple of problems with your link. First, you are linking to a press release - basically a marketing ploy by Sun saying, "Hey! Look at what we are doing! Don't forget about us!" Although it doesn't happen all the time, I'm sure the Slashdot editors would prefer to get links to other news sources instead of just links to corporate propaganda.

Secondly, how is what they are developing with their chip multithreading technology any different from what Intel has already PRODUCED with their HyperThreading enabled processors? Intel HT chips are not only already on the market, but being used in production machines. Sun's processors, on the other hand, won't ship until next year.

I'm sorry, but this is NOT "News for Nerds", and it was correctly rejected. To be honest, I'm surprised that this got moderated up. After all, when you submit a story the submission page states, "Note: grousing about rejected submissions is Offtopic and usually gets moderated that way. It happens, don't take it personally."

Re:Slashdot editors are on crack (what else is new

[illumin8]

According to Sun's press release, they will release hardware in "2005/2006" that is "expected" to increase throughput by 15 times for "Web, application serving, simple databases".

The 2005 number I took from this other link, which you might want to check out. The goal is 2 years to produce a CPU with 16 cores on its die, which would give you the "15 times current performance" figure (you probably lose a little performance due to scheduler overhead). That would be right around October 2005, but you're right, the other page says 2005/2006, so who knows? If they can offer half that increase in performance in two years I think there will be a sea change in the industry. Those little blade servers would be pretty amazing if each blade had the equivelant of 16 CPUs in it and the power/cooling requirements of only 1.

>> Vaporware

"You keep using that word. I do not think it means what you think it means."

Touche... I must admit that the Ultrasparc IV chips aren't released yet and are technically vaporware, but Sun hasn't really been known to wildly miss ship dates and I think it's fairly certain that they will at least hit the first couple of milestones.

Re:Slashdot editors are on crack (what else is new

[GoBears]

first, Moore's Law is about density. processor speed improvements are just a historical corollary.

second, everyone's roadmap leads to a multicore design (an inevitable development as increasing transistor count gets easier relative to shrinking).

third, improving "throughput" performance is not the same as improving serial performance. you have to find and then exploit parallelism, and neither is easy outside of data centers (which is what the US-IV and friends are aimed at). personally, i have lots of problems that are bottlenecked on serial performance.

it's not clear that throughput computing is going to shift things dramatically in sun's direction.

(and this is from someone who likes sun gear. this week i tried to propose a V440 as a VLM server for my department and lost out to a pile of 8GB dual G5s on price/performance. the difference in serial performance at this point is just ridiculous.)

Nice, but unfortunately...

[Anonymous Coward]

One of the major Transmeta users - the Compaq (now HP) TC1x00 Tablet PC range - has just moved across to Centrino.

I own a TC1000, and to be honest, it is a versatile little machine, but its CPU power did let it down a bit. Granted, my tablet did outlast others in extra long meetings, with the battery going on way past the power-hungry Acer's and Toshiba's, but the reduced power was noticable - tolerable, but noticeable.

The strange thing was that once it got going, the Crusoe in it was actually quite nippy and would happily run VS.NET, SQL server and IIS fro development purposes simultaneously. From what I understand the way in which code morphing software works though, the biggest delay is when you open an application - and unfortunately, this tends to be the most perceptable point of a machine's speed. HP have even opted to build a Celeron low-end tablet, shunning Transmeta totally across their range. Sad but true. :o(

I always considered tablets to be a major niche where Transmeta could clean up quite a bit. I have yet to play with the Centrino model, but if their battery length has been improved, then I see Transmeta having long term problems competing effectively in the tablet PC market in the near future.

Re:Nice, but unfortunately...

[mocm]

Considering that the Centrino was Intels answer to the Crusoe and the Efficeon is Transmetas answer to the Centrino, you would expect that Transmeta will still be a player in the tablet market. Don't be surprised if the next HP tablet will use Transmeta again.
Here you will find a list of Efficion products.

Re:Nice, but unfortunately...

[MikeFM]

In the least Transmeta and Via and forcing Intel to seriously work on the power usage and heat issues of CPU's. Without the competition there would probably be no Centrino. If we could just start a real competition among those companies that supply the batteries and such then maybe we could finally get decent uptime for laptops and such. I believe that the other major power consumer, the screen, is already an area of competition and seemingly we'll be seeing some improvement there soon.

*sighs* Crusoe was actually kind of a cool name I thought. Efficion is just a bad (even if more trademarkable) name. Via's Eden name is fairly cool still. I think the name alone will benefit sales somewhat.

Re:Nice, but unfortunately...

[hackstraw]

FWIW, Centrino is not a processor, its a mobile technology "suite", I guess you could call it. Its a bundling of the MB, the Pentium M processor, wireless networking, etc.

Re:Why don't you all buy Motorola, you GNU hippies

[toddestan]

Doesn't Apple now use IBM chips... which incidently partnered with Cyrix back in the day, and Cyrix got sold to VIA, which makes the Effecion. So you see, the Effecion *is* a Motorola chip, just with 4 degrees of seperation!

Hey

[Mr.Zong]

Why not PDA's? This is what i never understood about Transmeta, if the darn thing is so efficent, why not use it one of the fastest growing markets out there (Scalability seems kinda moot considering what they're cramin in this current gen of pda's)? I would love to gobble up one of those funky named procs and make it into my best new wardriving buddy. And I mention this, because who the heck uses those tablet pc's anyhow?

Re:Hey

[MikeFM]

The Via CPU's and even mobo's could be targeted for the PDA market pretty easily. Given that the nano-ITX mobo is less than 7 inches per side and the actual CPU is about the size of a penny. It'd be a good idea for them to offer a version with the external connectors and any other extra fluff removed that could be used for a PDA and similar apps.

I was wondering if it'd be possible for them to offer a mobo that could have multiple CPU's so I could get a dual or quad system. With the CPU's that small and power effecient why not cram more of them into the available space. A quad-CPU system in the size of a PDA would be pretty awesome.

If you're into doing it yourself you can build a really small portable system easily enough. You can get a power conversion board that fits snug to the mobo and removes the need for a bulky power supply. Also you can use CompactFlash as the hdd so that there are no moving parts and everything can fit into a space about the size of my dvd box for Fellowship of the Ring. I like tablet PC's but they are more useful if you can custom build them to your own needs.

Re:Linus at Transmetta?

[ashp]

Finally, a use for my account! I hope to make millions from this, and retire immediately.

compare to PPC?

[panserg]

how is it different from Apple Powerbook based on PPC?

Re:compare to PPC?

[Anonymous Coward]

it won't burn your jewels?

Re:compare to PPC?

[panserg]

I'll anser you just to make sure that some idiot would not mod you up as an informative one.

Apple Powerbook doesn't burn your jewels.

Re:compare to PPC?

[ignorant_newbie]

>Apple Powerbook doesn't burn your jewels.
the only reason you could say that my wife's tibook doesn't burn your jewels is that it's so damn not that you'd be nuts to put it on your lap

Excuse me?

[Doktor Memory]

While god knows I'm the first to agree with the general sentiment that the slashdot editors are sucking the glass dick (see my sig, etc), are you seriously maintaining that the release of a white paper (ie: "We plan for our next generation of computers to be EVEN FASTER, woo!") detailing a series of products with no ship dates attached is much more important than a product that has actually shipped?

The Efficeon (god, what an awful name) and the new Eden are both real products that I can now order in batches of 1 or more. The press release you cite is just Sun saying -- again -- that this time, really for sure uh huh they've whipped the UltraSparc's performance issues...in the next version...real soon now.

Re:Excuse me?

[illumin8]

... are you seriously maintaining that the release of a white paper (ie: "We plan for our next generation of computers to be EVEN FASTER, woo!") detailing a series of products with no ship dates attached is much more important than a product that has actually shipped?

I guess I should have included news sources in my links, because there sure are a lot of them.

All sarcasm aside, your point is well taken. I like the article on the Sun site because it explains more about how the technology works than any of the news sites, but it seems that Slashdot editors are much more likely to accept a story if the link points to an "impartial" news source rather than a press release. Thanks for the tip.

The Efficeon (god, what an awful name) and the new Eden are both real products that I can now order in batches of 1 or more.

Actually, neither one of those products have shipped:

From Via's press release:

The VIA Eden-N processor is sampling now and is expected to start appearing in secure networking, entertainment and communication devices in Q1, 2004.

From Transmeta's press release:

... the Efficeon processor family will be competitive with, or outperform competing microprocessors operating within critical thermal limits such as the 7W limit for typical fanless notebook designs. Systems based on the processor are expected to begin shipping in the fourth quarter.

Re:Excuse me?

[salimma]

Systems based on the processor are expected to begin shipping in the fourth quarter.

In all likelihood the chip has been released, it's products based on it that have not.

Re:Excuse me?

[i_r_sensitive]

But, did you read the white paper?

Sun makes a compelling point. Upping processor performance in a vacuum is not the be all, end all to improve overall system speeds. Upping processor performance in view of the limitations of the whole system makes sense.

While this is only a white paper, not an actual chip release, it does have some very promising potential, particularly for web servers and their ilk.

<sucking_up>
I'm not so sure about the editors galss dick sucking either. After all, we only /. a site while the post is up, we /. the editors all day, every day.
</sucking_up>

Re:Slashdot editors are on crack (what else is new

[mic256]

Wasn't it $5+ billion before they lost that $1+ billion last quarter ?

Nice OpenBSD plug!

[Voivod]

Anyone else notice that Theo de Raadt was quoted in the VIA Eden-N press release? We're at the point now where PR departments of billion dollar hardware companies ask project leaders of open source projects for pithy quotes to improve their press releases. Quite a suprise... and oh my GOD is that VIA processor cool. Can't wait to see one.

Or maybe the VIA website noticed Slashdot in the referer when I clicked to read it, and inserted quotes appropriate to the audience? :-)

Re:Slashdot editors are on crack (what else is new

[illumin8]

Thanks for the link; that article was an interesting read. I agree with what you say about parallelism vs. serialism. If you have a single-threaded application it will probably run slower on a CPU with lots of cores. But Sun's primary market is large database servers that have hundreds or thousands of users accessing them at once, thus, many threads, and the perfect candidate for lots of cores.

(and this is from someone who likes sun gear. this week i tried to propose a V440 as a VLM server for my department and lost out to a pile of 8GB dual G5s on price/performance. the difference in serial performance at this point is just ridiculous.)

Price/performance the G5 probably wins hands down against the V480 (I think that's what you meant), but did you take into account support? I hope you don't have a hardware failure at 2:00 am on a production G5... Good luck getting Apple to fix the hardware for you at that hour.

Re:Slashdot editors are on crack (what else is new

[cmcguffin]

Finding ways to exploit parallelism and hide memory latency are important problems in processor design, and definitely interesting stuff, my counter-sarcasm notwithstanding.

"market-leading quantities of high-quality entropy

[gdeinsta]

From the VIA press release touting the Eden-N: "Unlike software RNGs or existing multi-chip solutions, the PadLock RNGs utilize electrical noise on the CPU to generate market-leading quantities of high quality entropy for use in the creation of security keys. The entropy is stored in a collection buffer where it can be accessed directly via a dedicated x86 instruction set without the use of vulnerable software drivers." Best new technology of 2003!

Re:Why don't you all buy Motorola, you GNU hippies

[Anonymous Coward]

Damn. Need to keep up with all this stuff.

Why don't you buy TSMC chips, you GNU hippies!

There, that one should stay around for a while.

Re:Slashdot editors are on crack (what else is new

[illumin8]

Wasn't it $5+ billion before they lost that $1+ billion last quarter ?

It was a non-cash loss, as shown in this excerpt:

Sun also announced Monday that it is increasing an allowance it had made for its deferred tax assets, resulting in a $1 billion noncash charge in the fourth quarter, which ended June 30. As a result, the company said it is revising its fourth-quarter results to a loss of $1.04 billion, or 32 cents per share.

Linus doesn't work for Transmeta any more...

[Anonymous Coward]

...so we can stop being excited about their processors now.

*whew* Keeping up the dogma was really getting tough. Of course, now there's the "Transmeta r0xx0rd until he left, now they suxx0r!" reality distortion to create, but that will be a cakewalk compared to some of the other OSS burdens, like pretending ESR is rational.

Why the Intel bashing??

[altek]

Really, why is it important that Transmeta jump on this before Intel? Just because Linus works (worked?) for them??

Competition drives this industry, and so far Intel has consistently put out good, reliable, fast architecture (at least for desktop PCs...). At least I have been happy with it. I'm sure if you write drivers or compilers or other assembly code, you may prefer the instruction sets of another arch, but for the end-user of Intel products (99.9% of you I'm sure), I see no problem with Intel.

I'm not bashing Transmeta either, like I said competition is good, but seriously, why does the /. crowd seem to have a bias against Intel??

And I know this is /. so it's obligatory, but don't just reply to this and say asinine things like "because they suck" or "they're a giant corp like M$". They got that way for a reason, you know.

Just my $23.. inflation's a bitch ain't it!

Oh and what's the deal with the /. server(s) being so flaky today??

Re:Why the Intel bashing??

[Anonymous Coward]

Oh and what's the deal with the /. server(s) being so flaky today??

Actually, that's been for the last week or so... Not taking comments, Slashdot.org going down, 500 errors, etc.

--bhtooefr posting as AC because talking about /. problems can cause your Karma to go down (or UP, after all, /. is fucking up)

" Looks like a nice product, Linus and Co."

[psyconaut]

Linus left Transmeta and is now at OSDL!

-psy

For the sake of redundancy...

[keith.bronstrup.com]

Linus is no longet a Transmetite

Linus no longer works for Transmeta

Linus left Transmeta months ago

Linus hasn't worked for Transmeta in months

And it would majorly kick ass to put 16 of these bad boys on a mobo! Even though Linus had nothing to do with their making, unless they started the project before he quit several months ago!

Who exactly is on crack ??

[vlad_petric]

While I strongly agree with you that Clearspeed is bullshitware, everything else is simply incorrect.

1. A processor running at 1 GHz while dissipating only 7W of power is a major acomplishment (even if it's the speed equivalent to a Pentium 3 at 700MHz). It's not uncommon for a laptop processor to consume around 45W (i.e. half the total power consumption). 7W simply means that you have twice as much battery time ...

2. Sun processors, taken individually, suck big time. Their power come from scalability (putting 32 of them on a single board, for instance).

Furthermore, 15 times the performance should always be taken with a grain of salt. (and you were the one complaining about BSware ...). On what benchmarks ? I guess they always use MIPS (which is only meaningful for marketing), and never a real benchmark.

And why exactly are you comparing Sun processors to Transmeta ? They're really meant for different markets ... you don't put server procs in a laptop or mobile procs in a server ...

Not to mention that Sun starts to smell like a decaying corpse.

Transmenta Efficeon vs Pentium M

[Anonymous Coward]

Does anybody have some comparisons between the Efficeon and the new Pentium-M low power chips?

Why Google doesn't use Transmeta CPUs

[embobo]

I went to a U Washington CSE collqium given by a guy from Google. Their clusters have such a heat issue that they can't keep the racks next to each other. Somebody asks why they didn't use Transmeta CPUs. I beleive the guy's response was that, even taking into account the additional space and cooling requirements, Pentium-based systems still cost less than Transmeta-based systems.

This talk may be online...ah yes, here (search for "Google") it is:

Urs Hoelzle (Google)
The Google Linux Cluster
Windows Media Archive
Real Media Archive
November 5, 2002, Colloquium

Re:Why Google doesn't use Transmeta CPUs

[MikeFM]

I wonder if that'd still be true if Transmeta wasn't an underdog? If they could compete in production costs with Intel how much could they drop their prices? Or if Intel licensed the technology and used it in their own CPU's.

I don't get it

[Pan T. Hose]

Wouldn't it be faster if we programmers could target the real instruction set of Transmeta CPUs instead of writing the legacy x86 only to have it "morphed" into the real deal? Wouldn't it make more sense to work with gcc staff telling them about all of this software emulation voodoo magic and give us the option of turning it off, so we could use the internal beautiful RISC architecture and have even faster and more efficeont arch? Could someone please explain it to me? I haven't got any answers from Transmeta core team about those very issues.

Re:I don't get it

[Anonymous Coward]

LOL! This deserves to be held in the same regard as the 'BSD is dying!' and the 'it takes 20 minutes to copy this file on my Mac' family of trolls. I dip my hat to you, Sir Pan T. Hose.

Re:I don't get it

[bhtooefr]

Actually, Linus tried porting Linux to the Transmeta, and a Transmeta-optimized x86 compile was faster than a native Transmeta compile.

Re:I don't get it

[Mr. Jaggers]

Any references here?

Re:I don't get it

[bhtooefr]

I don't quite remember where I read it, but it might have been Just for Fun by Linus himself.

the NanoBGA EDEN-N

[sharkey]

That was the name of the bird that stole Baby New Year in the Christmas puppet-motion special, wasn't it?

ACPI for Linux

[jir0]

I actually haven't gone that far with Linux on Crusoes, but from what I've seen, it's not that bad. Putting aside the lack of ACPI support kernel-level, hardware itself works out for low power consumption quite well (at least better than regular Intel PIIs). Yes, it's a pity that Transmeta/Fujitsu don't do more to provide hardware support for Linux, especially since both have otherwise been participative in the whole Linux hype. (Fujitsu has, at least. AFAIK, Torvalds used to hack at Transmeta, not officially for Linux then, nothing more.) I don't believe Linus worked kernel-level at Transmeta, or on 32-bit x86 at that.

On the bright side, there's ACPI4Linux, which aims to provide a better power management interface between the kernel/OS and BIOS. Last I heard, they have quite active development (acpi at sourceforge).

Also, I have no preference for wide-screen laptops. I'm all for portability (light, small, thin, long battery life) The laptops I've been eyeing for a while are the Fujitsu Lifebook P-series (10", anyone? ^_^) (mostly on Crusoes), the new (anticipated) Crusoe line of 11.3" ones (mostly 6 hours min), and the Compaq Armada M300 (11.3", really cheap (USD 250), 3 lbs., too bad it's PIII-500 only).

Re:ACPI for Linux

[CAIMLAS]

I, too, have been drooling like a rabid cheerleader over the Fujitsu Lifebook P-series. I've yet to see one in person, and despite all their shortcomings in linux (as well as the sky-high price :( ), I'm still anticipating having the funds to purchase one. I'd love to be able to put one in a daybag and not have to worry about batteries, etc.

Oh, and lack of heft is nice, too. THough, like I said, I've yet to see one myself. :(

Copyright Infringment

[AvengerXP]

Intel might just sue for using "eon" at the end of the processor name. I'm afraid that what i just said will come true like that other story on slashdot the other day...

Re:transmeta processor plus mobo

[frank_adrian314159]

I don't know. My Fujitsu Lifebook p1120 seems to have one :-). Whether you can get the mobo plus chip sans the rest of the computer is still an open question. That being said, this little machine is great (in no small part due to the low power consumption of the Crusoe) - built-in wi-fi, runs for 6 hours on its extra-life battery, and I put on Linux as a dual boot. The only downside is that I still can't find a touchscreen driver for the Fujitsu-made touchscreen...

Re:Why don't you all buy Motorola, you GNU hippies

[Anonymous Coward]

Apple != Motorola, so your equation is all wrong.

Two Places To Look

[istartedi]

First, try BWI.com. There you can various types of boards that use the Transmeta CPUs (though Efficeon is probably not there yet). The most reasonably priced ones are made by Wincomm; but for some reason they aren't linked off BWI's main site any more. Last time I looked, I was still able to get the listing of Wincomm products by using their search function. Some projects such as the CharmIT wearable computer were based on the Boser HS-1600 board, which seems to be a popular choice. It costs something like $800-$1000. I have no idea why the things are so bloody expensive. Bear in mind that you are usually getting built-in memory and LCD controllers, video, sound, etc. It's almost a complete system. You still need a power supply, keyboard, mouse, and monitor. Also, the chip is soldered to the board, so it's not a true CPU-mobo solution. Clips for the Crusoe do exist, it's just that the board makers haven't used them. I seem to recall having stumbled across one, but it was by a manufacturer that's not well known and I lost the link... Sorry. Oh, I almost forgot the best thing about BWI: They quote prices for onesies and twosies right there on the website. No need to call sales. What a refreshing approach!

The other site is All American There you can actually get Transmeta CPUs, but without a mobo to plug them into this is only of interest to you if you license the reference design from Transmeta and contract for the fabrication of your own boards. Technicly that's not nearly as daunting as it sounds. With the proper files, you can usually send these things off to be prototyped for not too much money, and of course volume production is even cheaper. It's just that the startup cost is high--licensing from Transmeta, and expensive proprietary packages to manipulate the designs if you want to customize them.

It's rather ironic that Linus is associated with a company that throws up so many barriers to hackers. And I'm saying this as someone who owns Transmeta shares and is disgusted with the way this is handled, but I'm just like the guy on teh commercial who "owns Nike". My stake is so small that nobody would listen to me. So I vent this stuff on Yahoo's finance board, and sometimes here.

Re:Two Places To Look

[TPFH]

I remember durring California's rolling blackouts, weren't there some people blaming the internet; ie increased computer usage (instead of fraud) back then? Anyway, I remember thinking about the Crusoe chips and maybe it would make sense to build low power consuming desktop PCs.

And you don't need power companies shutting down the power stations for fraud to motivate this. It is about time we start thinking of efficiancy in all walks of life. Yea, there are people that will actually use the power of a 2ghtz PC, but even those people might want to have a fileserver running ML Donkey 24 hours a day without turning their computer room into a sauna.

I just got on the p2p bandwagon this summer (yea, I'm late but whatever) and my computer room is in the basement and normally cool year round. This summer it was between 80-90 degrees most of the time.

I imagine the cost for the motherboards is in that they are not being mass produced. Well, maybe Transmeta ought to think about marketing these at companies and home users looking to reduce their electric bills.

Question: If you were to do parallel processing or clustering with transmeta chips would you get more bang for your electrical buck?

Re:Slashdot editors are on crack (what else is new

[cpeterso]

With that much non-cash in the bank, Sun will be sitting pretty for a long time!

Any idea how much the NanoBGA EDEN-Ns will cost?

[WoTG]

Will they be higher priced than their current chips (a "size" premium) or will they be even lower priced (because they cost less to manufacture - probably)? Unfortunately, it seems Via's site has been /.'d...

PowerPPC and x86 instruction sets

[Anonymous Coward]

somebody should hackup some PowerPPC code morphing software for this chip. That way you have a cheapo mac clone. I wonder if it's possible to code morph mulitple insutruction sets and run multiple OS at the same time! that way you can switch between windows, linux and MacOS.

Since we're in the ryhming mood, why not try...

[Kashif Shaikh]

Decepticon!

Re:trasmeta processor plus mobo - yes, soon

[Anonymous Coward]

look at transmeta's partner list. some of them are whitebox and OEM motherboard manufacturers likely to produce something MicroATX or PC104.

transmeta sells a very expensive MicroATX "dev kit" themselves featuring the old TM5800 crusoe. but thats not what you want.

intel bashing due because...

[Anonymous Coward]

Intel never would have produced their Pentium 3 ULV and modern Pentium M ULV + LV (banias, the cpu in centrino) parts if it hadn't been for transmeta.

there was absolutely no competition or reason for them to produce such a low-wattage thing because there were no viable alternatives. Even if they had the price would've remained at their usual high monopoly levels rather than falling rapidly towards good commodity prices.

Re:Slashdot editors are on crack (what else is new

[nelsonal]

Accounting rules are goofy sometimes, but it's the language of business and it does bear learning. They aren't inconsistent, but there are occasionally loopholes, they're fairly logical (once you grasp the basic logic) and were one of the first applications of computers in business. In this case, the change dealt with a tax rebate that the US government (it could be a foreign gov't but let's keep it simple) worth something between $1 an $2 billion in taxes that SUN won't have to pay due to losses they already recorded (the government lets you swing losses around a short year period as a benefit to highly cyclical businesses). Think of it like catch up deductions. The rules of accounting require that management estimate the likelihood of actually earning what would be required to meet the tax write off and then apply that discount to the value of the tax benefit. SUN's mangment decided that the likelihood of reaching that level of profitability in the required period of time was less than 50% and has to reduce the value of that potential benefit to $0. They still have the benefit, but the value of it that they report to their owners was lowered. This is a pretty arcane subject and probably still confuses many of the investment professionals that are supposed to understand how it works (The accountant's, of course, will call this child's play. They save their anger for things like pensions or derivatives). If something wasn't clear, I'd be happy to try again with more numbers or a diagram.

your hunch is wrong

[Anonymous Coward]

Intel and Microsoft control the horizonal and vertical. Intel makes money on new CPU sales. Microsoft makes the bulk of its money in new computer bundled windows licenses.

The both have full incentive to design, develop and shove down your throats software, OSes and systems that demand significantly more CPU power so that the three-year PC upgrade cycle is maintained. Without that both companies would cease to be profitable.

Windows XP executes two-times as much code during system boot and many normal OS win32 api tasks as Windows 2000. Is it faster? yes, on modern machines. but not on the same old hardware that ran 2000.

you don't even wanna know the shit they're putting into longhorn...

Nice Trend

[Dr.Seltsam]

As stated before, Transmeta wa sone of the first companies that bring the low power processors from the embedded market to the desktop market. And, having a close look towards energy reserves and enviromental stuff it seems a good idea to go on with this kind of inventions.
Besides, no one really needs those 2+ GHz, 70 Watt monsters to type the company letters, right ?

Cautionary note

[Space cowboy]

Via's Nehemiah core (what a name!) is not as efficient as the modern x86 cores, so running an Eden at 1GHz is approximately the same as a P3 at 6-800 MHz, if you could buy such a thing.

There are reviews (envynews for example, against an Athlon 1900) which show the cpus at 1104:4696 for example, makeing the Nehemiah roughly the equivalent of an Athlon 450 (!)

Now, the CPU has other things which make up for it, hardware-assisted mpeg-2 playback etc, so it can playback your VOB's and DiVX's even with its weedy cpu, but don't expect this one to be too powerful, it's normally rated from 'ok' to 'disappointing' in the business benchmarks.

I'm getting one anyway - I intend to make a *quiet* PVR (Myth-TV) with network streaming (ffserver) mpegs and write-DVD (cdrecord) functionality that I can sit in the living-room, but know your machine before you get too impressed with the 7watt/1GHz hype ...

Simon

Re:Cautionary note

[TeknoHog]

Via's Nehemiah core (what a name!) is not as efficient as the modern x86 cores, so running an Eden at 1GHz is approximately the same as a P3 at 6-800 MHz, if you could buy such a thing.

It's the C3 processor that now has the Nehemiah core. Eden is a different CPU altogether, and AFAIK less powerful per MHz than the C3. (To be precise, the Nehemiah version is called C3-2 to separate it from the older C3 with the Ezra core.)

Now, the CPU has other things which make up for it, hardware-assisted mpeg-2 playback

Via's EPIA motherboards have a _chipset_ that does hardware MPEG2 decoding. It's not in the CPU.

Re:Cautionary note

[bluGill]

Yeah VIA has hardware to assist in a lot of things you are likely to do, but can you use it? Last I checked the MPEG decoder was not supported by any program running on linux, so you have to run windows. Might not be a problem for you, but it is for me. (Not so much that I hate windows as I like to do everything myself, and that might one day mean I play with the source code to something)

I too have ideas for what I would do with VIA (mini-itx) motherboards. Fortunatly for me, mpeg isn't of interest, but other abilities are.

Re:Cautionary note

[Space cowboy]

There have been steps in the right direction, though I'm not sure it's fully open yet. You can certainly get xine to play mpeg2's, but you may have to use the via-binary version.

There has been a reverse-engineering of the binary-only module though, which I think is waiting to be integrated into the main applications.

Simon

Re:Cautionary note

[bhtooefr]

a P3 at 6-800 MHz, if you could buy such a thing.

P3-600 (all models): http://tinyurl.com/r1nr
P3-650 (all models): http://tinyurl.com/r1nz
P3-667 (all models): http://tinyurl.com/r1oa
P3-700 (all models): http://tinyurl.com/r1oh
P3-733 (all models): http://tinyurl.com/r1oo
P3-750 (all models): http://tinyurl.com/r1or
P3-800 (all models): http://tinyurl.com/r1p9

I'd say you could buy such a thing (yes, I used TinyURL addresses, but the Intel addresses are ~140 chars). Of course, I'd rather have a 1.4GHz P3, but that's just me...

Speed?

[pesc]

So we have chips running at 1.1GHz, 553 MHz, 800MHz or 533MHz. And almost all people seem to think that these numbers are some kind of speed indicators, even at slashdot. Jesus!

How about a link to some actual benchmark figures? Have anyone seen any? I just looked at SPEC, but there was nothing there. What is the speed of these chips, really?

As a side note, I believe that the really interesting news is that the new EDEN chips use a nanoBGA packaging and are to be used in the new nanoITX form factor which is smaller than miniITX. At 120x120mm, the new motherboard is just slightly smaller than a CD enclosure!

Re:transmeta processor plus mobo

[mocm]

Have a look at this forum and the link to the touchscreen driver.

moving parts

[TeknoHog]

If you add a ide->compactflash converter and use one of their fanless 55w psu's you've got a machine with *no* moving parts.

BZZT! Wrong! It will still have moving electrons. :-)

Besides, switching power supplies sometimes make high-pitched noise (at the switching frequency). But well designed systems have this frequency outside the audible range.

Re:moving parts

[Mattsson]

Hehe... Forgot about those. =)
And also, we have the keys on the keyboard and the mouse. ^_^
And the screen moves photons too...

Via's Energy Efficient Chips... not powerful

[Nazmun]

The only way via seems competitive is with their somewhat misleading (and probably biased benchmarks) based on poweroutput-processor usage. Unfortunately in reality their 1 ghz processor performs around the speed of a 750 mhz celeron. Via covers that up somewhat for multimedia by having a lot of onboard decoding software for mpeg standards.

Let's hope Intel does catch up

[hankaholic]

Let's hope they can capitalize on this before Intel starts filling the same niche.

That's got to be in the running for the stupidest thing posted to the homepage, EVER.

Why would anyone hope that Intel doesn't produce low-power chips at reasonable clock speeds? Cooling, efficiency, product life (less overheating and temperature cycling can't be a bad thing)... and this guy hopes Intel doesn't catch up?

Transmeta hired Linus at one point. That's no reason to hope that Intel doesn't design a really low-power, high-performance chip that'll reduce power consumption of both desktop and mobile PCs while removing the need to have tons of (loud!) case fans.

Re:Let's hope Intel does catch up

[bluGill]

Markets are interesting things. I want Intel to compete in the low power market because it lowers prices for everyone. (enviormentalists will also point out that a lower power CPU would be a good thing in general) However if VIA and Transmeta cannot get sales before the giant Intel crushes them with what they can do, in the end prices end up higher because there is less compitition.

Re:Let's hope Intel does catch up

[hankaholic]

if VIA and Transmeta cannot get sales before the giant Intel crushes them

That's a verrrrry big if.

Transmeta is going after a market that Intel doesn't really seem to care about. Examine the trend in power consumption from the Pentium through the P4. Intel doesn't seem to be showing much of a general interest in lowering power consumption.

It would take a lot for Intel to start producing ultra-low-power chips. I really would be surprised if they decided to seriously consider trying to compete in that market unless they saw somebody else making a significant profit first.

This all leads to the stupidity of the statement which I had quoted earlier -- "Let's hope they can capitalize on this before Intel starts filling the same niche." Not only is it unlikely that Intel will attempt to beat them to the ultra-low-power ~1GHz punch, but to suggest that we should hope that Intel doesn't end up "filling the same niche" is suggesting that we should hope that the most widely known chip manufacturer doesn't produce cool, efficient processors.

You suggest that without competition, prices will be higher. If Intel doesn't attempt to target the same markets as Transmeta, doesn't that mean that those markets bear less competition?

Compatibility (Re:I don't get it)

[salimma]

Transmeta wants to reserve the freedom to change their internal instruction set at any time, without breaking anything.

They have already done it, in fact - the original TM3200/5400/5600/5800 have a 64-bit VLIW instruction set, the new Efficeon TM8000 has a new 128-bit one.

Parent information is stale

[Anonymous Coward]

You're thinking of the FPU.

VIA chips that support "3DNow!" instructions instead of SSE instructions also run floating-point instructions at half the clock rate of the CPU.

VIA chips that support SSE instructions run floating-point instructions at the CPU clock rate. The "Nehemiah" CPU is the first of these. The Eden-N is a descendant of Nehemiah.

Here is VIA's press release on the Eden-N.

Benchmarks are the tool of the Devil, but here are a set from a review of the VIA EPIA M1000 motherboard.

Here is a review of subjective use of a 1 GHz Nehemiah.

padlock...

[andrew71]

hmm... padlock... VIA being one of these

is it me or something's rotten around here?

Re:Green destiny - Green My Desktop?

[Cycloid Torus]

If 10 of these were mounted on a backplane, what would be the theoretical net speed in comparison to a single CPU?

Could I start with 2 and work up?

How much energy (fans, power supply losses, etc) would I save if I could do this?

My HD and CD drives are about 25w and 10 on a backplane would bring me in around 100w or so. Or am I missing something important?

Re:fp, yo ---Battery Ran Down

[lcsjk]

I had written a long reply to tell you why the faster will always have a greater demand than the low power and somewhat slower chips, but my battery ran down before I could get it sent.

I still fail to get it

[Pan T. Hose]

Transmeta wants to reserve the freedom [sic] to change their internal instruction set at any time, without breaking anything.

What about our freedom then? We're not even allowed to use the very instruction set mentioned above, much less to break anything. I wouldn't expect something like that from Linus Torvalds who may encourage boycotting the FSF but still has done a lot for free software per se (I mean the technical rather then political aspects of his work).

They have already done it, in fact - the original TM3200/5400/5600/5800 have a 64-bit VLIW instruction set, the new Efficeon TM8000 has a new 128-bit one.

In the world of free software there is little need for binary compatibility, as long as there's source compatibility. What they should do is to provide appropriate gcc backends for their architectures and everyone would have the best of both worlds. I hope they are reading this.

Re:I still fail to get it

[salimma]

What they should do is to provide appropriate gcc backends for their architectures and everyone would have the best of both worlds

You might want to ask GCC developers whether they have the time to develop for yet another target :) Besides, with Intel developing a similar VLIW architecture (IA-64) you could forgive Transmeta for trying to keep certain cards close to its chest...

so where is my cheap but slow long-battery laptop?

[ukyoCE]

I've been interested in getting a laptop for a while, but nothing i've seen in the market interests me. I don't need something fast - it's not like i'm going to be playing Doom 3 on it, just web browsing and text files and MP3s and maybe DivX at the most. So it can be pretty slow, 500mhz-1ghz at the very most. But I want the battery life to actually last a significant amount of time. Not this 2-3 hours crap. I'd also like it to be small and portable...not as small as a PDA, but just something with a maybe 10-12" screen at the most (which would also help keep battery consumption low). And of course lastly, I want this thing cheap, like 200-400$. When I was looking at laptops recently the cheapest was 1000+$ for a laptop including a CDRW drive.

Of course I could get an old laptop on ebay for fairly cheap and slow, but the battery life will suck just as much as new laptops - what I want is a new laptop running slower processors using modern technology to reduce power consumption.

I'd think a lot of people would be interested in a low-cost laptop that isn't meant to be a desktop replacement and doesn't need to be constantly plugged into a wall to use .

Re:Slashdot editors are on crack (what else is new

[bhtooefr]

I thought that HT SIMULATED dual cores, whereas this technology will have the equivalent of 16 CPUs in one package. Take 16 Pentium 4 w/o HTs, put them in one Socket 478 package instead of 16, and you have something like Sun's technology.

Re:transmeta processor plus mobo

[frank_adrian314159]

Many thanks!

Benchmarks?

[Pan T. Hose]

Actually, Linus tried porting Linux to the Transmeta, and a Transmeta-optimized x86 compile was faster than a native Transmeta compile.

Where have you got those benchmarks from?

Re:so where is my cheap but slow long-battery lapt

[SmartCookie]

Here ya go. Fujitsu has a line of notebooks that should be up your alley. With the expansion bay occupied by a second battery you can get 9-11 hours battery life (5-6 hours with one battery).

The problem is ...

[Anonymous Coward]

I've thought the same thing. The Transmeta chips would be a great fit for a PDA. Fast, low power, floating point. The problem is cost. Transmeta is trying to sell their chips for roughly the $100 range. This price just won't work for the PDA market where $10 a chip is more acceptable.

Re:so where is my cheap but slow long-battery lapt

[ukyoCE]

Thanks for the link. While those are interesting, the price is still $1200+. Maybe what I'm dreaming of just isn't possible, and that's why no one has made it.