Transmeta's New Smaller, Faster Chips Announced
billstewart writes "Transmeta announced their new 5900 and 5700 CPUs. They're 50% smaller than the 5800, intended for low-power, low-heat, high-speed applications, and contain an integrated Northbridge. They're sampling now, production in January 2004, and expect to have a mini-ITX board out in 1Q04. The core chip is a 128-bit VLIW hidden by x86 emulation (as opposed to their new Efficeon, which is 256-bit VLIW.) The difference between the 5900 and 5700 seems to be L2 cache size.
There are several other stories on Google News."
Comment removed based on user account deletion
What do they use those chips for? Microwaves and stuff? Toaster ovens?
(-1, Raw and Uncut is the only way to read)
I've always ranted here about how we could use an industry standard chassis and AC/DC power spec for mini-ITX. If LCD monitor vendors could simply stick their panels into an open spec laptop chassis, we'd have oodles of cheap, interchangable laptops out there. And they wouldn't cost $900 to fix when you spill your free beer on them...
Life is the leading cause of death in America.
robots ?
i suck at teh html. LINK
People who run home servers and get reamed on electric bills.
-
ping -f 255.255.255.255 # if only
Say what you want, but these people have found a niche and deserve credit.
Their CPUs are sufficient for most tasks and not seldom run three to four times as long as comparable CPUs with the same amount of power. They are the equivalent to the 'kaizend' motors in the late generation portable cassette players ('walkmen'), seriously optimized for a specific goal: to consume as much minimum power as possible.
My friend has a Fujitsu Lifebook P with a 900 Mhz transmeta and it runs 16 hrs of the grid! And he even watches DVDs with it. Try that with a Pentium Mobile.
We suffer more in our imagination than in reality. - Seneca
not to respond to your non-troll but, this is being posted on a laptop using a transmeta TM5800 at 876mhz.....
i hope that these new chips fit in the old slots. it would be a nice upgrade for my laptop......
All of the above was encrypted with a Quad ROT-13 method. Unauthorized decryption is in violation of the DMCA.
There's at least one Transmeta powered Tablet PC. And I think Fujitsu sold an ultraportable laptop with a Transmeta CPU.
Granted, I haven't checked out the market for a bit, since I've pretty much gone "console only", and the only PC games I play anymore I just wait until they hit "OS X" - or do without. (Not that I don't have an oversized old games library as it is - I don't need to buy anymore....)
But I have friends who do LAN parties, and I've wondered about getting a Shuttle kind of machine, or preferably something the size of a Cappachino computer. Small, slip it into a backpack, show up with just that and a flat screen (keyboard, mouse, etc) - but it would be a small machine just for PC LAN gaming. It wouldn't need a huge video card - anything that can run most games published 2003 at 800x600 would be fine.
I wonder if these Transmeta chips could be used this way.
52 Weeks, 52 Religions with John Hummel
I care. I'm looking for something to replace my Athlon/1Ghz Linux box. It has done fairly well, but I setup something that has far more horsepower than my little website requires. I'm sure it's a waste of energy and I'd like to find something that fits in a small case, uses comparatively little power, and will work with RH. I'm sure I'm not alone...and so far my research has come up with fairly wasteful systems.
Could a low end Intel-based system do it? Maybe, but I'm actually interested in a lower power system more than initial cost.
From the article:
"The Crusoe TM5700/TM5900 processors are another significant step in advancing the cause of efficient computing," said Dr. Matthew R. Perry, president and CEO of Transmeta. "By delivering a solution that is 50 percent smaller than our existing Crusoe TM5800 processors, Transmeta allows system designers to further leverage the high performance and low-heat dissipation characteristics of Transmeta's proven hardware and software architecture for a wide range of new smaller form factor, fan-less designs."
Important tidbit not in the article, but needed to be:
Dr. Perry then proceeded to explain the seemingly confusing numbering scheme, "Well, since we had cut down the form factor some of thought we should also cut the model number down. But, we didn't want to alienate those who are used to seeing newer products with higher model numbers, so we compromised and named it higher and lower than its predecessor."
---
Two fish swim into a wall, one turns to the other and says, "Dam".
So, how do these Transmeta chips compare to the VIA C3's, in terms of computing performance, and power/heat requirements?
VIA has been doing a very nice job with the C3, with several varieties, speeds, and sizes to be used in all sorts of commercial or hobbyist applications. They have a new mini-itx board, with dual ethernet ports for network gateway usage. And, their new C3 processor includes hardware AES support, with incredible performance for network or filesystem encryption.
It would be great to have an alternative. The TM chips seem to have some really interesting features. But, I have not seen any of these boards/chips available retail. They seem to be essentially OEM solutions for embedded devices. This positioning puts them head to head with many excellent non-x86 solutions, like the ARM, PowerPC, and Hitachi SH processors.
Imagine having a cheap, low power, fanless, quiet computer, running a variety of convenient things for a home network. You know, DNS, HTTP caching, file serving, email, the works. Put this in a small and attractive case, pop in a processor that really is quite fast, and you have something worth drooling over.
Summing up the important issues about Transmeta chips (I'm posting this from my Fujitsu P Transmeta-based laptop),
a) Transmeta's biggest problem is the lack of speed. It runs most productivity software and normal browsing (not Mozilla - Opera and IE are fine) at comfortable speeds. Don't try to run it as a J2EE server or something like that, though.
b) Transmeta's biggest advantage is the battery life. As another poster mentioned, I regularly get 10 hours from my battery, and that's *real life*, not some artificial benchmark.
In sum, it's the best laptop chip ever if you don't have more than moderate speed needs. Perfect for the casual user - and for people doing lightweight HTML/CSS development, like me.
"It is, I hate to admit it. I've found that I don't get very excited over low power, lower everything CPUs for mobile use.
Give me that socket sucking power of my P4 any day.
Does anyone else feel that we are pass due for another speed revolution?"
No. In fact we have been long overdue for a plateau (relatively) such as we are in now. It has allowed alternatives to Intel to be taken seriously
More importantly, the current stall in processors speeds will mostly likely lead to more efficient software, particularly from Microsoft, who tends to rely on Intel's huge speed jumps to justify more bloated version of Windows (not that I use it).
Finally, it's hard to find a PC these days with no moving parts (fans) and I've decided, if at all possible all my future PCs will be either fanless or at least passively cooled under normal use conditions. I don't like the noise, the added RF interference, or the ultimate need for repair when the fan bearings wear out. (Yeah, I know disk drives have moving parts, but they are generally easier to replace, and I think solid state disks are a ways off yet).
I never understood coding outside. The sun makes it hard to see a screen, so you'd have to find a perfect shady spot, then you'd have to go and find a table and chair comfortable enough to not get sore from coding for several hours.
Just leave work at work and enjoy doing outside things outside.
no?
-- taking over the world, we are.
link (has pictures too): http://www.kurnspatrick.com/sharpmm2.htm
t re v237.htm
Ubiq Computing from Akiba Hotline wrote a review on the Sharp PC-MM2-5NE a couple days ago (unfortunately in japanese ):
http://pc.watch.impress.co.jp/docs/2003/1209/ho
(use a translator service e.g. world.altavista.net or any others)
Some notes from the review:
They used a PowerPoint 2002 file at 4.02MB and timed opening times
Model 1st time 2nd time 3rd time
Efficeon 28.04 18.95 18.78
TM8600 1GHz
(PC-MM2-5NE)
Crusoe 51.91 29.44 29.78
TM5800 1GHz
(PC-MM1-H5W)
Full starting time with the MM2: 43.70 seconds
With the MM1 (Crusoe): 58.25 seconds
(I'm assuming this is Windows XP Home with nothing tweaked, and the same application setup - it would be silly to compare startup times with different application suites in start-up.)
The 256MB on board appears to be a permanent fixture and not upgradable, in the goal of making the MM2 as light as possible.
The hard drive model is a Hitachi (HGST) DK14FA-20; 20GB, 1.8".
That "MOBILE switch" I mentioned earlier, according to the review, can increase the battery time up to 40%. It would be interesting to see actual numbers with this (40%? that's a lot).
The keyboard is 17mm pitch, 1.7mm stroke. Compare this to the P-2k series with 17mm pitch and 2mm key stroke. This means the Sharp keyboard will be shallower. The LCD doesn't have a latch, so I assume it'll be like the P-5k and Sony models with a spring-loaded screen.
At a brightness level 3 (whatever that is) and using both Office XP (whichever apps, I don't know) and Netscape Communicator (and "etc." - whatever that means!) - the review managed to get 2.4 hours off the standard battery. The standard battery is 19.98Wh with 11.1V/1.8A. They mentioned in passing that the Sony 505 Extreme (X505) got 2.8 hours, but that's not a good comparison since that battery is 22.2Wh. The Sharp model is also much cheaper than the Sony.
They also have an MPEG movie, but my download was corrupted or something (though I suspect it's some kind of powerpoint presentation in Japanese so we wouldn't be able to understand anyway).
Another thing - the TM8600 supports AGP4x, but the AGP operates at 2x on the Sharp with ATI mobility radeon; the article cites Transmeta as saying that the AGP bus with Efficeon isn't entirely stable. I'm not sure what to make of this, but surely it'll be better than the video in the P-2k.
I'd like to see a desktop system built with maybe 8 of these running in SMP. You'd probably have about the same raw computing power as a high-end Intel or AMD dual-processor machine, and probably less power consumption. Where you'd really win is with usability and interactivity - a good SMP OS would handle multitasking properly among the CPUs. Your web browser would never interrupt your mp3 player again, and the UI would be unhindered by background processes. This may especially be the case with the on-die memory controllers.
The only problem being the fact that they could never sell it... only high-end server versions of Windows support high numbers of SMP CPUs. Obviously this isn't a problem for Linux users.
-3Suns
~~~~
The Revolution will be Slashdotted
I remember when they first came out with their crusoe chips they were marketting them as viable alternatives to pentiums and k6's (or a k6 variant). The problem was that their performance loss couldn't justify the battery life increase and so few manufacturers took the risk to built laptops with them or market them as heavily as their pentium laptops.
I'm surprised transmeta lasted this long and so I guess that's an indicator that they weren't dot com vaporware. However, I hope to see this time they try to market them not as laptop replacements but just really fast chips for embeded applications or portable devices. Battery life is a very big consideration in designing mp3 players, cell phones, cameras, etc. What this may bring soon is smaller devices that rely on less chips since they can take advantage of transmetas more powerful chip than what it's replacing. If not, it could simply allow more features in handhelds that already exist instead of trying to invent new markets (tablet's to some extent).
In 1998, some engineers at Corel took 10 StrongARMs and connected them on a custom backplane, made a couple of modifications to Apache, and were able to dish out close to one million web pages per minute.
I'd love to see someone put 8 of these on a board with a gig of memory, and two ethernet jacks. One would go to the network, the other would go to your file server/SAN/NAS/other_buzzword.
Put 2 gigs of memory on it for disk caching, and for a pretty low amount of money and electrical power, you could dish out VERY large numbers of web pages.
Shoot, take it farther: Have another unit based on them that runs LVS as a load-balancer, and put several of the servers behind it. All of the sudden, for $2000, you'd have the capability to dish out a billion web pages per day (or more), with load-balancing and realtime failover to boot!
steve
Oh, you're not stuck, you're just unable to let go of the onion rings.
You make that sound funny [and it sorta is] but think about this. If fax machines start having say >500MIPs [or whatever the 5900 offers] then we can put better compression algos in there and more flexible protocols, etc...
Tom
Someday, I'll have a real sig.
I have a dell inspiron 366 MHz PII with the dual battery packs. It weighs over 10 pounds, and it has 155 WHr of capacity total. The battery lasts eight hours.
I love to code at coffeeshops, and in fact, took a year's sabbatical where I did most of my coding in coffeeshops. In the four years of owning this computer (it shipped with win98), only once did I ever run down the batteries. Eight hours of coding and I was pretty beat - six hours of creative thought was all I was good for at a time. I hope you can survive to 12, but you might want to try it before buy a machine to that spec.
(p.s. I love that machine, but it was too heavy for anything but a backpack)
HIV Crosses Species Barrier... into Muppets
I know that the idea of a chip which runs java natively has been bandied around already, but I've always wondered why Transmeta hasn't released other architectures under their code morphing software, specifically java. All the arguments I've heard against a java machine have been due to the fact that java is more than just a series of byte codes, it's also an api. It seems to me that a combination of a crusoe chip, the right code morphing software, and the equivalent of JNode as an OS would allow for some fast and efficient java machines. Is this possible?
First, Ditzel wanted to do a fast VLIW--the great wide hope--faster than Intel. It didn't quite work out but someone at Transmeta lucked into the low power idea. Great idea. It took awhile but with enough perserverance and capital they made it work.
But at the end of the day, they get to compete with Intel. This is sort of like winning a bunch of thumb-wrestling contests and, as first prize, getting to go a few rounds with Mike Tyson. Intel doesn't play nice, has a multiple ear appetite, *deep* pockets and can out-manufacture anyone.
I wish I could buy a Crusoe; I really think as an idea it rocks. But life has slapped enough sense into me to be skeptical. They have less than a year of money left. But someone, AMD?, will buy them.
Ditzel reminds me of the bad guys in Bond movies. Instead of killing Bond when they have a chance, they have to tie him up with a beautiful girl and leave the room. Ditzel lacks execution skills.