Efficeon/TM8000 RELEASE etc.

Printed from: LeoG.net Ultra-Portables Forum
Topic URL: http://www.leog.net/fujp_forum/topic.asp?TOPIC_ID=3710
Printed on: 05/31/2010

Topic:

Topic author: oion
Subject: Efficeon/TM8000 RELEASE etc.
Posted on: 10/01/2003 10:59:47
Message:

Transmeta posted "news" on their site today (hahaha) about the various industry companies now affiliated with them. Absolutely no real news, however, besides the (same old) blurb that the Efficeon will be discussed at the Microprocessor Forum in San Jose in October 13-16 of this year. But we already knew that. Honestly, I was expecting a bit more excitement from them if they were shipping in bulk to OEMs supposedly last month - at least some company announcements. Perhaps the companies are waiting for the MForum. Or, pessimistically, Transmeta is running into production problems yet again.

http://investor.transmeta.com/news/20031001-119083.cfm

Meanwhile, the Fujitsu company is reorganized and launched a new site, too:

http://www.computers.us.fujitsu.com/www/news.shtml?aboutus/pressreleases/pr_100103

What this might mean for their product lines/customer service, I don't know.

There have been a few posts asking about new P-1k/2k iterations; my opinion is that it's extremely doubtful now. The TM5800 is being phased out and no new iterations are being made for it.

**Re:Die size etc. - see later posts**
Even though the Fujitsu P-series has been relatively popular, one must remember that the subnotebook/ultraportable market still only takes up a smaller percentage of all laptop sales (in the US); it's a particularly risky market when a number of people don't have easy physical access to specific brands in-store. Any rise in price for a Transmeta-based notebook will likely kill its already struggling sales under the Intel cloud. My prediction, off-base or not, is that the P-2k series will be completely phased out forever, while the P-1k series may be retained due to its more unique touchscreen feature. We'll see... Other people have other comments?

I suppose at this point I'd say - if you want a P-1k/2k or Transmeta laptop, buy it now. But who knows, I could be wrong; perhaps the TM8000 will blow us away. Still, in technology industries - "better late than never" really doesn't cut it. The Pentium-M already has a huge chunk of the subnotes. If Transmeta had released Efficeon in June/July, things might look different.


==============ADDENDUM==============[10/18/03]

I've put together a small page with portions of the webcast that detail what we really want to know - the benchmarks. (Note that you can download the full PDF webcast presentation -i.e. all the slides- from Transmeta's news page. Something like 75 pages, over 7 megs.)

http://www.oion.net/qnd/tm8000/

Replies:

Reply author: blehmann
Replied on: 10/01/2003 15:02:14
Message:

Hi Oion:

As always, you found the information first and provided an insightful analysis of it. The only change I would make concerns your recommendation to would be p1/p2 buyers: they might as well wait a couple of weeks to find out for sure what is up. That said, I am disheartened about this non-announcement at a time when tm would be building toward a crescendo of ultralight laptop announcements in two weeks if things were going well.

bnl

Reply author: oion
Replied on: 10/01/2003 16:19:27
Message:

Good point - if Fujitsu and other stores end up clearing out P-1k/2k inventory, they may even slash prices (considering another P-5k iteration is on the horizon). Something to think about...



______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: gellul
Replied on: 10/01/2003 16:28:09
Message:

Oion, This is a very good post with a lot of insightfulness. I hope the T8000 performance is high and the price low. Maybe it will have a chance in the marketplace. Maybe Fujitsu will use the product.

Reply author: mgglim
Replied on: 10/01/2003 22:05:30
Message:

Oion, good to see this post from you. And blehmann,I share your disappointment with Transmeta's pace.

Transmeta seems to have learned from the past, acting less like a tech geek and more with aplomb as a tech marketer. And since they've been burned before with their unchecked enthusiasm for their product, ala newton-like, they would now rather surprise us this time on the positive side.

You can hear the Transmeta cheerleader in me once again. There's only 13 days left before their official launch date. To date, they have kept everything they don't want out still under wraps. It's important to notice what you don't see or hear, as this is atypical of the industry- Apple with enthusiast website leaks, AMD with the leaks of their marketing plans bordering on the oops variety. This to me looks like Transmeta runs a tight ship, and that to me is good.

Also significant is what you are seeing on the other side- Intel. There is a lot of noise, a lof of money spent on ads- on Centrino, on Centrino Day last Sept 25 - the effort of making just another sleepy day into a splashy event. Plus the incentives they dangle to makes OEMs go fully onboard with their chip. Motorola was like that a decade ago, when they were king of cellphones. Nothing they could do in marketing could surmount the technical superiority of qualcomm's cdma nor the feature sets of Nokia. Now Motorola is a has-been. While Intel is reacting in a more paranoid manner, it still is reacting to a threat that Ditzel (Transmeta's founder) makes more real with each day approaching the Efficeon launch. Bear in mind, Intel with its huge marketing budget, is making us think they invented efficient computing for portable wireless use. It may get away with it. I hope not.

In reaction to Oion's comment on the implications of a large die size, I agree that the cost will go higher, but if performance as well as efficiency increases tremendously, that would take them to the desktop replacement market, which seems to be the preference anyway of Americans. And this is what matters, since the U.S. market holds great sway.

We may not see the Efficeon is the beloved form factor of the P series, but I am holding hope that whatever bulk it takes, it will not become a blumbering mastodon.

At the very least, I expect to not see a notebook that has a touchpad because it's needed to make an excuse for a large battery (as in Centrino) and give us the option of an IBM trackpoint because it can and will do the job.

Reply author: blehmann
Replied on: 10/01/2003 22:44:04
Message:

The devil must be given his (or her) due and so Intel should be credited with producing a Pentium M that is so much better than the p4. In doing so, Intel did not set a trivial bar for TM to clear. I have been a cheerleader too and I remember convincing myself that Apple's ten year lead time on windows and Motorola's two year lead time on Intel would make for a clean sweep for the Macintosh world. It was a clean sweep but the Mac ended up (well not quite) in the dustbin. Would that it were otherwise.

bnl

Reply author: oion
Replied on: 10/02/2003 11:08:08
Message:

quote:

---

Originally posted by mgglim

You can hear the Transmeta cheerleader in me once again. There's only 13 days left before their official launch date. To date, they have kept everything they don't want out still under wraps. It's important to notice what you don't see or hear, as this is atypical of the industry- Apple with enthusiast website leaks, AMD with the leaks of their marketing plans bordering on the oops variety. This to me looks like Transmeta runs a tight ship, and that to me is good.

---

quote:

---

While Intel is reacting in a more paranoid manner, it still is reacting to a threat that Ditzel (Transmeta's founder) makes more real with each day approaching the Efficeon launch. Bear in mind, Intel with its huge marketing budget, is making us think they invented efficient computing for portable wireless use. It may get away with it. I hope not.

---

quote:

---

In reaction to Oion's comment on the implications of a large die size, I agree that the cost will go higher, but if performance as well as efficiency increases tremendously, that would take them to the desktop replacement market, which seems to be the preference anyway of Americans. And this is what matters, since the U.S. market holds great sway.

We may not see the Efficeon is the beloved form factor of the P series, but I am holding hope that whatever bulk it takes, it will not become a blumbering mastodon.

---

Reply author: blehmann
Replied on: 10/02/2003 12:43:32
Message:

wrt product announcements, I think Oion has gotten it right. TM repeatedly announces non-laptop sign-ups by manufacturers as it did in the recent press release. There is a clear absence of laptop announcements in this release and the one a month or so ago. However, this is just so much (educated, I hope) reading of tea leaves and we should have a much clearer idea in a couple of weeks.

bnl

Reply author: mgglim
Replied on: 10/04/2003 03:04:04
Message:

I agree, the absence of designs wins is disurbing. But I would rather that the cause of my anxiety be that Intel is a feared supplier, ready to wield the stick to any OEM that dares support its enemy, than the thought of Intel having technologically mastered Transmeta in its game causing no TM8000 takeup with the OEMS.

Puting myself as a product manager faced with the perceived threat of Intel withholding supply of their erstwhile ascendant Centrino chips, I would hate to lose my job because I made a strategic blunder by selecting Transmeta- one that is struggling, begging for scraps at my table and thus incapable of wielding its stick.

If it turns out Transmeta has a superior product, I can easily backtrack from Intel and Transmeta would still gladly supply me its chips. It cannot punish me because it has no choice - it has to build volume and achieve scale and profitability with it. Furthermore, it needs all the friends it can get, as it may have won the battle, but it has not won the war. Intel will keep stepping up the pressure until it reaches parity or surpasses Transmeta. This reminds me of what I read somewhere "In a cat and mouse game, the cat only has to win once." Intel the Cat and Transmeta the Mouse.

Thus, the OEMs are going to back Intel out of necessity. If Intel cuts me off because I wasn't loyal to it, imagine being cut off from Centrino chips and losing the laptop market. If i were Fujitsu, I will think really hard.

The most promising customers for Transmeta thus will be the second- or third-tier notebook manufacturers who have no market share to lose and thus the most to gain. Luckily, there is no shortage of scrappy notebook makers - China, Taiwan, and Korea have mfrs. that will have a field day taking the market by storm with the TM8000. That is, if Transmeta will deliver the performance.

In the next months, maybe this group will be reviewing a lot of these notebooks, eh?

Reply author: werner_jo
Replied on: 10/04/2003 12:20:54
Message:

Transmetas Efficeon:
http://www.transmeta.com/efficeon/efficeon_faq.html

Transmeta is not disclosing the die size or clock speed of the Transmeta Efficeon Processor at this time.

Yes the die size will be bigger (1 MB L2 cache) but this will have no effect on the TDP. The die size is measured by mm², for example the the 0.13 ìmNorthwood PIV has 131.4mm². That is not size of the ceramic or metal plated chip which is fixed on the socket. So it will be possible to build a P2000 sucsessor with an efficion (together with an ATI or Nvidea GPU with 4*AGP) using the same place and cooling system like crusoe.

My speculative prediction: a 1.2 GHz Efficeon (TDP 8W, 2.1 times higher performance than current crusoe systems), DDR333 Ram, Geforce4go in a subnotebook will be anounced in oktober. Maybe a lifebook but i think it will be an other company.

Reply author: oion
Replied on: 10/04/2003 12:33:30
Message:

That would be great; I like the P-2k form-factor so much over the P-5k that I can't bring myself to upgrade even if the P-5k is a hundred times faster. Just hoping...

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: mgglim
Replied on: 10/05/2003 00:06:46
Message:

quote:

---

Originally posted by werner_jo

Transmetas Efficeon:
http://www.transmeta.com/efficeon/efficeon_faq.html

Transmeta is not disclosing the die size or clock speed of the Transmeta Efficeon Processor at this time.

Yes the die size will be bigger (1 MB L2 cache) but this will have no effect on the TDP. The die size is measured by mm², for example the the 0.13 ìmNorthwood PIV has 131.4mm². That is not size of the ceramic or metal plated chip which is fixed on the socket. So it will be possible to build a P2000 sucsessor with an efficion (together with an ATI or Nvidea GPU with 4*AGP) using the same place and cooling system like crusoe.

My speculative prediction: a 1.2 GHz Efficeon (TDP 8W, 2.1 times higher performance than current crusoe systems), DDR333 Ram, Geforce4go in a subnotebook will be anounced in oktober. Maybe a lifebook but i think it will be an other company.

---

Reply author: oion
Replied on: 10/07/2003 09:12:44
Message:

No laptop manufacturer announcements, yet Transmeta has already snagged Fujitsu as a foundry for its 90-nm version Efficeon for the end of 2004. (Intel released their own news of their 90-nm version in August. -Anandtech) Before people suggest it, I don't think this automatically means Fujitsu would use the Efficeon in their future notebooks - after all, Fujitsu didn't even use their own hard drives (e.g. MHT line).

As for die size, the information was posted at several sites, so I don't know how they got information about the TM8000's die size besides possible press releases I haven't seen or some insider info or whatnot (Vanshardware, Impress.co.jp/Transmeta - outdated, Endian). Those readings may be completely off, and I don't know how reliable they are, but it's stated thusly: TM5800 = ~55mm^2, TM8000 = ~100mm^2, Pentium-M = ~83mm^2, AMD mobile XP = ~84mm^2. Not being an engineer, I know there are many other factors concerning price and form factor; at the least price would likely increase with the larger die, however. How much is an interesting issue from a marketing standpoint, especially in comparison to the increased price tag of a Centrino-branded notebook. Not sure where AMD will stand on this whole issue.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: Crimsona
Replied on: 10/07/2003 12:37:12
Message:

I remember reading that the Efficeon will have a integrated northbridge, which would add quite a bit to its die size. This would be a similar move to AMD's integrated memory controller in their Athlon 64's. An integrated northbridge makes it 1 less chip to produce... which could reduce costs on the whole motherboard.

Reply author: oion
Replied on: 10/07/2003 14:51:47
Message:

Ah right, I completely forgot about those. Revamp: TM8000 has the integrated northbridge AND AGP video, DDR controller, and a few other things. Reminds me now of the Shuttle series with all the integrated stuff - space-saving and cost-saving indeed. Perhaps the overall costs will end up evening out nicely. There's a nice thought... I'm much more optimistic now.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: Chris Snyder
Replied on: 10/07/2003 16:49:34
Message:

One more bit of news:

http://www.theregister.com/content/3/33263.html

Apparently Fujitsu is going to be producing some of the Efficeons. This doesn't necessarily anything - they're obviously completely separate divisions of Fujitsu - but it might give Fujitsu more incentive to use the chips, especially if it means cutting into Intel's sales, as they'll have a stake in the sales of Efficeon chips as well.

Reply author: Chris Snyder
Replied on: 10/07/2003 16:51:02
Message:

Oops, never mind - I didn't see oion's post above.

Reply author: wright45
Replied on: 10/10/2003 14:02:23
Message:

New on the Transmeta web site, they are going to have a live webcast of the launch announcement at the Microprocessor forum.

the wording of the news says "as they unveil the details of the Efficeon processor"

No hint for systems coming out then, but this seems like alot just to give us more "details"

Well, only 4 more days til the "historic launch"

Reply author: newpbx
Replied on: 10/10/2003 16:43:07
Message:

One of the Fujitsu or Transmeta web sites quotes the relative performance of the "first release" 1.2Ghz Efficeon as 50% faster on "real world" aps, and 80% faster for multi-media aps, (than the 1.0 Ghz Crusoe 5800).

My guess is that the new P series (Efficeon-based) will also feature a slightly thinner case, an improved keyboard, perhaps with the arrow keys dropped to slightly below the rectangular QWERTY block, a more natural location for the right shift key, and the same overall footprint.

The new video chip/bus arrangement should produce noticeably better video performance (for DVD playback), and maybe even play out HDTV images in real time when an accessory RF receiver card for broadcast TV is used!

I also read that Transmeta is already talking about a 2nd version of Efficeon chip with better performance, (due 3Q04 as I recall).

The Microprosser Forum presentation by Transmeta will be on 10/15.

Reply author: oion
Replied on: 10/10/2003 17:24:01
Message:

The Microprocessor Forum blurb isn't news - the actual press release for that was back in August 12, though they gave only a four-day date range back then. The perf specs aren't news either (see the "TM8000" thread on this forum concerning that). The new 2004 iteration was the news I posted that concerning using Fujitsu as a foundry; unless you're talking about something else, please be more specific (and give the actual press release URL if possible).

There has been no rumours of a new Transmeta-based P yet, so you can't say "the new P-series will feature..." unless you have a real source somewhere. I'm hesitant that Fujitsu PC will even continue to use chips from Transmeta at this point, but we'll see. I really hope they do. For me, keyboard isn't an issue since I can relearn relatively easily (though I wouldn't want a keyboard like IBM that doesn't have the Windows keys). In fact, I'd be right happy just to have a normal P-2k with all the normal quirks and just the processor upgrade. Keeping my expectations slightly low....


* P.S. - I don't suppose anyone is actually GOING to the Microprocessor Forum?

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: oion
Replied on: 10/13/2003 15:13:46
Message:

P.S. - Actual presentation is on 10/14, not 15. And you have to sign up.

October 14, 2003 at 03:30 PM Pacific Daylight Time

Too bad I'm on modem...

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: christh21
Replied on: 10/13/2003 22:14:41
Message:

oion, I signed up. I hope to watch the broadcast tomorrow. I'll be sure to post anything that doesn't get posted on the transmeta website. I'm excited to see what's up after all this time waiting!

Reply author: oion
Replied on: 10/13/2003 23:05:22
Message:

Jealous.

Well, I haven't decided if I'll vulturize all the sites tomorrow; I'm sure the information will be flying every which way, anyway. So if someone else posts all the specs/tests/blurbs/etc. before I do (and I can't watch the live webcast anyway), I'll change the thread title to "RELEASED" ...

Oh, I suppose I'll watch their stock just for amusement, too. hehehehe. (No, I don't have any.) Maybe even Intel's reaction, if any...

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: Crimsona
Replied on: 10/14/2003 00:06:40
Message:

Transmeta is too small of a competitor for Intel to care, really. Intel's more worried about what's AMD doing... I doubt there will be much in terms of design wins, sorry to say.

Reply author: blehmann
Replied on: 10/14/2003 00:54:37
Message:

My prediction remains that an efficeon p will be announced. I really hope I am right. Intel does not care about TM per se, but I do.

bnl

Reply author: oion
Replied on: 10/14/2003 09:35:01
Message:

AMD used to be small. Now look where they are - they took a chunk the market by sheer virtue of being better and cheaper. It's too late for Intel to corral AMD back to its small K6 corner. Granted AMD is still only at about 20% share, but that's pretty significant and with greater pressure, they can improve. Now, AMD and Transmeta have been working together behind Intel, as well. It's no secret that Intel is in direct competition with Transmeta at this point, therefore anything Transmeta does in this particular market must be perceived as a threat, even if small. If it's better, cheaper, and they manage to convince OEMs, it doesn't matter how "small" they might be.

Obviously Intel must have felt pressure from Transmeta in the mobile realm, otherwise there would be no need for them to market a "battery-friendly" chip so heavily when they already have a huge percent of the overall market. No, I think Intel's response to Transmeta was to in fact develop the Pentium-M. I don't think it would be so far-fetch to say - if it weren't for Transmeta's publicized light-energy and relative popularity in the Japanese market (a least in the pioneer sense), Intel would not have bothered trying to create a more energy-efficient chip.

Obviously at this point the trick is to see whether Transmeta can win over any OEMs; it may or may not have to do with marketing power. More will help, certainly, but AMD didn't exactly have a marketing budget early on either, in their war with Intel. I'm not sure they do even now, besides relying on benchmarks and pricing.

If anything, hopefully Transmeta can be a player and we'll have a three-way competitive battle that can only make the chips better for consumers. That's the real issue, anyway.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: oion
Replied on: 10/14/2003 14:03:57
Message:

Oct 14, 2003

quote:

---

Efficeon features a number of new high performance interfaces integrated into the same chip as the processor, eliminating the need for a separate Northbridge chip. Efficeon includes a high speed AGP-4X graphics interface that enables the use of industry standard high performance AGP graphics solutions. Efficeon includes a high speed DDR-400 (Double Data Rate SDRAM at 400 Megatransfers/second) memory interface, with ECC (Error Correcting Code) as an option. Efficeon includes a high performance HyperTransport interconnect for connecting to the rest of the system. Efficeon's HyperTransport interface can send data at up to 1.6 GBytes/ second aggregate, which is up to 12 times faster than the I/O throughput of Crusoe's PCI interface. Efficeon also includes a high speed LPC (Low Pin Count) bus interface for communicating with the latest generation of FLASH memory.

With integrated on-chip Northbridge core logic, the Efficeon TM8000 family reduces chip count and decreases the size and cost of the PC board, enabling innovative, smaller form factor designs. Transmeta and NVIDIA have been working jointly together on small package solutions. The Efficeon TM8620 21x21mm small package combined with NVIDIA's 22x22mm nForce3 Go 120 Southbridge will allow the combination of CPU, Northbridge and Southbridge to fit in a space nearly four times smaller than that of its nearest competitor. The standard model Efficeon TM8600 processor will be offered in a 29x29mm package, also smaller than the competitive processors.

---

Reply author: grandall
Replied on: 10/14/2003 14:07:30
Message:

From the article at http://siliconvalley.internet.com/news/article.php/3091711:

"The company says it has already signed up partners for its new chip including ROX for blade servers, Sharp and Fujitsu for a new version of a thin and light and a handful of Tablet manufacturers."

Reply author: oion
Replied on: 10/14/2003 14:11:01
Message:

Ooh, good good. I particularly like the quote, "Transmeta has got to win an award for the number of most subtle hints dropped about a product prior to its release."

But "thin and light" - that's not the same class as an ultraportable like the P-series, in classic terms. We'll see.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: grandall
Replied on: 10/14/2003 14:27:57
Message:

Maybe I was reading too much into "thin and light", but my interpretation based on the two manufacturers listed (Sharp and Fujitsu) was that the form factor would be similar to those used in those companies' current Transmeta offerings. In Fujitsu's case, that would be the P series, and in Sharp's, the 2.1 lb Actius MM10.

At least, that's what I hope. I have a credit card warmed up for a Efficeon version of the P1000.

Reply author: oion
Replied on: 10/14/2003 14:37:06
Message:

I've always considered Fujitsu's "thin and light" as their S-series, not P-series. Actually, they market it that way as well: "LifeBook S Series notebook is thin and light" while the P-series used to be marketed as the ultraportable. Though I doubt they'll add another processor branch to the already two-armed S-series. Maybe they'll make something between the S and P? Who knows. I certainly wouldn't mind a 12-incher like the Panasonic W2.

By the way, I didn't notice when Fujitsu stopped selling the P-2000 series on their site.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: The Censor
Replied on: 10/14/2003 14:59:33
Message:

http://www.crn.com/sections/BreakingNews/dailyarchives.asp?ArticleID=45196

Transmeta Unveils New Processor, Efficeon



E-mail this article
Print this article
Link/reprint this article

Breaking News
Transmeta Unveils New Processor, Efficeon
MySQL Buys Alzato, Clustering Technology
Sony Unveils New All-In-One Destkop For Retail
Plumtree Nears Profitability
EDS, Opsware, 20 Other Companies Back New Open Standard For Utility Computing
IBM To Buy CrossAccess
Intel Unveils Flash Memory System For Cell Phones
EMC To Buy Documentum In $1.7 Billion Deal


By Edward F. Moltzen, CRN

2:21 PM EST Tues., Oct. 14, 2003
Transmeta formally took the wraps off its next-generation processor, the Efficeon TM800 - a chip the company says doubles some performance measurements over its predecessor Crusoe processor.

The new offering from Transmeta, Santa Clara, Calif., includes both the physical processor and "code-morphing" software that helps provide for longer battery life.

"Crusoe was a good chip," Heinlein said. "It delivered good battery life, (but it's) performance was not universally praised."

Efficeon was built as an entirely new processor -- with both hardware and software - largely to improve performance. While Crusoe could run four instructions per clock cycle, Heinlein said, Efficeon can run eight instructions per clock cycle.

And the company said the Efficeon will perform at 1.1 GHz, compared with the 900 MHz offered by Intel's Mobile Pentium M in the Centrino platform.

Transmeta executives also said the new Efficeon chip uses a fraction of the power on "standby" as a similar Centrino-based processor from Intel.

According to Transmeta's roadmap, the company will ship a 2 GHz version of Efficeon next year, running at 25 Watts and built on a 90 nm process technology under a new partnership with Fujitsu. The roadmap calls for a transition to an even smaller, 65 nm process for Efficeon in 2005, when the company said it will begin shipping a third-generation Efficeon at higher performance.

Pricing was not immediately available, nor was a list of original equipment manufacturers who would adopt the new technology. Transmeta executives have said they also intend to seek out the custom systems channel for their processors over the next several months.

Reply author: The Censor
Replied on: 10/14/2003 15:03:58
Message:

more articles:

http://www.infoworld.com/article/03/10/14/HNefficeon_1.html

http://www.reuters.co.uk/newsArticle.jhtml?type=technologyNews&storyID=3611898§ion=news

Reply author: oion
Replied on: 10/14/2003 19:12:39
Message:

A few notes and stuff I caught on the webcast (via my puny modem connect ):

(Integrated stuff, 128kb/1MB L1/L2 cache, compatible with MMX/SSE/SSE2... etc.)

Price range: $20 Crusoe, $100 Efficeon

Encryption benchmarks: relative to P-4
RSA 1.4x P-4
..(missed the two other encryption benchmarks)

Sharp notebook system comparison: thermal limit of 7 watts w/o fan: Centrino is 900MHz, TM can do 1100MHz at same power level.

Benchmarks: (missed the test names)
Integer: 2.7x throughput of Centrino
Floating point: 1.3x Centrino
System: (TM ahead)
3Dmark: 1.6x Centrino

Standby power:
8x lower power than Centrino on idle

(Partner with NVidia compact board space solution):
total area combined CPU/bridges: 925mm^2 (~4x smaller in board space than [x])

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: christh21
Replied on: 10/14/2003 19:29:55
Message:

Here are my notes while watching the WebCast:

4 years of work!

new information:
SPeed: 1.0ghz/5w - 1.1ghz/7W - 1.2ghz/12W - 1.4ghz/15W (then up to 2ghz in 2004 with 90nm)

benchmarks: Floating Point / Integer computations 1.3-2.4x faster than Centrino
.18 watts in idle state (vs 1.8w on Centrino)
smaller board space

2x higher integer
1.3 higher floating point
1.6x higher 3D graphics performace vs Centrino
fanless
2x the framerate on DVD playback than Centrino

System benchmarks (CPUBench, etc)
Efficeon provided very similar results to Centrino, higher in CPU intensive tasks and slightly lower in others. (they claim that performance was hindered because they were forced to us DDR-266 because Centrino doesn't support DDR-400)

CMS/Long Run:
4 gears of translation (compared to 1 with Crusoe) which allows translation at the instruction, region, and cross-region levels, as well as "HOT SPOTS" (instructions that are commonly used)

Future of Intel processors: Leakage power and voltage fluctuations
Transmeta solution: LONGRUN2
- software dynamically changes voltage and MHz
- Software Controlled Leakage Management
- Dynamically adjusts threshold voltate
- reduces leakage power, active power, and standby power
- dynamically adjusts threshold voltage (which is hard-wired in other chips)
- LongRun2 can reduce standy power to 2mW (144mW without)
- won't be available in first generation Efficeon

Reply author: christh21
Replied on: 10/14/2003 19:32:55
Message:

Fujitsu was onboard with Transmeta, so it's likely that we WILL see a Fujitsu Efficeon based machine.

I have to say that I was surprised by the presentation. My preconception was that Efficeon would rate just at or just below Centrino performance, but that doesn't seem to be the case. 1.6x 3D perfomance! 2x DVD framerate! amazing. I am very excited.

Reply author: grandall
Replied on: 10/14/2003 19:58:21
Message:

Not to be pessimistic, but I would suggest taking these benchmarks with a grain of salt until someone actually makes a machine with one of these guys in them. Benchmarks are always in the eye of the beholder. 1.6x the 3D graphics performance sounds great, but is that because they're comparing a system with an Nvidia GPU to the bland Intel integrated graphics? What does 2x the framerate on DVD playback mean - I have a Pentium II notebook that plays DVDs without generally dropping frames. I don't see how you can get twice the frame rate of no dropped frames.

Nevertheless, it's nice to see those kinds of numbers. My concern now is that if it's being pushed as a higher-end, faster chip, that it won't end up in ultraportables like the P-series.

Reply author: Crimsona
Replied on: 10/14/2003 20:02:24
Message:

Most worrying is the problem of no design wins. I was hoping for at least 1 or 2 OEM partnerships being announced... but seems like that wasn't the case

Reply author: oion
Replied on: 10/14/2003 20:05:09
Message:

I agree about the benchmarks - the problem is that we don't have enough details about the actual benchmarks and testing conditions, I think. I don't recall the DVD/framerate thing. Even if the benchmarks appear "skewed" like Intel sometimes does for its own processors, the TM8000 should come out at least about par, which is where it should be.

Not sure how the cost thing will factor in, though. Would have to know Pentium-M prices.

Just remember that the Efficeon is being marketed for the ultra-mobile market as well as normal laptops - after all, they talked a great deal about footprint size as well as battery life. The high-end chip in ultraportable form factor combination is precisely the one that's generally unattainable. - Therefore if Efficeon can break this traditional barrier, great.


Actually, they do have some OEM partners on the Efficeon; I don't remember the slide exactly, but Fujitsu, HP, and Sharp were in there. Likewise Sharp already had an Efficeon notebook on display at the Forum. Don't know the details there, though.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: christh21
Replied on: 10/14/2003 20:51:30
Message:

Well that was my point, of course you have to assume that the benchmarks are going to show the Efficeon as a better processor. But even with a grain of salt, I can't see the benchmarks being so off that we'll be disappointed by the new processor. One of the slides showed standard computer benchmark programs, as I mentioned earlier, and they showed the Efficeon better on same and just under Centrino on others, but if you average those benchmarks out the Efficeon will likely be comparable to the Centrino. I think this is a huge step for Transmeta, and with speeds reaching 2ghz by the end of next year, Efficeon will definitely be a strong contender.

The other point that the benchmarks point out is that the Efficeon supports high end 3rd party chips that even Intel doesn't so far, or at least not in mobile form. Efficeon supports HyperTransport, AGP 4x, good stats even for Intel. If Efficeon computers come with nVidia chipsets (which are going to be Southbridge, not just Graphics) that is a huge improvement over current Crusoes or even Intel machines.

Personally, I don't do anything that really needs a gamer style power desktop. I think that Efficeon will do quite well. The major thing is they need a big OEM to get the processor out to people who only buy the big names (everyone I know owns a Dell, HP, Compaq, or Gateway).

My personal opinion, Transmeta is opening the way into the future. The presentation about leakage issues just showed that processors cannot continue on the same path that they are on right now. In order to take on those issues, there will need to be a radical change in the way processors are developed. Transmeta has the right idea and is opening up a whole new realm of computing.

An Oion, yes there was a slide full of OEMs, Fujitsu, Sharp, HP, and several others. To me, it looked pretty similar to their current list of OEMs.

Reply author: grandall
Replied on: 10/14/2003 21:55:28
Message:

With regards to costs, current (as of Oct 6) Pentium M prices are as follows (as reported by CNET at http://news.com.com/2100-1006-5086860.html):
1.7GHz $423, 1.6GHz $294, 1.5GHz $241, and 1.4GHz $209

So... if the Efficeon is coming in at $100, and even assuming that $100 is for the low-end 1GHz version, then it looks like it's at a better price per unit speed regardless of the benchmarks.

As I understand it, though, Intel is planning on releasing a new low-end notebook chip based on the Celeron, a "Celeron M" so to speak, in the next few months. Clock speeds will start at 1.2GHz. Prices haven't been announced, but I'd be surprised if they're much more than $100, and they could be less. Even though the performance might be lousy, having an Intel "M" chip close to $100 could make things harder for Transmeta.

Reply author: oion
Replied on: 10/14/2003 22:05:25
Message:

I've condensed a recent press release: (10/14) Products "sort of"

Amoi Electronics: V3 ultraportable notebook
Amtek: 12" Tablet PC
Antelope Technologies: Module Computer Core
Densitron: Single-board computers
DTR: embedded and mobile computing devices
Fujitsu: No general plans.
GBM: 12" 2-spindle notebook
HP: No general plans.
IBASE: embedded products...
JMNet: Blade servers
MoBitS: General product designs...
Plexus: No general plans.
Sharp: Actius
Xybernaut: wearable/mobile computing...

The only bigger name in this mix that has actual plans is Sharp; Fujitsu and HP basically only "support" Transmeta, but there haven't been any word on a specific product line or product type in development (those "No general plans"). I've heard of the low-end "M" - it would definitely cut into the price range Transmeta is aiming at. Looks like they feel threatened enough to take notice, at least...

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: grandall
Replied on: 10/14/2003 22:32:39
Message:

By the way, in case people haven't been following, the "other" processor brought out today looks pretty good too: VIA's Eden-N, which packs a 1 GHz CPU into a 1.5x1.5cm space. That's smaller than a penny and several times smaller than a Pentium M. Power consumption is similar to that of the P-M or the Efficeon, and the cost is supposedly $30 or so. I'd bet you could make a pretty decent version of the P1000 with one of those inside.

Reply author: blehmann
Replied on: 10/15/2003 01:25:11
Message:

I am disappointed that no agreements were announced. I thought (and still think to some extent) that the death of the p2 meant the birth of a replacement, not the last of the line. The price of the efficeon is not too high at $100 and efficeon performance is in the right ballpark at lower power consumption than the pentium m. I guess I am not sure that it pays for Fujitsu to slice the salami thinner by having both a p2 and p5 unless p2 power consumption is much lower or its performance is much better. Maybe there will be good news soon but it is a bad sign that Fujitsu did not announce today. You have to figure that Fujitsu had every opportunity to work with seed units.

bnl

Reply author: mgglim
Replied on: 10/15/2003 02:15:48
Message:

quote:

---

Originally posted by blehmann

I am disappointed that no agreements were announced. I thought (and still think to some extent) that the death of the p2 meant the birth of a replacement, not the last of the line. The price of the efficeon is not too high at $100 and efficeon performance is in the right ballpark at lower power consumption than the pentium m. I guess I am not sure that it pays for Fujitsu to slice the salami thinner by having both a p2 and p5 unless p2 power consumption is much lower or its performance is much better. Maybe there will be good news soon but it is a bad sign that Fujitsu did not announce today. You have to figure that Fujitsu had every opportunity to work with seed units.

bnl

---

Reply author: shanemac
Replied on: 10/15/2003 02:57:59
Message:

Well said mgglim, my wallet is burning a hole in my pocket i'am not sure if i can hold out much longer but i said that 6 months ago, so again time will tell.

Reply author: oion
Replied on: 10/15/2003 08:20:13
Message:

Remember that they said that OEM shipments were supposed to begin in September. So it was supposed to be in production - but given the all the previous "partner" press releases, it makes me wonder if they had fallen behind schedule yet again and had not released real press releases prior to Microprocessor precisely to cover up that fact. It's also possible the other bigger names were simply waiting to see how TM would do overall anyway (Forum, etc.) before investing. Make no mistake - it's certainly not TM's fault for no product announcement from Fujitsu; they probably will have the 90nm version released next year. It's a pretty big gamble, especially when these OEMs already have contracts with Intel. Anyway, I never expected to see end-user TM8000 products until November to December anyway, so I'm neither surprised nor upset. Once those months roll by, I'll be complaining more.

Well, I'll be watching the Actius now.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: gellul
Replied on: 10/15/2003 09:26:03
Message:

My P2120 will be a year old this November 2003. I am not ready for a new P just yet. But next year I could pass my P2120 onto my daughter who will be starting college in the fall and purchase a new P for myself.

I vision my new P to have:
Transmetta Efficeon TM8000 at 1.1GHz
LongRun2 software
512M DDR333 or faster
APG graphics chip set with 32Meg of Ram
DVD -R/RW +R/RW
the same size/form factor as my P2120
4-5 hour primary standard battery

Oion, I couldn't resist posting my want list on your excellent thread.

Reply author: oion
Replied on: 10/15/2003 10:00:59
Message:

Hehehe.

My ideal P-series (or Y-series or whatever they might call it):

TM8000 1+GHz
1GB DDR (not too picky about type)
Gfx with vid in/out (24+MB ok); DVD hardware decoding (right...)
DVDRW optical, slot-load (but then I can't use the mini-cdrs...)
BIOS-bootable USB, Firewire, flash slot (hehehe)
Same size/ff as classic P-2ks, possibly thinner (~11")
5-hr std battery (surfing, mp3s, docs, 60% brightness)
2 lbs w/std battery
Green coloured-case (inside joke)
No fans ANYWHERE (well.... mebbe PCMCIA)
Other: 3-yr std warranty, about $2k or less

Actually, my want list certainly doesn't have to be from Fujitsu.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: christh21
Replied on: 10/15/2003 11:10:43
Message:

My want list:

Something under $1500!! I'm not a rich man, I just want enough to get the job done. Let the extras be extras

TM8000 1.1+Ghz (the faster models are much higher wattage and probably require a fan)
5400RPM Hard drive
512MB DDR RAM
CD-RW Hotswappable (With a DVD option available of course)
MemoryStick/Compact Flash/etc slots
BIOS bootable USB flash/floppy and CD
TFT screen (I would love to study outside) with TouchScreen? Maybe tablet style!
A MUCH better Wi-Fi than current P (the range SUCKS on the Orinico the P2 has) - must be 11b compatible
Slightly thinner than the P, a little lighter, and maybe just a touch bigger screen (16:9 instead of 15:9, maybe 11" to 11.5")
Again, any brand that puts this out I would buy, even if it wasn't Fujitsu, but I would rather it was.

Okay, so a little carried away in the end... frankly, I would buy ANY ultra-portable Efficeon with batterlife up to 10 hours, and with some form of CD-RW (external would be fine)

Reply author: oion
Replied on: 10/15/2003 11:18:56
Message:

Actually, for outdoor use that should be the transflective or reflective TFT LCD, not just "TFT" (since pretty much all LCDs these days are active matrix TFT anyway). I don't even have the opportunity to use wireless at the moment, but I suppose I should get a model with it in the future. Of course by then they may all have wireless anyway.

10 hours and lighter than the P? Or do you mean with "optional" battery. Hm, I don't recall the webcast having actual battery life figures anywhere. I don't think sub-3lb 10+hr with 1GHz+ performance is feasible right now.

My minimum featureset still stands - a standard P-2000 series with an upgraded chip. C'mon, can't be too difficult to build...

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: Crimsona
Replied on: 10/15/2003 12:19:05
Message:

I realize how much I use CF/MS/SD slots... My next laptop MUST have all 3 slots (or at least cf or sd/ms so i can use a pcmcia reader)

That's probably the main reason for the P5k over pretty much anything else right now

Reply author: blehmann
Replied on: 10/15/2003 13:35:47
Message:

I have a feeling I am going to buy a p5 soon since it is hard to see if some oem will produce laptops based on 130 nm efficeon chips or wait for the 90 nm chips that Fujitsu will make for tm sometime next year, allegedly. I have two specific questions for those who caught the web cast that arise from an eetimes article I read at http://www.eetimes.com/semi/news/OEG20031015S0020. First, it mentioned that both Fujitsu and Sharp showed some sort of prototype laptops using efficeon chips. Assuming they got that right (not always a reasonable assumption), what did the Fujitsu prototype look like? Second, they refer to the mobile Athlon used in the Fujitsu S2 and a recently announced chip by Via Technologies called the c5p. The article said the latter has comparable performance at half the price. Does anyone know anything about this chip and who might be using it?

bnl

Reply author: oion
Replied on: 10/15/2003 14:06:22
Message:

Well, Sharp is definitely going to have an Actius using the 130nm version, as mentioned before, so that's still good. I don't recall the Fujitsu prototype at all - someone remember that at the webcast? Of minor interest, Transmeta will release their current financial results today after market close. I wonder if companies will really hold out for the 90nm - if the current TM8000 doesn't do well right now, where will they get funds and the impetus for the upgrade? Too much is riding on this chip as it is. All we can do is wait and see, as usual.

As for VIA's chip, I assume you're talking about the one Grandall posted about earlier. I haven't followed it myself, and I haven't heard of the Via chips being used in any notebook configurations. Someone confirm? I wonder what products they're used in.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: mgglim
Replied on: 10/15/2003 14:31:40
Message:

quote:

---

Originally posted by oion

Well, Sharp is definitely going to have an Actius using the 130nm version, as mentioned before, so that's still good. I don't recall the Fujitsu prototype at all - someone remember that at the webcast? Of minor interest, Transmeta will release their current financial results today after market close. I wonder if companies will really hold out for the 90nm - if the current TM8000 doesn't do well right now, where will they get funds and the impetus for the upgrade? Too much is riding on this chip as it is. All we can do is wait and see, as usual.

As for VIA's chip, I assume you're talking about the one Grandall posted about earlier. I haven't followed it myself, and I haven't heard of the Via chips being used in any notebook configurations. Someone confirm? I wonder what products they're used in.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

---

Reply author: gellul
Replied on: 10/15/2003 14:45:25
Message:

Best to wait until the LongRun2 software is out. LongRun is the Transmeta software that varies the frequency and voltage of the Crusoe processor to get maximum battery life. LongRun2 is for the Efficeon processor. The performance with 130nm is still going to be good. The 90nm will allow the price to come down since they can get more processor per wafer.

Reply author: grandall
Replied on: 10/15/2003 15:09:05
Message:

I know I'm getting off topic, but with regards to VIA and their chips: VIA's been producing some fun chips and motherboards for the last year or so, targetting the embedded computing and "small computer" market. They have a series of mini-ITX (17cm x 17cm) motherboards complete with CPU, PCI slot, USB, Firewire, the other normal ports, etc. Many of their motherboard/CPU combinations are fanless as well. People have put these into a lot of neat places, like having a full computer in their car stereo slot, or as a tiny multimedia set-top device, or running kiosk, or in a toaster, etc. VIA's got some great pictures of things people have done with their stuff on their web site. They haven't shown up in a lot of notebooks (at least in the US), but I think VIA would like to go there. They don't make great performance systems, but if you're looking to build a computer into a stuffed animal, VIA's your chip.

Reply author: christh21
Replied on: 10/15/2003 15:19:28
Message:

Oion, I agree with you about the P2000 successor... while I went a little overboard there at the end with wishful thinking, If I could get my same machine with the new Efficeon, I'd be overjoyed. Heck, if they just sold the new motherboard that I could throw into my current laptop I'd go there too, especially if we're looking toward December.

With all these new chips coming out, at such low prices, I can't believe the OEMs aren't jumping on the bandwagon to get out from under the Intel domination. Their chips are the most expensive on the market. I understand the OEM's reluctance, I just think Intel is too big and has its paws in places they shouldn't be (like forcing OEMs into corners).

Oh well... maybe in 10 years we'll see more processor competition out there.

Reply author: oion
Replied on: 10/15/2003 15:56:36
Message:

I think a major issue on the reluctance of OEMs is end-user marketability. Let's face it, the average consumer recognizes "Intel," not "Transmeta." Despite everything, brand names sell stuff; just look a "Centrino," which isn't even a special chip per se. A fundamental law of consumerism, perhaps. The same goes with Microsoft and other heavyweights in any imaginable industry, of course.

Let's look at Via (oh yes, I did hear about this insanely tiny custom mini-pcs, very interesting): ultra-cheap, from a name not popularly known as a CPU-maker. What does the fundamental law of brand names say? You pay the premium price on the branded product because of the implied guarantee that the product is top-of-the-line, won't fail, and will perform as it should. VIA may or may not make it to mainstream laptops (I doubt it, it seems very niche right now), but it's sort of how people look at low-end chips in general. For the average working adult with the money to spend, they will undoubtedly opt for the traditional, classic, known names. This is why Dell, HP, etc. do well with their cookie-cutter systems. Students and younger people with more limited spending power can "afford" to try other things. Then there's the education variable - most people don't compare the intricacies of CPU architecture very closely. It's still mostly about GHz.

I'm thinking Transmeta has to somehow work on its brand image; it's all marketing. If they could steal a few key people from the bigger companies, that would be nice. (Yea right) The mere fact that Linus Torvalds was at Transmeta was in itself a built-in marketing strategy that made the brand a bit more visible early on...

10 years is awfully pessimistic.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: grandall
Replied on: 10/15/2003 16:12:04
Message:

I think some OEMs also fear the unknown, or only want to deal with as few a number of options as possible to help fit a cookie-cutter approach to production. Here I am specifically thinking of Dell, which has yet to consider an AMD chip, let alone Transmeta. When recently asked if they would be offering Athlon 64-based systems, a Dell executive responded that they wouldn't do that because the chip is too "ice-breaking."

Reply author: oion
Replied on: 10/15/2003 16:17:11
Message:

..... STUPID Dell....

And yet, their sales power sure illustrates something. (sigh)

I forgot to add that Dell may have special pricing deals with Intel, anyway; haven't kept up with such rumours/news, though.
______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: christh21
Replied on: 10/16/2003 00:29:19
Message:

Some of the things that Intel gets away with sure seem unethical at best...and should be illegal. They are creating a monopoly by forming special princing deals that require the OEM to purchase processors from them exclusively. I think this contractual exclusivity deals should be illegal because they remove choice from the market. That's just my 2 cents (all my 2 cents on Transmeta and Intel are sure adding up recently aren't they)

Reply author: thetenken
Replied on: 10/16/2003 20:02:10
Message:

Here's a press release of the Efficeon with a nice shot of the processor itself.
http://www.transmetazone.com/releaseview.cfm?releaseID=1050

Here's the Via Eden processor and some products forthcoming...
http://www.linuxdevices.com/news/NS7542683131.html?foo=VIA%20wows%20with%20nano-sized%20x86,%20entropy-based%20security,%20tiny%20PCs%2010--16

Reply author: newpbx
Replied on: 10/17/2003 17:30:58
Message:

The URL below has some interesting analysis and "live" screen shots from the Transmeta Efficeon announcemnet earlier this week. Transmeta shows oblique references to Transmeta's performance metric for Efficeon, but no actual real world comparisons.

The initial basis for comparison for Efficeon vs. Pentium M is "What will each do at a maximum power drain of 7 watts", the upper limit to running a laptop without a fan.

Also, the planned speeds for future Efficeon versions are shown.

See:
http://www.extremetech.com/article2/0,3973,1342823,00.asp

Reply author: oion
Replied on: 10/17/2003 18:11:17
Message:

What's interesting are the benchmark/power draw discussions, though they don't give as detailed an analysis as I'd like (since they didn't have the chips to play with). More analysis and information than I could get from the webcast on my limited connection, certainly. Hope to see many more technical articles this month.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: grandall
Replied on: 10/17/2003 18:34:40
Message:

So let me get this straight... their comparison of speeds are not based on performance per clock speed, but essentially on performance per watt? What speed of Pentium M comes in at 7 watts? If Transmeta is comparing a 1.1 GHz, 7 watt Efficeon to a 900 MHz, 7 watt P-M, that would be useful to know.

Reply author: christh21
Replied on: 10/17/2003 19:29:48
Message:

grandall,

from what I saw in the webcast, they did some MHz to MHz comparisions and some performance/watt comparisions. The benchmarks were based on MHz if I remember right, but the power consumption benchmarks were based on wattage. So they did compare the Efficeon running at 1.1ghz at 7watts, and the P-M running at 900mhz. I think the later point was that the Efficeon can out perform the P-M when restricted to power consumption.

EDIT:
I just double checked, and the "real world" benchmarks, that is "Business Winstone" and "PC Mark" were running at 7watts of power, which means they were comparing an 1100Mhz Efficeon to a 900MHz P-M. The other benchmarks were "per clock." (and Efficeon will win because it can execute twice as many instructions per clock)

The interesting point here, is that people seem to want a processor that is faster than the Intel, but still provide the power savings. Personally, I don't mind giving up a little performace for longer battery (and no fan).

We'll just have to wait for real laptops to come out and do our own tests.

Reply author: mgglim
Replied on: 10/17/2003 21:29:29
Message:

Like y'all, I'm waiting for the benchmarks to come out. Who will be the first company to come out in the open and moon Intel? We know who's definitely last- button-downed Dell. But that doesn't matter as Dell for me is a non-entity when it comes to the notebook world. They don't really own the process of manufacturing their notebooks, and that being the case, are beholden to Intel to guide it. Intel gives them specs, and they pass the specs to their contract manufacturers in Taiwan, who produce a "safe" product. Hence, Dell is a non-entity. They do save on R&D, but their notebooks are mediocre, design-wise, and unreliable, quality-wise. Just go to their website and look for how many refurbished notebooks they have. Note: Refurbished = Previously returned.

This begs the question: Which of the notebook brands really have control of their notebook design and development,is capable of tweaking processors to optimize their performance, and is capable of in-house production to control product quality? The answer, I believe, lies in the ability to put the most capabilities, in the smallest form factor, at an excellent quality level, and at a low cost and thus low price point.

This still points to the Japanese brands. I think it is safe to say that Japanese-branded notebooks have risen to the forefront in the same way that Japanese cars, pickups, and SUVs have dominated the auto market. Because my Fujitsu P is so reliable, for example, I am loyal to it, would make repeat purchases, and would recommend to friends to buy Fujitsu, whether new or used. If I were to take chances with used notebooks, I would buy a Fujitsu. But not a Dell, for example.

This leads to my thesis that Transmeta would need the Japanese more than ever. Also, the American preference for size would limit the market for ultraportables, thus limiting the market in the U.S. for the initial production of Efficeon processors. Thus, the Japanese would spearhead Transmeta's push in the Japanese, Chinese, Asian, and European markets.

So, my guess is that Transmeta would pull all stops to get Fujitsu, Sharp, Toshiba, and Sony to push the Efficeon. Not to make light of the Taiwanese and the Chinese, but they lack name recognition, and it would be hard to count on them to push volumes that approach the economies of scale needed to be profitable for Transmeta.

Luckily for Transmeta, Intel is facing competition on three fronts:
- Servers: AMD with its Opteron
- Workstations and Desktops: AMD with AMD65 and FX51
- Ultraportables: Transmeta

(However, the sweet spot are desktop replacement laptops, and Intel has no real competition here.)

This makes Intel weaker, and pyschologically, this is what Transmeta needs to counter the fear of Intel's wrath- of withholding key processors from those who patronize Transmeta.

I believe the Japanese manufacturers have plans drawn to quickly ramp up production of Efficeon. They just need a sign.

The sign: That Taiwan Semiconductor is producing good yields of the Efficeon, and that they can ramp up production of Efficeon chips and provide a consistent and uninterruptible supply of chips. If Transmeta can give this assurance, which is hard to do given past experiencek, then Transmeta, thru the Japanese, will be able to sell huge quantities of Efficeons.

Transmeta may be a classic victim of its own success. It is a fabless semi company. It has moved from IBM, then to Taiwan Semi, and then will go to Fujitsu. This doesn't look good from a strategic standpoint. With the impending success of the Opteron and the Efficeon, look for Transmeta to be merged with AMD once AMD gets its profitability in order.

In the meantime, let us wish Transmeta success in producing its Efficeon chips in a timely manner.

Cheers.

Reply author: oion
Replied on: 10/17/2003 22:21:55
Message:

Agreed, as suggested somewhere on the first page. The Crusoe was also more popular with the early-adopting Japanese makers, so that looks like the market to watch, not the US market (at least not right away). The days are ticking by...

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: mgglim
Replied on: 10/18/2003 12:23:17
Message:

)Thinking it over, Intel's bargaining chip still is the Pentium 4-M, which is the desktop replacement laptop market. AMD and Transmeta doesn't have a competititive answer to them. For that matter, NO ONE.

Intel, because it has strategic control over this crucial sweet spot of the computing market, will have great sway over the Japanese notebook manufacturers, and thus can tell them to go f--- themselves if they should support Transmeta in a big way with their ultraportables.

Sadly, the antitrust enforcement in this country is asleep, as Bush pro-big business, and very loyal to big business donors, especially in a run-up to an election year. What is worse is that he sees no reason to aid the Japanese, as this is like supporting Toyota against GM.

So Transmeta sadly will have to hope people buy Taiwanese to get volume. As for us P users, we'll keep our fingers crossed that I am wrong.

Stealing from Oion, time is ticking by...

Let's hope for some good news:

1. The Japanese say "Sc..w you" Intel
2. The EU enforces antitrust on Intel
3. Apple decides to use Transmeta as its notebooks are powerhogs (heresy? look, iTunes came out with Windows version already, Apple's current OS shares code with Next, which run on x86)
4. Palm uses Transmeta to produce an equivalent to an instant-on subnote of the HP Omnibook of the early 90's
5. HP does #4 as well

So, don't get me wrong. I'm still a Transmeta bull. Transmeta has a lot of fight left, especially when trapped in a corner.

Now, the floor is open for more outrageous ideas for the Transmeta (no Al Qaeda blows up all the Intel fabs

Reply author: Crimsona
Replied on: 10/18/2003 13:59:21
Message:

Al Qaeda blo... damn.

I've used a Via 1 ghz on my aunt's computer. Oh my god I have never seen 1 ghz go that slow before. Took forever just to load IE (well I guess with Win ME and 128 mb RAM it wasn't going to be very fast). But I could swear my P2 400 could match it in speed... but i could be disillusioned

Reply author: oion
Replied on: 10/18/2003 15:59:25
Message:

Yes, quite true that mere megahertz can't be the deciding factor in overall performance... My 800MHz 2040 certainly feels nothing like 800MHz, and appears easily beaten by our K7-700MHz.

Anyway, I was perusing the PDF version of the webcast (or trying to - my P2040 was complaining about the huge file ), and put together a simple page of some of the actual benchmarks. Comments welcom, discussion encouraged.

http://www.oion.net/qnd/tm8000/

(Also edited my anchor post so people wouldn't have to poke through the thread pages it in later searches.)

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: christh21
Replied on: 10/18/2003 17:44:04
Message:

On the transmeta website, you can now view the archived version of the webcast.

Reply author: mgglim
Replied on: 10/18/2003 17:50:47
Message:

Oion,

Thanks for posting those specs. What is that about the smaller chip the 8620? Is that the 90 nm version?

These specs look very good, worth noting is how well it stacks against the P4 in the integer performance/clock part. Now, I'm thinking, can the 1st gen Efficeon then be suitable as a chip for desktop replacement laptops, going against Pentium 4-M's?

Reply author: christh21
Replied on: 10/18/2003 17:51:33
Message:

mgglim,

that's an interesting speculation about Apple... of all the major manufactures, they are the least influenced by Intel, because they use IBM chips. The other interesting thing, is that Transmeta could likely come out with a CMS that is PPC compatible. Because of the design, Transmeta could use the same chip with only a different code morphing engine to support any instruction set.

I just noticed in the webcast slides, that BOTH Sharp AND Fujitsu had "prototype Efficeon development machines" on demonstration. That may mean that Fujitsu hasn't given up on Transmeta all together.

The other interesting thing that I've been thinking about is the possibility to provide dual-CPU laptops. With an Efficeon combined with a Crusoe, or two Efficeons, you could get great performance and still have power savings (one CPU would remain idle until needed). The design of the chip really provides great possibilities.

As mentioned above, one thing I would like to see is a Transmeta chip for PDAs, or "ultra-personal computers" as Transmeta calls them. Use XP/Linux Embedded with 256MB Flash and 128MB RAM, you could have quite the nice little system in the palm of your hand.

Reply author: christh21
Replied on: 10/18/2003 17:53:18
Message:

Did any else notice that Transmeta claims a "2x power savings over previous Crusoe chips"? I find that extremely interesting.

Reply author: christh21
Replied on: 10/18/2003 18:01:38
Message:

oion: on your site you missed a closing link tag just before the list of benchmark software. :D

Reply author: oion
Replied on: 10/18/2003 18:51:13
Message:

Thanks, fixed.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: mgglim
Replied on: 10/18/2003 20:39:03
Message:

quote:

---

Originally posted by christh21



that's an interesting speculation about Apple... of all the major manufactures, they are the least influenced by Intel, because they use IBM chips. The other interesting thing, is that Transmeta could likely come out with a CMS that is PPC compatible. Because of the design, Transmeta could use the same chip with only a different code morphing engine to support any instruction set.
/[quote]

That seems like a good idea, but would not be a priority, because Transmeta is strapped for cash and could not possibly take on a project that would drain its resources for something like Macs, which has a limited market share. But if Apple could tweak its system to run on the Efficeon, that would put be a big plus for Apple, as it could enter the power/efficiency game, and Transmeta's spirits would be lifted. Apple still wields enormous effect on the mindshare of the computing community, as witnessed by the iMac, in how it catalysed the adoption of USB, and how even its native Firewire got into mainstream PC computing for video. Imagine how having a Efficeon processor could excite the computing community and legitimize the Efficeon.

[quote]
The other interesting thing that I've been thinking about is the possibility to provide dual-CPU laptops. With an Efficeon combined with a Crusoe, or two Efficeons, you could get great performance and still have power savings (one CPU would remain idle until needed). The design of the chip really provides great possibilities.
/[quote]
Good idea, that may just be what blade servers can use the Efficeons for. All the more reason for AMD to buy Transmeta.

[quote]
As mentioned above, one thing I would like to see is a Transmeta chip for PDAs, or "ultra-personal computers" as Transmeta calls them. Use XP/Linux Embedded with 256MB Flash and 128MB RAM, you could have quite the nice little system in the palm of your hand.

---

Reply author: blehmann
Replied on: 10/18/2003 22:01:08
Message:

I must say that I am more pessimistic than most of you. First, the only oem that announced intentions to build an efficeon machine was Sharp. HP and Fujitsu are only endorsers, whatever that means. Second, Fujitsu had a damned box there, but it was obviously not close enough to a real computer for TM to show it. Third, Fujitsu had a box there and it was not close enough to a real computer for Fujitsu to show it. Fourth, they spent 20 of those 71 pages talking about an idea for the future that could be demonstrated in a mockup. This means that they did not have more to actually show us about the chip. You want to focus on the product in a product announcement. Fifth, the Apple idea is interesting in principle but has virtually no chance of happening in practice. Apple should have done something with other cpus a long time ago. Having failed to do so, they are not about to start now.

In short, the real news in the announcement was the absence of a laptop that will actually ship soon. I will probably wait a little while but I will likely buy a p5 soon. I really wanted a 1 gb 1 ghz efficeon p2 but it looks like I won't be getting one.

bnl

Reply author: christh21
Replied on: 10/18/2003 22:15:05
Message:

Yeah... it's likely that we won't see anything until until the end of the year or later even... I mean they mentioned in the web cast they are ABOUT to ship to OEMs... so OEMs haven't even started production yet. Personally, I'm going to stick with my P2110 until an Efficeon comes out, but I'll probably have to wait a while.

Reply author: oion
Replied on: 10/18/2003 22:29:32
Message:

I tend to agree with blehmann in criticism; I don't recall the webcast part where they mentioned the shipping timeframe, but assuming christh21 is correct - well that just really, really sucks. Consistent with my suspicions earlier in this thread. Grrrr. Despite optimistic benchmarks, part of this much-anticipated release is rocky. Not quite as polished as I hoped. However, let's not bury Transmeta just yet. There may yet be hope (as I look to the Japanese).

Fortunately I'm not in a hurry for a new laptop at the moment.

Damn, I'm tired...

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: mgglim
Replied on: 10/19/2003 02:25:05
Message:

...me too

since I just saw the webcast, I'm hoping that Ditzel remains a real scientist and has not turned into a marketer of the lying order, and that what he's saying can be backed up by real products. I really think he can. He was instrumental in creating the Sparc architecture, and was an early proponent of RISC. And by large measure, it was a success. He realized the folly of Intel in using brute strength, using an inefficient architecture and compensating by adding more and more transistors to increase computational abilities, but at the price of diminishing returns. He was however practical enough to not challenge the x86 architecture. He thus created Transmeta, assembling the best minds, to improve the x86 architecture. Thus, his vision and leadership, has produced Efficeon, a chip that challenges Intel's wrongheaded approach and gives performance in an efficient manner. Now, he still faces great odds against Intel. But technical prowess is on his side. Look at the presentation again and think this through: can Intel be shaking and scrambling?

Can you imagine a $100 chip outperforming the Centrinos that are being sold at 3-4x as much? Will Intel be forced to cut prices and lose the margins that gave them such good profits last quarter?

Forget about Apple and Palm and what I said earlier. Those are mere icing on the cake. Just think the successor to the Fujitsu P, running so much better, bottlenecks removed : fast graphics with nVidia, fast RAM with DDR400, longer battery life. And of course, fast processor. Folks, do you think you will be disappointed this time? This is no cute Fujitsu p2040 with an underpowered engine. Get it?

This is the real thing. Just wait. The chip will come out quickly. Its contract fab, Taiwan Semiconductor, is no slouch. 130 nm is not pushing the edge anymore, yields will be great, and so will the cost.

I think the Japanese sense a great opportunity to trounce the Dell-HP-IBM troika. They smell bloood. They will tell Intel to do the anatomically impossible.

Remember Field of Dreams? Build it. They will come.

Go ahead. Make my day. Buy a P5 and make Intel happy. But will you be? OK, so you can always give that to your daughter.

Reply author: blehmann
Replied on: 10/19/2003 03:04:13
Message:

I hope you are right, mgglim. Unfortunately, the problem in Field Dreams was not getting it built, but keeping it running. The problem here is getting it built. We can all keep it running.

bnl

Reply author: mgglim
Replied on: 10/19/2003 13:02:39
Message:

it's still a story that harks back to David and Goliath.

What is Intel's weakness? Ironically, it's product. This product, processor chips, account for last quarter's roughly 80% of its $7.9 Billion of sales. For a chipmaker, a 21% return on sales is phenomenal. It makes me drool. But I'm drooling at the prospects for Transmeta.

Transmeta only has $2.5 million in sales last quarter. It didn't have a realy competititve product. Now it has. That is its stone. With Taiwan Semi and Fujitsu as its sling, it will beat Intel. It will produce a product that is not only superior, but at least a quarter of the price of Intel's.

Since laptops are the moneymakers for Intel now, and naturally account for the nice profits, think what damage Transmeta will inflict on Intel.

And think at how much Transmeta stands to gain in the spoils of this war. A mental Fermi calculation says Transmeta will be at least a $1.5 billion/year company, making at least $150m per year in one to two years.

And my point is this: with such bright prospects for itself and such a big threat for Intel, I would not bet against Transmeta's not being able to build its business.

So, let Transmeta rain down sweet Efficeons on the barren desert that has been scorched by Intel.

Reply author: christh21
Replied on: 10/19/2003 13:11:36
Message:

Ah sweet prose... well done. My concern is how far will Intel go to shut down Transmeta? Right now they may not be too scared, but if Transmeta really starts doing well... who knows what Intel might do, not to mention they probably are working on another version of their low power processor to fight the battle. I hope Transmeta gets the financial power they need to edge into the market.

Reply author: oion
Replied on: 10/19/2003 13:26:00
Message:

I don't know if "scared" is the right word to use for Intel vs. Transmeta. From a marketing perspective, no company is worth its salt if it ignores smaller companies that will potentially elbow their way into a market.

But I remember this quote: "For its part, Intel admitted at the recent Spring Intel Developer Forum that it decided to develop its Pentium-M processor, designed specifically for a mobile environment, as a response to Transmeta's low-power approach to chip design with the original Crusoe processors." (PCWorld, etc., March 2003, in a Transmeta Astro article)

Scared? Maybe, maybe not. But they're definitely paying attention.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: Crimsona
Replied on: 10/19/2003 13:52:58
Message:

1.5 billion/year revenue? Highly unlikely. AMD is barely a 900 million/year revenue company, and they are much more widespread than Transmeta could ever be in 1 or 2 years

Reply author: mgglim
Replied on: 10/19/2003 14:34:57
Message:

quote:

---

Originally posted by Crimsona

1.5 billion/year revenue? Highly unlikely. AMD is barely a 900 million/year revenue company, and they are much more widespread than Transmeta could ever be in 1 or 2 years

---

Reply author: christh21
Replied on: 10/19/2003 21:57:02
Message:

oion: yeah, I remember reading that same thing about the P-M. Basically, Intel just needs to build a processors that's "good enough" to compete and advertise it like crazy, and edge out it's small competitor. Interesting idea bouat Transmeta combining with AMD... I think I would look favorablly on that, because that means they could implement their approach on a much broader scale and with bigger name recognition, and more dollars.

Reply author: blehmann
Replied on: 10/19/2003 22:37:05
Message:

I think that it is easy to underestimate Intel. The P4 was a mistake in which Intel went for a chip to be marketed by megahertz and not performance. It would have been easy to underestimate Intel at that point and think that they might be caught and easily passed. I made this mistake during one of the episodes in which apple/ibm/motorola went way ahead of Intel in performance; I thought that Intel would be placed in the position of permanently playing catchup. Instead, Intel performed a major overhaul and came up with a version of the Pentium that was more than competitive in terms of performance.

The P4 was a mistake but Intel has done an excellent job (much as I hate to admit it) correcting that error with the M. The 0.9-1 ghz ULP M that powers the P5 is much more competition for the still forthcoming Efficeon than I would have thought possible a year ago. A year ago, I certainly would not have imagined the bevy of M-based subnotebooks that is now available. The desktop M's completely correct the flaws in the P4 by going back to the better base architecture of the P3.

Intel is something of a monopoly and, as such, it engages in heavy-handed marketing tactics that makes it appear as though it is only successful because of its position, not its products. This makes it easy to overlook Intel's long record of meeting competitive threats at times when numerous pundits were quite skeptical of its ability to do so. It is easy to overlook the fact that the M is a pretty good chip, not merely an adequate one. It is certainly better than "good enough."

In truth, somebody should have acquired TM a long time ago. TM should specialize in its comparative advantage: making near hardware code (the analogue of microcode in old mainframes; a way in which I am dating myself) that solves problems that are hard to fix with hardware alone. If TM made the code morphing and anti-leakage software while AMD designed, built, and marketed the chips, the result might be a serious competitive to Intel.



bnl

Reply author: mgglim
Replied on: 10/20/2003 00:22:38
Message:

Intel owes its success up till now to luck and mindshare. Luck- because IBM chose Intel chips to be used to counter competitive threats from Apple and from the CP/M OS using Z80 chips. From this, Intel was able to parlay itself into dominance by owning the x86 architecture.

IBM and Motorola's PowerPC, DEC's Alpha, and Sun's Sparc could not beat Intel, just because they presented competing architectures, that even if superior (arguably), that could not overcome the momentum of the x86 architecture and the code base that is invested in it. So, Intel could just be mediocre, and it truly still is, and still easily stumble and regain lost ground.

AMD and Transmeta is using the same x86 architecture and doing it better. Intel's mediocrity is now starting to show. The next few quarters will be its last hurrah. It cannot take cover behind the mighty x86 fortress, as the fortress itself is devouring it.

This internecine struggle itself is strengthening the x86 architecture, playing to the demise of the Sparc and possibly, the PowerPC.

With little competition left, it can be easy to argue why Transmeta and AMD will capture a large share, if not the lion's, of the CPU market.

It will also be hard to see Intel, steeped in its old ways and habits, go any better than Xerox, which ceded its copier dominance once it lost its patent protection, and it has been touch and go ever since.

Reply author: mayukawa
Replied on: 10/20/2003 01:55:59
Message:

I think you're underestimating Intel. They have billions spent on R&D. The latest semiconductor technologies need a lot of money to implement (process plants) as the industry go toward smaller and smaller traces. With so much clout, Intel can play hardball with chips just like Microsoft can with Windows. I can see AMD gettting a chance to take more market shares from Intel...but Transmeta? Transmeta's CPU design is inherently designed for a niche market (very low-power applications), so unless they can also compete on performance, I doubt they'll move out of this niche market.

Reply author: mgglim
Replied on: 10/20/2003 09:02:48
Message:

quote:

---

Originally posted by mayukawa

I think you're underestimating Intel. They have billions spent on R&D. The latest semiconductor technologies need a lot of money to implement (process plants) as the industry go toward smaller and smaller traces. With so much clout, Intel can play hardball with chips just like Microsoft can with Windows. I can see AMD gettting a chance to take more market shares from Intel...but Transmeta? Transmeta's CPU design is inherently designed for a niche market (very low-power applications), so unless they can also compete on performance, I doubt they'll move out of this niche market.

---

Reply author: oion
Replied on: 10/20/2003 09:12:52
Message:

quote:

---

The next few quarters will be its last hurrah. It cannot take cover behind the mighty x86 fortress, as the fortress itself is devouring it.

---

Reply author: oion
Replied on: 10/20/2003 09:16:54
Message:

quote:

---

So I may have, as did Rome was underestimated. Or was GM against Honda back when Honda just made motorcycles. Or was Motorola against Nokia. Or even say, Reichhold Chemicals, who eventually got eaten up 15 years by Sun Chemical, a former joint venture subsidiary in Japan.

---

Reply author: mgglim
Replied on: 10/20/2003 09:31:46
Message:

Companies run by schmoozers eventually fail. They have misguided hopes fed by the rah rah's and pom poms of their leaders, who are given to loud talk and speeches heavy on emotions but sorely lacking in substance. For it is vision and leadership they lack, and find comfort in leaders that can give them the illusion of a secure future. Changing the leadership is futile, as their leadership embody the culture and preferences of its constituents. It is much better to create a new company to challenge the old, than try to make change within.

Intel had its run. It did its job. It will be forced by history and by circumstance to cede the torch.

Reply author: oion
Replied on: 10/20/2003 09:44:28
Message:

Without being overly dramatic and grandiose about it, I'd say it's too early to say when Intel will fall, even if at all. If there does become internal unrest, it's up to the business heads whether they decide to reorganize, concentrate on new products, etc. That's up the in air, Intel may or may not make the correct decisions, but one way to look at it is this - yes, the leadership may be bent to their consumer will. However, at the same time Intel may use their incredible marketing clout to make these same constituents change to a different technology. "We are Intel. You know our products are the best. You shall accept our new and even better stuff." A classical business consumer may be more willing to take Intel changes and spinoffs than an entirely "new" and untested brand.

No point in placing Intel in preterite form just yet... We have yet to see if the TM8000 will even be adopted in further OEM designs.

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: mgglim
Replied on: 10/20/2003 10:25:23
Message:

Rome did not fall in a day. GM still lives to this day. Motorola still showing signs of life. Xerox is on ife support. Digital Equipment is a mere footnote of history. IBM alive and kicking, only in a different business- services. Intel, a fab contractor for Transmeta chip? Hmmm...

Reply author: mgglim
Replied on: 10/20/2003 10:36:06
Message:

Folks,

http://www.theinquirer.net/?article=12234

This proves Transmeta is on to something. We may be on to something. I'm missing Intel in the news. It may still be that Intel has a very effective way of keeping its trade secrets, naaahhhh!

Reply author: mayukawa
Replied on: 10/20/2003 15:36:12
Message:

I don't know...I personally think that the only other company that can compete with Intel on the semiconductor front is IBM. I think a lot of us want to support the underdog (to maintain competition in the market), but I'm not sure Transmeta will be the David to take down Goliath...

Reply author: oion
Replied on: 10/20/2003 17:19:34
Message:

I don't think that article actually "proves" anything, but it is interesting. Let's not let all this grandiose anti-Intel sentiment get out of hand, even if I don't particularly like them. Wasn't AMD supposed to take down Intel? But really, I'm only (mostly) interested in seeing TM8000 news here.

Van's Hardware will hopefully have an article by next week (same place as the earlier roadmap article).

______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: mgglim
Replied on: 10/20/2003 23:20:48
Message:

Just so you know, Fujitsu is clearing out its P2120's, along with other items on sale. Is this a sign they have new products based on the Efficeon coming out? I just can't wait.

Even some P50xx's are being cleared out. This is even more intriguing.

Reply author: jherber
Replied on: 10/21/2003 12:20:49
Message:

quote:

---

Originally posted by mayukawa

I think a lot of us want to support the underdog (to maintain competition in the market), but I'm not sure Transmeta will be the David to take down Goliath...

---

Reply author: mgglim
Replied on: 10/21/2003 15:32:38
Message:

human beings rule. dinosaurs don't. however, the ebola can be very scary, even if you rule.

Reply author: oion
Replied on: 10/21/2003 15:49:56
Message:

(.... Whaaaaat?? )

Er, anyway... I need to stock up on P-2000 batteries. Hm, but the bay is still compatible with other models. Guess I'd better go visit their site to see what they're doing. Perhaps even scrounge ebay.

It's too bad Comdex is already past; if Transmeta had released earlier, we might've gotten interesting notebooks to look at there. Possibly even awards - more awards helps sales. Oh well.

Anyway, as a side note, at least it looks like the AMIBIOS8 supports USB1.1 and 2.0 host controllers; USB keyboard and mouse support at POST, setup, and runtime (legacy emulation); and Boot and/or BIOS recovery from USB mass storage (HDD, floppy, ZIP, LS120/140, MO, flash). Just a quick read-through of the BIOS that was paired with the Efficeon; hopefully the "optional" settings will be given by Fujitsu or some other OEM. Full BIOS support would be nice, anyway. I wonder what BIOSes have firewire support. (Don't know the details of P-5k's BIOS.)


______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

Reply author: oion
Replied on: 10/22/2003 16:41:00
Message:

Uh, this wasn't specifically posted, but it should've been...

http://www.theregister.co.uk/content/3/33426.html

Current financial analysis of Transmeta. Doesn't look very optimistic.

______________________
America's favorite news source

Reply author: gellul
Replied on: 10/22/2003 17:25:36
Message:

Do you think Fujitsu is going to stop selling the primary battery for the P2120?

quote:

---

Originally posted by oion

(.... Whaaaaat?? )

Er, anyway... I need to stock up on P-2000 batteries. Hm, but the bay is still compatible with other models. Guess I'd better go visit their site to see what they're doing. Perhaps even scrounge ebay.

It's too bad Comdex is already past; if Transmeta had released earlier, we might've gotten interesting notebooks to look at there. Possibly even awards - more awards helps sales. Oh well.

Anyway, as a side note, at least it looks like the AMIBIOS8 supports USB1.1 and 2.0 host controllers; USB keyboard and mouse support at POST, setup, and runtime (legacy emulation); and Boot and/or BIOS recovery from USB mass storage (HDD, floppy, ZIP, LS120/140, MO, flash). Just a quick read-through of the BIOS that was paired with the Efficeon; hopefully the "optional" settings will be given by Fujitsu or some other OEM. Full BIOS support would be nice, anyway. I wonder what BIOSes have firewire support. (Don't know the details of P-5k's BIOS.)


______________________
To have no errors
Would be life without meaning
No struggle, no joy
-- Brian M. Porter (from the Salon 21st Challenge: error haiku contest)

---

Reply author: oion
Replied on: 10/22/2003 20:04:07
Message:

You know what, originally I was going to say "probably" - because the P-2000 series is discontinued, but then I double-checked and the main/hicap batteries are compatible with the P-1000 series, which they're still selling. Not to mention the bay battery is indeed compatible with several other Fujitsu models. Guess there's no hurry now.

______________________
America's favorite news source

Reply author: mgglim
Replied on: 10/22/2003 21:45:25
Message:

What has the world come around to? No leaks of any sort at all? Even if Transmeta staged one, I wouldn't mind. This quiet period has the elements of a tornado brewing. And we can be thankful that we are well-stocked with batteries :-)

Reply author: oion
Replied on: 10/23/2003 00:15:35
Message:

It's sad that they're expecting to lose money in the next quarter already. Halved? That's just sad. I don't think it'd be too far-fetched to say, if the TM8000 doesn't fare well, it's the end of Transmeta. Though, perhaps their technology will be bought... hmm.

Just as their pre-Efficeon press releases were suspiciously silent, it still is now. This doesn't bode well at all; design wins are always publicized. I suppose it's really only up to AMD now.

______________________
America's favorite news source

Reply author: wright45
Replied on: 10/23/2003 01:48:33
Message:

Before the Efficeon release, the only thing I had to compare a chip release with was the PentiumM release. I had heard of Centrino (or Banias) before the release, but the day that I actually looked into it (searching the web and all that) just happened to be its very launch day. And that day a number of new systems were announced with the PentiumM. I was hoping that Transmeta would do the same for the efficeon.
Now I realize that, if Transmeta did have any OEMs ready to sell a product with their chip, they surely would have released the product. But since they didn't, that tells me that the efficeon launch day and what they did to promote it was all to sell it to the OEMs, not consumers like most of us. And if thats true, then thats not good (this processor has been rumored about for about a year right?)
IMO the best thing for Transmeta right now is to get the efficeon systems out ASAP, and advertise loud and clear that they are available. And I'm sure if they could, they would.

I know there's nothing much new said here, but I thought I'd share.

Reply author: mgglim
Replied on: 10/23/2003 02:59:30
Message:

that was a good assessment and I agree with you there must be something wrong with not having announcements on shipping products during the chip launch. they are behind on their 3rd quarter 2003 release date, and I wonder if that LongRun2 chip leak control announcement is an effort to deflect their failure to meet the timeline.

since there is a lot of software involved in the chip, I'm hoping the delay can be fixed through a software revision rather than on process yields.

nvidia lost a lot of ground this year, and I read somewhere that it was because they went out for broke by going with an unproven 90nm process,and yields weren't good and this caused them to not ship. i hope transmeta didn't end up having process problems once again just because they had put too many things into their chip like agp, hypertransport, etc. doing that gave them many points of failure, and if they failed to manage these critical problem sources, that would cost them dearly. i wonder if there is a cub's fan out there who can save them.

Reply author: thetenken
Replied on: 10/24/2003 17:36:04
Message:

As discussed in the Other Subnotebooks forum in this thread:

http://www.leog.net/fujp_forum/topic.asp?TOPIC_ID=3891

Seems like the first major design win for the Efficeon, a new Sony ultraport, the Sony PCG-UX5.

Ripped shamelessly from lawrencecamera's post:

Transmeta Efficeon CPU 1Ghz
Nvidia nForce3 Go 120 Display
7.1" XGA CG Display (1024 x 768)
256MB DDR SD Ram
40GB HDD
Built-in WiFi 802.11b/g (11 & 54 mbps)
2 x USB 2.0
OS: Windows XP Home (SPI)
Bundled Software:
Thumb Phase 3.0
VAIO Sychronizer 1.2
Price: 169,800 Yen
Date available: Mid December
About the model number: ?
PCG-UX5 Ubiquitous Extreme 5
Here is link from the Chinese User Group with the post:

http://www.hk-vaio.com/forums/viewtopic.php?topic=7387&forum=1&6

Reply author: NigelS
Replied on: 10/24/2003 19:27:13
Message:

That's pretty darn exciting. Oh well, if Transmeta can't count on Fujitsu for support, ya can't beat Sony for big name clout! (Altho' it feels like Luke giving in to James Earl Jones' sales pitch for the dark side...)

Reply author: mgglim
Replied on: 10/24/2003 19:47:17
Message:

Mid-December? Oion, you guessed right. Just barely in time for Christmas, late for Hannukah and Kwanzaa. But will make it for Chinese New Year.

Reply author: thetenken
Replied on: 10/24/2003 20:55:54
Message:

Who knows if we'll see this stuff on our side (U.S.) of the pond before Christmas, or will it just be asian/Japanese releases?

Reply author: blehmann
Replied on: 10/24/2003 23:03:01
Message:

This is indeed good news. I really don't understand why Sony did not announce at the TM rollout. It would appear to be in their best interest. It also makes me more uncertain as to how to interpret Fujitsu's silence. My presumption was that no product announcements meant that either (a) manufacturing was still sufficiently dicey that reliable seed units were not available for the oem's to work with or (b) oems had evaluated efficeon units and found them wanting. I guess I will give Fujitsu a little more time to announce before I buy a p5.

bnl

Reply author: oion
Replied on: 10/25/2003 00:16:46
Message:

What I'm more interested in is why Transmeta didn't announce this, even if Sony didn't. Sony doesn't really have much to lose compared to Transmeta; if their line falters (which it apparently hasn't in Japan), they'll just rely on other lines and chips. Unless there was a blurb on the Japanese Transmeta site (assuming there is one)... I suppose a chip iteration in an existing line isn't really a design win. Still, one might think it would've helped Transmeta marketing, in which case it seems more like a flub. Since the Sony U-series wasn't released to the US, I'm assuming the UX5 won't see real retail light of day here either. Normally I don't expect Transmeta to separate international markets - after all, they did list international models on their site. It's one thing for Sony to not market the U series here in the US, but an entirely different thing for Transmeta not to add it to their "resume." Really makes me wonder. Or, perhaps Transmeta does intend on mentioning this closer to the UX5's release.

As for other OEMs, we can only wait with bated breath; if more models aren't announced around the Xmas season or very soon thereafter, I don't know what to expect from Transmeta. I'm sort of expecting the worst at that point. Time is against them, along with Intel.

______________________
America's favorite news source

Reply author: mgglim
Replied on: 10/25/2003 00:59:01
Message:

My own 2 cents on this is the uncertainty of Trasnmeta's production timeline. Hence Sony is jumping the gun for "me-first" mileage but only by talking of a late release date. This means that Efficeon chips are a matter of when in the near future (1-3 mos.) and not a matter of "if it will ever be produced."

So Fujitsu and Sony, as well as Sharp and Toshiba, and even HP, are playing safe. They don't want potential buyers in the October to December quarter to withhold any notebook purchases in anticipation of a near-availability of an Efficeon notebook. Transmeta,in order to be on the good side of its OEMs, is cooperating by not feeding on this expectation, just in case something beyond its control happens. And not being in control has to do with having a very cutting edge chip challenging the production process capability of Taiwan Semiconductor. Transmeta, by being less verbose, is guaranteeing its future credibility with its OEMs and its end-users, us.

I, for one, while not yet in the market for a notebook, am telling friends to hold off buying notebooks until the Efficeon comes out. Many people already have laptops, they can still deny themselves a few months of gratification just so they do not regret buying notebooks that are going to just be passe a few months down the road and sold for a song on Ebay.

Reply author: Alt
Replied on: 10/26/2003 00:03:10
Message:

quote:

---

Originally posted by oion

What I'm more interested in is why Transmeta didn't announce this, even if Sony didn't. Sony doesn't really have much to lose compared to Transmeta;

---

Reply author: oion
Replied on: 10/26/2003 10:32:50
Message:

Hrm, leaks. I just don't trust them. Guess that model is just a theory to me, though Sony did/does have several Transmeta models that have done relatively well in the (jp) market. Of course, I'd like to know who leaked that info in the first place... No idea about Sony owning Transmeta shares, though I wouldn't be surprised. On the plus side, Sony does have a penchant for small innovative designs, even while a bit too proprietary sometimes.

______________________
America's favorite news source

Reply author: oion
Replied on: 11/03/2003 11:01:53
Message:

It's not on the Transmeta press release page yet, but they're going to have a live presentation at the AeA Classic 2003 conference in San Diego tomorrow and Wednesday, Nov 4-5.

http://www.oion.net/e/TMTA110303.pdf

http://www.aeanet.org/events/fstm_Classic2003HomePage.asp
______________________
"Be what you would seem to be - or if you'd like it put more simply - Never imagine yourself not to be otherwise than what it might appear to others that what you were or might have been was not otherwise than what you had been would have appeared to them to be otherwise." - Lewis Carroll

Reply author: oion
Replied on: 11/09/2003 22:50:53
Message:

Perhaps no one is reading this anymore (and I apologize for unilaterally resurrecting if that's the case ), but those who didn't bother with the first webcast can now see the proposed TM8000 roadmap converted on my site. The Centrino roadmap is here.

Reply author: christh21
Replied on: 11/10/2003 00:45:25
Message:

Looking at the Centrino roadmap, it looks like Transmeta might pull ahead in the next 2 years. The Centrino only has plans up to 1.7ghz so far while the Efficeon is looking at 2ghz by the end of next year. Also, I can't image that the 1.7ghz centrino will be as energy efficient. Efficeon 1.6ghz will run at 7w. If Transmeta has the cash to hold out for a few more years, they just might make it. Also, Transmeta's website indicates that at least 1 OEM is now in production with the Efficeon chip, and they expect 12-14in laptop production to start by January.

Reply author: blehmann
Replied on: 11/10/2003 13:32:12
Message:

Oion, I for one, am always interested in this thread. It is the first one I look for when I look at the active threads when I log on to this site. That said, it is the overwhelming lack of specific design wins and other product announcements regarding efficeon that I find so disheartening. Part of this is doubtless due to securities law that makes it dangerous to make some kinds of announcements; this is the reason for all of the boilerplate language at the end of all of their press releases that falls under the rubric of safe harbor statements. However, TM can say more than vague statements that they expect notebooks to be shipped based on the efficeon chip in the first quarter of 2004. This statement has absolutely no information content; TM can say it irrespective of whether they have specific knowledge of notebook manufacturers adoption plans. I think that people understand this point when they ignore statements like "With this launch, Transmeta pushes Efficient Computing to a new level, through major innovations in its proven hardware and software architecture, increased on-chip integration and new high performance interfaces." This sort of "new era" language that the pr department always comes up with.

It is the lack of specifics that is depressing. It is the announcement on the tm website that oion noticed regarding Perry (the tm ceo) speaking at the AeA conference in San Diego that is not followed up by any information as to what he said and what its implications are for the future of efficeon. Sharp and maybe Sony will be shipping Efficeon machines soon but no news is not good news in this setting.

Finally, I am worried by statements that tm might pull ahead of intel in the next two years. The lesson of the competition between intel and the powerpc chip family is that intel is quite capable of catching up extremely quickly in technological arms races. If Intel thinks that it is at a pronounced competitive disadvantage, it would not be wise to bet strongly that they will fail, however much I might wish for it to happen.

bnl

Reply author: oion
Replied on: 11/10/2003 14:55:06
Message:

I need to see a real unit for sale in December - be it here or Japan - or I'll likely just give up on Transmeta. I listened to the AeA conference but there wasn't that much interesting information in it. Perry mentioned that they're concentrating on the Japanese market. That's both a good decision and the only decision, given there seems to be very little interest in the US. Perry mentioned, twice, that they have 3 design wins with HP; I assume those are blade servers. However, I don't know if he was talking about past Crusoe wins or current Efficeon wins; I'm assuming the former, with the potential that HP will upgrade to TM8000. In mentioning Transmeta's intent to spread its marketability into mainstream laptops, Perry commented that the Crusoe had been limited to 10"-under subnotes, which is only 8% of the entire laptop market. Imagining Intel taking another large percent of that 8% on top of whatever percent they already have in mainstream is a little disheartening; Transmeta really has their work cut out for them..

I don't really trust roadmaps myself, but there were some comments about future Transmeta designs so I added it to the page, though right now I'm thinking of removing it! Perhaps I'll just annotate a bit more. So much can happen in a few months. Remember the TM5800 debacle in 2001 due to delayed release...

Intel isn't about to sit still, either, considering how popular Centrino is. I'm still hoping, assuming in worst-case-scenario that Transmeta tanks, that the technology will be sent onto AMD or another larger corporation with better means to compete. Or perhaps it's still too early to speak of it.

Reply author: christh21
Replied on: 11/10/2003 19:56:25
Message:

One frustrating news coming from Transmeta is their new technology licensing idea. They appear to plan on licensing LongRun2 to other chip companies, probably AMD, but I doubt Intel will fall for it. It's a good business strategy in once sense, but on the other hand it also shows that Transmeta is really hurting. Did you all see Transmeta's earnings report? $2.7million, that's the least they have ever sold, they sold more 3 years ago. They are going downhill fast. At least if they license the technology, even if they tank consumers will still benefit.

Reply author: oion
Replied on: 11/10/2003 20:03:33
Message:

Right; their stock has been doing dismally for a while, and they don't have good forecasts for this quarter either. I don't know how their shares would react to one or two design wins at this point, but it'll be interesting to watch either way. TMTA dies, technology is sent where? TMTA doesn't die, we get more designs. But the waiting stinks!

Reply author: blehmann
Replied on: 11/11/2003 01:07:46
Message:

Without intending to sound ghoulish (in a corporate sense), the problem with a growth company dying slowly is that it does not engage in proper care, feeding, and nurturing of its r&d portfolio. If tmta dies quickly, someone will buy the technology and we can reasonably hope that they will know what to do with it. If tmta dies slowly or limps along, their r&d portfolio will be standing still while others will be making progress, most especially intel. I do hope I am unduly pessimistic and I do fear that I am not.

bnl

Reply author: oion
Replied on: 11/12/2003 09:37:12
Message:

A press release today (that I can't seem to find on their site at the moment) about a partnership with Xybernaut for wearable computing:

http://oion.net/e/TMTA1112.pdf

While this is a more solid press release about an actual partnership for products instead of a "this industry company supports us!" blurb, it's still not a design win per se. Maybe it's getting there...

Edit: The press release as duplicated by Xybernaut is here, though the URL year is wrong:

http://www.xybernaut.com/company/public/press/2002/press_release.asp?PRESSID=202

Reply author: blehmann
Replied on: 11/12/2003 15:21:58
Message:

Sorry, but this is a thud, not a wow...

bnl

Reply author: newpbx
Replied on: 11/13/2003 14:02:26
Message:

I think some of us are being too hard on Transmeta. They released a really new product, Efficeon TM 8000, only a little behind schedule and it promises to be a significant performance upgrade over their previous TM 5800 processors.

Contrast this with Intel, who constantly announce a new processor which is only marginally faster than the one it replaces and sometimes slower for the same clock rate (P3 is faster than the same clock rate P4 for the early release versions).

The Version 1 Efficeon may be a place marker, whose role is to allow laptop and mobo manufacturers to "kick the tires" on TM's new technology, and also let TM insure that it is compatible with major operating systems, and optimized for CPU-hungry aps like Photoshop and MPEG coding software.

The Version 2 Efficeon will no doubt include performance and power saving improvements that were not fully tested for the initial announcement, and which may emerge during the Version 1 tests.

I for one, hope that TM will take their time testing to get a clean release, unlike other industry vendors who push questionable products into the marketplace with only cursory testing. (Think about a large software vendor most of us know).

Reply author: oion
Replied on: 11/13/2003 14:35:15
Message:

Good points, though Transmeta is at a very difficult position right now; at least their stock had risen a bit after the initial release. Intel has already taken much of the subnote market with its first release of Centrino, and it has far bigger branding power than TMTA. Those OEMs who went with Centrino generally have successful Centrino products now (why change a good thing?). In a competitive industry like this, being late is enough to kill a company that was already floundering to begin with. At this stage it looks like they may have a little more time, but not much at all.

While it's true that the Efficeon is a whole other class over the Crusoe, everyone tends to compare to Pentium-M, not Crusoe, and even with those benchmarks the first Efficeon is barely par with the lowest P-M iteration (900MHz) - though obviously most of us who read these forums care for more than mere *hertz speed. I don't know about details, but it also wouldn't be far-fetched to assume whatever contractual obligations and special deals certain OEMs have with Intel will blunt Transmeta attractiveness. Who knows just how dirty Intel plays, though.

As for marginal iteration releases as Intel has done - mind you many consumers wouldn't know the difference between clock rate and salami besides 'bigger is better.' (I'm sure these are the same people who always buy SUVs. kidding, kidding.) I suppose Transmeta has the added work of debugging software instead of working on an all-hardware architecture as other microchip companies do, but this doesn't mean they can allow themselves the luxury of too much time. Very difficult balancing act - great product, tight timetable, low cost - just to compete with Intel. I'm not saying it's impossible (indeed, I really want Transmeta to succeed), but it's an uphill struggle.

I'll remain completely cynical and pessimistic until I see the first retail units. After which I'll (probably) be drooling and scrambling after the latest cool models.

Reply author: thetenken
Replied on: 11/14/2003 14:30:53
Message:

An article rehashing Transmeta's differing approach for the Efficeon to the Crusoe on EETimes.

http://www.eet.com/semi/news/OEG20031113S0017

Reply author: Alt
Replied on: 11/14/2003 21:59:20
Message:

Thats a nice article.

|Tab

Reply author: jeffro
Replied on: 11/15/2003 22:26:43
Message:

That article doesn't make me feel encouraged for Transmeta. If performance/power is only equal to Intel, then even if Efficeon costs much less it will be hard to gain market share. Intel can always resort to price slashing if they want Efficeon to go away in the meaty segment of the notebook market, and leave Transmeta fighting for scraps as they have been. However, I'm interested to see if Efficeon sparks more interest in silent desktop computing, since Intel seems determined to keep Pentium-M out of desktop computers.

Reply author: oion
Replied on: 11/18/2003 16:35:19
Message:

HA HA MADE YA LOOK.

http://investor.transmeta.com/news/20031118-122940.cfm

"Antelope Technologies and Transmeta Announce the Availability of the First Product Based Upon New Modular Computing Architecture"

Posted today, but the most interesting part is that the new design is with the Crusoe, not the Efficeon. (And I thought they gave up with the Crusoe by now. ) I suppose that means they had been working on it for a looooong time, well before Efficeon samples or so.

jeffro: The important issues for me will be heat and battery power, with performance coming dead last in my list. I feel this is true for every P-1k/2k owner out there who actually likes their P-series. That doesn't mean performance isn't important, just that for me I want performance to be on par with other chips in the subnote niche, but with better everything-else. Having the nvidia graphics subset would help with other things, too.

Reply author: jeffro
Replied on: 11/19/2003 11:34:56
Message:

Maybe my post wasn't clear. I was talking about performance per power consumption. The article said this is more or less equal between Efficeon and Pentium-M. So I don't see Efficeon's advantage in heat & battery life. The only real advantage appears to be in production cost, which does not seem enough for Efficeon to really take off, at least in mainstream notebooks.

Reply author: oion
Replied on: 11/19/2003 13:18:12
Message:

Sorry, I interpreted your post as just performance or power i.e. computing power, my bad. I can see where you're interpreting that the TM8000 and P-M are equivalent in perf-per-power - you're talking about the 7-watt benchmarks, yes? The article I think summarizes those as par. But there are two variables which Transmeta intends on using to up performance on the same power next year: 90nm process and Longrun 2. I don't know the details about those, though. Efficeon does have much better idle consumption than a P-M, however, so that may play into the whole picture, but there have yet to be 3rd party tests, which is what really matters - but the EE article didn't even gloss over that. The advantage in heat and power consumption is more obvious when looking at the proposed Transmeta roadmap; one question is whether Intel can maintain their performance on a lower wattage. The EE article, mind you, didn't really say anything new; the information was from past press releases, general Transmeta information, and the conference, but they didn't provide all benchmarks. The real advantages are not merely (potentially) production costs, but the better graphics subset and much smaller package, along with proposed power savings and the 7 watt limit. Are there fanless Centrino laptops? (dunno) Of course, this is all theoretical until actual retail units appear. It's mid-November already...

Reply author: jeffro
Replied on: 11/19/2003 13:39:36
Message:

I'd like to believe Transmeta has a good roadmap and Intel's up against roadblocks, but given how late Intel got into the efficient processor game and dominates, it's hard to bet against them.

The Efficeon news articles usually mention fanless desktops as a targeted market. If a market materializes there (particularly on corporate desktops), it could get interesting for Transmeta and Efficeon. Low production cost would be a real advantage and if Intel entered the game with Pentium M they'd have to do so at the expense of their bread & butter Pentium 4.

Reply author: oion
Replied on: 11/19/2003 13:49:50
Message:

I don't know about the desktops, but it's possible; right now I think they'll do best in the blade server market and mini-notebooks/other small things, tablets. Probably subnotebooks as well, but that depends on getting at least 1-2 heavyweight names with large markets in the US. If people running the hot and noisy server farms discover a great heat/noise advantage from Efficeon along with performance, then it would be more natural for the application to go to desktops.

It's probably safest to bet on Sony coming out with an Efficeon laptop model first, since they were the first with Crusoe as well.

Reply author: christh21
Replied on: 11/19/2003 15:29:00
Message:

You mentioned that the Efficeon and Pentium-M has similar performance at the same power consumption. You asked "where's the advantage"... the advantage is in lower than maximum perfomance situations. The Pentium-M has 2 settings> full speed (900mhz) and slow-speed (600mhz). The Efficon can dynamically change speed in several steps/voltages... all that we know.. So don't forget that most power savings comes from less than 100% CPU usage. The point of Efficeon is to provide equal performance at max levels, and much better power efficency and less than max. Don't forget that.

As far as silent desktops... what about dual-Efficeon systems? You could still have a silent/fanless (or at least minimal heat control) and have much more processing power.

Reply author: Crimsona
Replied on: 11/19/2003 17:26:26
Message:

The Pentium M actually has 7-11 steps in terms of speed settings. Can't find the reference or the exact number, but i'm pretty sure I'm right. That's as close to dynamically changing as you're going to get without creating 1 step for every mhz I guess.

Reply author: mm42
Replied on: 11/19/2003 21:45:50
Message:

Pentium M number of operating states depends on the max speed.

The 1.6 GHz chip has 7 speeds (and corresponding voltages), every 200 MHz from 600 MHz to 1.6 GHz.

The 900 MHz chip has only 3 speeds, 900 MHz, 800 MHz, and 600 MHz.

Reply author: thetenken
Replied on: 11/19/2003 22:24:22
Message:

quote:

---

Originally posted by oion

Are there fanless Centrino laptops? (dunno) Of course, this is all theoretical until actual retail units appear. It's mid-November already...

---

Reply author: blehmann
Replied on: 11/20/2003 01:47:36
Message:

The other thing about the (alleged) heated battle between the Efficeon and the M is that Intel is not standing still during the Efficeon delays. Intel has done a great job of playing catchup in the past (unfortunately).

bnl

Reply author: christh21
Replied on: 11/20/2003 13:16:57
Message:

That's interesting... I didn't know that P-M had several steps. But, in defense of Transmeta, Intel's SpeedStep only changes speed after prolonged time at above/below a certain level, and then switches. Someone else mentioned that above, while the Transmeta can do that dynamically "100s of times a second" I believe is what Transmeta has stated. It does seem that the GAP is narrowing, but Intel still refuses to design a completely new processor. The P-M is just a modified P3, they haven't made anything new in a long time.

I'm curious how Intel plans to address the voltage leakage problems that Transmeta claims to have overcome with LongRun2. If Intel can't be innovative enough to overcome that, then they may find themselves falling behind. At least we can see that competition is necessary to bring about innovation. (anyone thinking Microsoft?)

Reply author: oion
Replied on: 11/20/2003 13:27:10
Message:

OT a bit, but the problem with Microsoft is that it's the interface layer between a person and the hardware. A company can change hardware relatively easily if they wanted to (not talking about costs and logistics), but to change the operating system or even the major office application suite requires more money and time (and complaints) for training on how to use the thing. So, even if, say, a Linux distro is just as easy to install and is just as functional (supports as many applications) as a Microsoft OS, the fact that business people are so damned used to M$ will be the major obstacle, not the quality of the software itself. As for Transmeta vs. Intel, however, the hardware layer is invisible - a user really doesn't see it or interact with it directly. Innovation seems easier when you don't have to retrain the average joe/jane how to use it. Obviously then we get to the whole branding issue, but it's still easier for a company to strike forth with new designs in that regard. Intel doesn't need to really innovate while it still has (whatever huge percent) of the market.

Re: W2
Another point for the W2... I'm still eyeing that while I wait for an Efficeon retail unit.

Reply author: mgglim
Replied on: 11/20/2003 14:55:36
Message:

http://arstechnica.com/cpu/003/mpf-2003/mpf-2003-1.html

Finally, an article on new processors coming from Jon "Hannibal" Stokes" from information gleaned at the recent Microprocessor Forum.

Reply author: oion
Replied on: 11/20/2003 15:24:21
Message:

Interesting read, nice to get some new discussion out there, though this may be a bit more technical than business-minded for some people. Most of it is still theoretical in application, unfortunately (especially concerning Efficeon). I definitely agree with "All in all, Efficeon is a welcome improvement over Crusoe, but I really expected something like it to be out much sooner." *tick-tock*

Reply author: mgglim
Replied on: 11/20/2003 20:27:21
Message:

I glossed over the technical, but I think the discussion was framed in terms of performance, and so it was not surprising to find the Efficeon failing in that regard.However, my expectation of the Efficeon lies elsewhere, and it's in a good ratio of computing strength to power consumption. In the realm of power consumption, it still points to the Efficeon as being a viable, if not superior,alternative to Intel's architecture.

What's not lost on us is, as stated also in the article, how Transmeta can emerge from its underdog position to counter Intel's marketing dollars, to convince us consumers to buy computers based on its processors.

Even if the Efficeon has its merits, it would be easier for Transmeta if it could portray its processors as unquestionably far superior to its competitor Intel's. If there is any shadow of a doubt as to its superiority, Intel's marketing dollars could easily vanquish any message or solution Transmeta is offering.

Fortunately, there is still a lot of mystery in the black box called the Efficeon. Let's hope that as the mystery is unraveled, the Efficeon will come out shining.

Reply author: wright45
Replied on: 11/20/2003 20:28:54
Message:

Just a little blurb about Intel:

http://www.dvhardware.net/article2133.html

"Taiwanese notebook makers have confirmed that Intel will introduce three new models of its next-gen Pentium M CPU (codenamed Dothan) on the 4th of February 2004."

"Intel will also debut its Pentium M 1GHz ULV processor at a price of around US$262 per unit in the first quarter of next year. The new 1GHz ULV model is to counter Transmeta’s recently released 1.1Ghz Efficeon processors, said the sources."

I wonder if Efficeon systems will be out by 02/04/04

Reply author: oion
Replied on: 11/21/2003 10:39:35
Message:

Efficeon systems really must be out by December, IMHO, given its public release in October. Intel's release of Centrino was followed very closely (if not in the same press release) by Centrino laptops. January would be the absolute latest I'd hope for, but frankly at that point I'm quite ready to give up on Transmeta. Remember that bulk production to OEMs was supposed to be by August! Efficeon was just released too late. Too late, possibly not enough. But then there's still hope from Sharp and Sony, the question is when (and in Sony's case, whether it's not just a rumour).

Though it's still nice in a strange way to hear Intel releasing stuff "in response to" Transmeta.

Reply author: mgglim
Replied on: 11/21/2003 15:49:12
Message:

This article, I hope, won't be too technical. It gives us wary onlookers much hope in wishing Transmeta success in proving that indeed it has a much much more superior solution in terms of performance/power consumption metrics.

http://www.acmqueue.com/modules.php?name=Content&pa=showpage&pid=80&page=4

To sum it up, this article shows Transmeta processors used in supercomputing clusters, called Green Destiny, and illustrates how well it performs on metrics the writer deems more appropriate. In terms of power efficiency, reliability, and cost savings, the use of Transmeta processors is compelling. Moreover, it was interesting to see that the lab used their own CMS (code-morphing software) over that of Transmeta's to give a boost of 50% to computing strength. I see two implications here:

1) Transmeta must still be fine-tuning its CMS now and would rather risk launch delay than risk any chance of faulty code ruining its launch, as first impressions are almost impossible to recover from (remember Apple Newton), especially given Transmeta's disadvantaged position vis-a-vis Intel's marketing dollars; and

2) Tailoring CMS to different end markets (subnotes, desktop replacement laptops, desktops, workstations, servers, server blades, supercomputing clusters) makes for a very cost-effective way for Transmeta to extend its reach

Reply author: oion
Replied on: 11/23/2003 11:22:13
Message:

I've updated my TM8000 page with a (not exactly new) 3rd-party roadmap from PC Watch. It's an interesting chart comparing speed/power between Intel and Transmeta. The original article is here (Japanese). It's not really new information we couldn't get from separate sources, but it's nice to have a graphical depiction.

http://oion.net/qnd/tm8000/#roadmaps

Reply author: christh21
Replied on: 11/24/2003 00:59:14
Message:

That article about the Crusoe supercomputer was very interesting... I wonder if we'll ever get any performance enhanced CMS? I doubt it, but 50% increase? Transmeta still has a long way to go.

Reply author: oion
Replied on: 11/25/2003 18:04:31
Message:

"Transmeta needs money, please buy our stocks"
* http://investor.transmeta.com/news/20031125-123410.cfm
(to save bandwidth )

Oh yes, and the article with Green Destiny and the 50% increase.... 50%? Wow, that's impressive, and seems surreal. I don't remember the technical details for how CMS works, but it would make sense for Transmeta to have a cookie-cutter CMS software for OEM companies to work on. It doesn't necessarily mean Transmeta failed to make a good CMS, it could just mean that that group has excellent engineers for their own systems, and/or their systems aren't so complicated that reworking a CMS patch would introduce too many bugs.

Reply author: mgglim
Replied on: 11/25/2003 18:53:41
Message:

I think people invested in SCO should shift their funds to Transmeta. If they have equal chances of failure (or success), going with an innovator is infinetely better than going with an interloper. Plus, Transmeta will make Intel work hard for the benefit of us consumers, while SCO will plunder off the honest collaborative work of many well-meaning developers.

Reply author: jeffro
Replied on: 11/26/2003 07:27:27
Message:

People are going to invest where they think they will make money. I hope SCO takes it on the chin for what they're doing. But I wouldn't invest in TMTA right now. TMTA was well under $2 a few months ago and went up over $4 on a barrage of news, now seems to be headed the other way again. I agree with Oion that the their issuing new stock looks desperate. Guess it all depends on Efficeon, and the wait is getting pretty tiring. Q3 is long gone and Q4 passing fast.

Reply author: blehmann
Replied on: 11/26/2003 23:53:48
Message:

Firms issue stock when they think the company is overvalued, not undervalued. There is no way that TM is undervalued now. Moreover, TM does not need more capital now, it needs finished products. And Q4 is most definitely almost gone.

bnl

Reply author: mgglim
Replied on: 11/27/2003 03:26:33
Message:

Transmeta right now is very much a speculative bet and I admit I can't put blinders on my bias for its success. Why I support Transmeta is the same reason progress is made - competition. I wish its stock price would go up, not out of speculation, but because there is real, and not simply perceived, value in the company. And that value has to come from having a very good product. Having a continuing series of improved product cycles, arising from having improvements both in design (e.g. CMS, Long Run) and in production process capability (90 nm, 65 nm, high yield) will keep it on Intel's gunsights, and spur more innovation.

Transmeta is hurting right now. Not so much for lack of a good product, but because it is going to run of out cash. It is facing its darkest hour (I have visions of Britain being bombed by the Luftwaffe.). To stay on course with its vision, it needs to stay independent. If it runs out of cash, it could be sold for its IP (intellectual prorety) and its technology will be transferred to the acquiring company. And Transmeta's soul will continue on it a chip from either Via (which absorbed an earlier failed venture, Cyrix), AMD, IBM, or Intel. But the resulting chip will simply be an evolutionary improvement. But with Transmeta being independent, it has the ability and the determination to produce something truly groundbreaking.

I believe Transmeta will come out of this experience stronger, having experienced its baptism by fire. There are many companies that failed miserably even with great technology (3dfx for example), but many have succeeded because they had faith in their vision. Qualcomm was nothing only ten years ago, now it's primed to take a cut of every 3G mobile handset made. AMD trailed Intel every step but kept sharpening its prowess, now its Opteron is gaining rapid acceptance with IBM, HP, and even Sun Micro, and it can be argued it has Intel on the tail end.

When you think about it, Transmeta has very solid intellectual property. It has learned from its experience with Crusoe. It has the best minds assembled to tackle the complexities of using software to optimize processor performance against power consumption. It has proven it can bring its design concepts to product in the reliability of its Crusoe processors in both notebooks and in the case of Green Destiny, in cluster computing. And yet, it has barely demonstrated the range of devices its processors can run on.

What it doesn't have is billions of dollars stashed for a rainy day. And time. And the certainty of control of its processes, as is the case with fabless chip companies. All of that might seem like big odds for the faint-hearted, but this is actually the best condition for Transmeta to thrive and succeed. If Transmeta were a nation, it would produce another civilization.

Just keep your eyes and ears peeled.

Reply author: jeffro
Replied on: 11/27/2003 07:30:46
Message:

Don't know how one can say "There is no way that TM is undervalued now" If Efficeon is a success, TMTA stock will go up. If not, surely it will go down. Maybe Transmeta has well-placed confidence in Efficeon and the issuing of new stock is a sound action. But it wasn't perceived that way by everyone, based on the step dropoff in stock price just after the announcement.

Reply author: blehmann
Replied on: 11/27/2003 16:55:48
Message:

Jeffro, there really is no way that TM is undervalued now. One does not measure undervaluation relative to what the company will be worth if Efficeon succeeds or fails in the future. Undervaluation would arise of the market overestimated the prospect of failure or underestimated the possibility for success. These are statements about the probabilities for success or failure, not which outcome eventuates. You place your bets before you find out what is going to happen. Success will happen if Efficeon ships sufficiently quickly with sufficient advantages over the Pentium to get it nontrivial design wins and failure will happen if this does not happen. The market does not seem to be overly pessimistic about the probability of either success or failure.

And mgglim TM is failing precisely because it is not shipping a competitive product.



bnl

Reply author: mgglim
Replied on: 11/27/2003 18:50:40
Message:

Blehmann, no doubt about that. Maybe a little turkey dinner and they'll bounce back. I wonder how much of the original design team Ditzel assembled is still with the company. Having a competitive product not just now but for the long term depends on the edge coming from the strongest team in the business. Having confidence in Transmeta is as much a consideration for companies thinking of going Transmeta.

Reply author: jeffro
Replied on: 11/27/2003 23:28:41
Message:

What evidence is there that the market is overestimating the probability of failure or underestimating the probability of success? How does one estimate such a thing? We don't even know how well Efficeon performs yet.

I'm sorry to keep on this, though. The "no way" was too strong for me, but I don't think you're wrong to strongly believe TMTA is a good value. So that's all I'll say on this.

Reply author: mgglim
Replied on: 11/28/2003 02:06:07
Message:

I think we have to look at Transmeta at this point in speculative terms. This means that we can set aside discussing matters relating to valuation, so whether it is undervalued or overvalued is really moot. As an object of speculation, Transmeta as a stock can be framed in terms of probability. You ask what the probabilities of success and failure are. Let's say it's fifty-fifty. Now you ask, what is the cost of failure and what is the reward for success. If Transmeta fails, will its value go down to zero? Probably not, since it has valuable intellectual property that are of value to processor chip companies. If it succeeds, what is the value of the market it can carve for itself?

Let's assume a worst case of being wiped out if it fail. An investment of $1K would be reduced to nil, and I lose $1K. If however Transmeta should succeed, and I'm looking at a longer time frame, I could very well see Transmeta being a $50 stock in two years. That would mean a gain of say $14K ( 50/3.2 *$1K - $1K).

Given these odds, in my thinking Transmeta would be a good speculative bet on the power of its cost efficient processor architecture.

You can put in your own numbers as my numbers can very well be considered voodoo to you, but in essence this explains my thinking on the subject of valuation.

Reply author: thetenken
Replied on: 12/03/2003 18:44:35
Message:

Here's some more competition for Transmeta. If Intel wants to beat out Transmeta in the price race, they'll probably use the Celerons to do it...

http://news.com.com/2100-1044_3-5113002.html?tag=nefd_top

Reply author: oion
Replied on: 12/03/2003 23:33:21
Message:

*** As for battery life, we'd have to compare Transmeta vs. P-III to get an idea of competitive edge. Also, the mobile celerons would have to not only have similar battery life, but better performance than a TM8000, which I don't think would happen (not that we have test cases) - but that would be redundant with the P-M... It's feasible Intel may attempt to truncate the market from both ends against Transmeta. Sounds like something they'd do. *** And they certainly have the money and means. But another way to look at it is this - if the new Celeron is priced well below Efficeon and can perform relatively well - why would anyone want to buy an overpriced P-M/Centrino package? (messyedittoosleepy)

And sort of a side comment, the idea of Intel releasing a revamped chip reminds me of a Tom's Hardware article benchmarking the latest Athlon 64 chip against a P4:

quote:

---

Since Intel already has an inkling of what the outcome of the eternal duel between Athlon 64 and P4 will be, the manufacturer hastily introduced the "P4 Extreme" a few days ago at the IDF (Intel Developer Forum 2003) in San Jose. We were there: the processor is nothing more than an Intel Xeon with a P4 label tacked onto it, complete with a 2 MB L3 cache, now offered with FSB800 (200 MHz real FSB speed) and 3.2 GHz. << snip >> So was it a fair move for Intel to make such cosmetic changes prior to the actual launch of the Athlon 64? We see it as the infantile reaction of a monopolist who's naturally inclined to act like a general at a sand table exercise.

---

Reply author: christh21
Replied on: 12/04/2003 00:34:56
Message:

I find it somewhat ironic that for Intel to compete with Transmeta, they are releasing low power processors at very low speeds. If the TM8000 1.1GHz can perform as well as a P-M 900mhz, then it will surely trash an 800mhz Celeron-M (with its reduced cache/etc). This announcement actually makes me a little more optimitistic. But, I'm still nervous that there haven't been any design wins with Efficeon. Looks bad. Very bad.

Reply author: oion
Replied on: 12/04/2003 01:10:26
Message:

I didn't pay attention to actual speed/performance notes in the new mobile Celeron article (), but assuming you're right - "reduced" Celeron - it may compete more with a Crusoe than the Efficeon. But why bother competing with Crusoe... Why release a repackaged low-performance sort-of-low-power CPU at all? Perhaps the price would have to be very low to make people notice it, along with the branding weight. It makes more sense now that the new m-Celeron won't perform that great, otherwise it'd eat away the expensive P-M/Centrino market, which is the last thing Intel would want.

::Pokes Transmeta and OEMs with sharp stick::

Reply author: oion
Replied on: 12/05/2003 00:07:22
Message:

FINALLY.

http://investor.transmeta.com/news/20031204-123949.cfm

"The new HP Consolidated Client Infrastructure (CCI) consists of an access device -- an HP thin client, powered by a Transmeta's Crusoe® processor -- that connects with the HP Blade PC, powered by Transmeta's new Efficeon processor running Microsoft Windows® XP Professional, located in a centralized data center. Blades are dynamically allocated to only one user at a time, providing a truly individual computing experience."

At least it'll be used somewhere now.

Reply author: christh21
Replied on: 12/05/2003 15:30:08
Message:

quote:

---

Originally posted by oion

At least it'll be used somewhere now.

---

Reply author: oion
Replied on: 12/05/2003 16:06:10
Message:

I know, I know.

Mind you much of the HP press release is about the entire system of Consolidated Client Infrastructure, not the Efficeon-powered PC blades themselves (which will be sold as part of that system, apparently). And they won't even be available until March 2004. ~sigh~ Still, the thin clients area will be an interesting market. We'll see if corporations who use this system will like it so much as to provide impetus for more individual products.

http://www.computerworld.com/hardwaretopics/hardware/story/0,10801,87786,00.html
http://www.infoworld.com/article/03/12/04/HNbladePC_1.html

quote:

---

Thin-client pricing will start at $349, and the PC blade will list for $799. HP said a full product suite -- including customization, implementation, training and support -- will start at under $1,500 per seat and at less than $1,000 per seat for a company that wants to replace tens of thousands of its desktops with blades.

Although HP sees big potential for its blades, the reality is that thin-client shipments, most of which are server-based systems, amount to only 1% of PC shipments. Market research firm IDC in Framingham, Mass., is forecasting about 1.45 million thin-clients shipments this year but has said the market will grow at more than 20% annually, reaching 3.3 million shipments per year by 2007.
(-Computerworld)

---

Reply author: thetenken
Replied on: 12/05/2003 23:33:53
Message:

Related story, blade servers for Wyse.

http://www.eweek.com/article2/0,,1403887,00.asp

Now where are the subnotes!

Reply author: oion
Replied on: 12/05/2003 23:43:02
Message:

Ah yes, I missed the Wyse press release on Transmeta's main site because - it's not even linked on their actual press release page! (Why?) Oh well. Yes, still waiting on Sharp and Sony, and maybe even Fujitsu. uuuuugghh

http://www.wyse.com/about/pr/2003/1203_transmeta.htm (as linked from Transmeta's site)

Reply author: mgglim
Replied on: 12/06/2003 11:19:47
Message:

I don't know whether to feel good or distraught about news coming out lately. One one hand, TM is extending its reach towards other areas like blade PC's. On the other hand, it is finding it difficult to convince itself, much less us, that it could compete at this moment on the notebook space. I suspect that it may very well be that making CMS code is more tricky for notebooks than it is for devices in the corporate workspace. A lot of that has to do with having a narrow and defined set of applications in a corporate environment, which seeks conformity and predictability so that support costs can be minimized. Such a situation lends itself well to making CMS software that is easily optimized and thus will makes its chips competitive even on computing prowess. On the notebook front, TM may be finding it hard to be everything to everyone since "everyone" would be the different permutations of applications in the whole wide world.

So, maybe the next announcements would be "anything but notebooks" such as digital video recorders, web servers, web tablets, database servers, etc., all these having dedicated functions. That would still be a large market, although it would be preferable to get the big banana, which is the notebook space, since this is where the present opportunity is and where there is most growth.

Like most of you, I am keeping my fingers crossed that TM comes out with a good flavor of CMS for the notebook, with excellent notebooks to boot, pretty, pretty soon.

Reply author: oion
Replied on: 12/06/2003 12:02:44
Message:

The question about CMS is whether Transmeta tailors for every notebook or lets OEMs work on their own version from a standard architecture - I think it's the latter from previous experience. Therefore I think it's less the CMS issue and more the Intel front, really, that is keeping Transmeta away from notebooks. Right now Transmeta really needs to do well on at least one front - it needn't be limited to notebooks; when its processors are successful in one area and we can see specific real-life advantages in its use, then expanding product lines will just follow naturally. I don't mind that right now the only non-vapourware is a HP blade PC/thin client solution, even if it's a little disappointing. Looking at the big picture, I don't believe TMTA is "finding it difficult to convince itself" that the notebook sector is the way to go; HP picked up the chip relatively quickly and it just so happens that the product will be a blade PC. That doesn't preclude anything in itself. The rest is up to the OEMs. Perhaps TMTA will last long enough for LongRun2 to be released, and then the market picture will change yet again.

Reply author: thetenken
Replied on: 12/06/2003 12:33:21
Message:

That may be the case. HP and others may be waiting for the new manufacturing process that could take the Efficeon to 2Ghz, along with Longrun2 technology before they make the leap to notebooks. The blade market should generate enough revenue for Transmeta for the short term and provide them with a testbed to optimize their CMS and LR2 tech in a more controlled environment. Who knows, maybe the next supercomputer in the Top 10 could be Transmeta? =)

Reply author: jeffro
Replied on: 12/06/2003 12:37:28
Message:

Nice to see a product, but it would be much nicer to see something those of us not running a big server/client system could get our hands on. Maybe HP will make the fanless desktop I covet.

A little off topic: I'm having a little trouble seeing the advantage of HP's "PC blade" concept. How is having each blade dedicated to a single user better than having the total processing power of the server dynamically allocated according to the needs of the clients? Is the advantage just the simplicity (and therefore cost)?

Reply author: christh21
Replied on: 12/06/2003 14:30:58
Message:

About the supercomputer... there is already a supercomputer built with Crusoe processors that is about 1/10th or less the size of one that has the same computing capacity. Due to low heat dissipation and energy consumption, the Crusoe supercomputer is much less expensive for the same computing power. There was a post about that earlier in here.

Reply author: oion
Replied on: 12/06/2003 18:58:32
Message:

quote:

---

Originally posted by jeffro

A little off topic: I'm having a little trouble seeing the advantage of HP's "PC blade" concept. How is having each blade dedicated to a single user better than having the total processing power of the server dynamically allocated according to the needs of the clients? Is the advantage just the simplicity (and therefore cost)?

---

quote:

---

Blade PCs will enjoy a centralized and easy-to-manage solution that offers numerous benefits such as increased compliance with software licensing requirements, cost-savings by reducing the likelihood of virus penetration, reduced power consumption with the Efficeon's energy efficient design, much simpler "moves, adds and changes" and a reduced risk of data and equipment theft.

---

quote:

---

A blade PC (BC anyone?) is a full PC on a small card that you stick into a special slot in a rack or cage especially adapted to Blade PCs. It's just like your nVidia video card, for instance, except it has a CPU, GPU, harddisk, network card, and so on, all on board. It's a full PC in a compact form.

In an office for 50 people, the IT dept has a cage which houses the 50 Blade PCs (BCs) for all the users and all the users see is a connectivity station on/under/near their desks into which they plug their monitors, keyboards and mouses. USB would do fine here.
The advantages are users still have their own PC but can't touch it nor mess with its hardware. The system administrators get centralized management, easy upgrade and replacement and relocating someone PCs becomes a matter of patching data cables slightly differently.

---

Reply author: jeffro
Replied on: 12/06/2003 19:19:29
Message:

The ClearCube description sounds different from the HP concept (though maybe this is due to oversimplified description of ClearCube). In the HP setup, each blade is dedicated to a single user at a time, but once the first user logs off someone else can use it. It's just providing personal processing horsepower, not really a PC in the conventional sense (processor, memory, storage, like the clearcube description). What I don't get is this: why is dedicating blades to each concurrent user is better than pooling the processing power of the blades in a scalable "mainframe" and distributing it as needed. Seems that if 50 users are just surfing the web or doing some word processing, then you're more or less wasting the power of 50 blades. And each user is limited to the computing power of a single blade. Seems like you could get away with far fewer blades in a mainframe design. Though maybe the maintenance advantages described outweigh this.

I'm sure there's a compelling reason to do this if HP is doing it, I'm just trying to find out what it is.

Reply author: oion
Replied on: 12/06/2003 19:44:46
Message:

I suppose the real question is why blades are better than traditional served-mainframes in terms of performance (knowing little of that area myself). Perhaps the cost also factors in a great deal. (Yes, I see ClearCube's description is a bit different than HP's CCI; it would seem ClearCube's blades are less dynamic than HP's CCI because they're described as "dedicated" to a user, therefore actually not as good, but can still be dynamically switched to another blade on a single point of failure. I think the differences aren't as great between HP/CC as mainframe/blade, though.) I'm sure the idea in corporations is that people won't be wasting time just surfing the web and doing word processing, however. Let's say, database programming and stuff like that.

Wish HP would get a technical page of the BX1000 up soon.

Reply author: jeffro
Replied on: 12/06/2003 19:58:03
Message:

Sure, corporate workers don't surf the web. Just kidding; I probably used a bad example. But no matter what they're doing there will be idle time where they're just looking at the screen and thinking. So it still seems to me that dedicating blades to each worker is not efficient if the number of users is not pretty small. But all this is certainly not my area of expertise either, which I'm sure is apparent.

Reply author: mgglim
Replied on: 12/06/2003 21:51:13
Message:

The real point of having blade PC's is manageability in an office with many users. Instead of having 50 individual pc's as we're used to, each with different problems arising from hardware and software unique to each user's degree of customization, manipulation, and of course, abuse, you will have 50 similar thin clients each connected to a blade PC. There will then be 50 blade PC's in a blade rack. The adminstrator has all 50 blade PC's under his control, and in the same place. Each blade PC will have a processor, memory, storage, and it will do the bulk of the processing of its corresponding thin client. The thin client is basically an input/output device, with a screen, a video card, a keyboard, some nominal memory and a less powerful processor. The user is kept from abusing the blade pc, and is given little latitude to muck the works. Thus, you have a fully functional and powerful PC in the form of a thin client and a blade PC connected via high speed connections.

At this point, we can either treat this PC as either a standalone PC or a client in a client-server setup. As a client in a client-server setup, it will rely on data from a server. Standalone, it will not be connected to a server at all and will rely on applications and data residing on the blade PC.

Now, in the age of the Internet or Intranet where the web is your interface, you really do not need a client-server model anymore since the web interface allows you to obtain data from a web server. This makes the blade PC acting in a standalone manner more relevant. With high-speed internet, and with a powerful processor in the blade, and a reliable thin client, you have a set-up made for the times.

This leaves one question: why not use a singular powerful multi-processor engine instead of many blade PC's? Many reasons:

1. Multi-processor computers are not as scaleable as blade PC's.
2. They are very proprietary and thus would be more expensive to purchase and to maintain.
3. They have a single point of failure and this against the grain of networks, where there is redundancy that minimizes the effect of a catastrophic failure.
4. Blade PC's can be easily upgradeable.

etc. etc.

Reply author: jeffro
Replied on: 12/06/2003 22:49:41
Message:

Thanks, I think I'm getting the point. I did not think the HP blade PC could be configured as a standalone with storage on/assigned to the blade. Guess it makes sense to have the flexibility to do so.

Reply author: mgglim
Replied on: 12/07/2003 10:21:41
Message:

I wonder how well a blade PC-thin client set-up will work if you have a 50-user setup of gamers?

Reply author: oion
Replied on: 12/07/2003 11:49:40
Message:

More info about blade PCs and thin clients:
http://www.fcw.com/fcw/articles/2003/0908/tec-repack-09-08-03.asp
http://www.interex.org/hpwnews/content/hpwn12.04.03.html

Don't know about gaming; interesting theory, though. All the articles I've seen so far are related to corporate use, of course. But people with experience on networked multiplayer gaming/LAN parties may be able to comment given the information we have about CCI/blades right now.

One limited discussion:
http://www.geocrawler.com/mail/thread.php3?subject=%5BLtsp-discuss%5D+thin+clients+for+gaming%3F&list=10022

Apparently some Linux users have set up thin clients for gaming, but there isn't that much information. (And no info on blade pc gaming right now that I could find.) Curious~

Reply author: oion
Replied on: 12/08/2003 15:32:23
Message:

And now, for something (not) completely different...

SHARP Mebius MURAMASA PC-MM2

Reply author: mgglim
Replied on: 12/08/2003 16:44:51
Message:

That was an excellent write-up. I won't have to learn Japanese anymore, :-)

Reply author: oion
Replied on: 12/08/2003 19:16:27
Message:

(I don't know what write-up you're referring to but I've clarified the label for pages that were machine-translated.)

Sharp-Japan updated their official PC-MM2 page starting here:
http://www.sharp.co.jp/products/pcmm25ne/text/p1.html

A quick rundown (also in the specs for 5NE, but I'll post here anyway):

10.4" XGA LCD, TM8600-1.0GHz, 256MB DDR (fixed? not sure what the specs say), ATI Radeon mobility 16MB

About 2lbs, less than 1" thick at the thickest point. There's a physical "Mobile/Normal" switch you can use to toggle CPU throughput and screen brightness states to save battery life (or run at normal speeds on battery, I think - can be useful). Three different batteries; standard battery is rated to around 3-3.5 hours at total weight 2 lbs, highest capacity battery is rated to ~11 hours at 2.6 lbs. Same type of docking cradle as the previous MM1 model for synching. Floppy and optical drives are external and optional (bleh). One VGA-out, 2xUSB 2.0, 1xPCMCIA, audio-out. No mic-in or modem. Will retail around 180000Y or under $1,700 at current rates.

While I'm happy an actual Efficeon subnote will be available in retail very early in January, I'm not fond of the lack of ports and other limitations (currently only 20GB, I think it uses a 1.8" HDD; I've love an S-video out). 11 hours at less than 3 lbs is delicious, though. Assuming someone doesn't require much input/output and can get by with perhaps only a USB flash key, that's a real mobile-warrior item.

Reply author: jeffro
Replied on: 12/08/2003 20:17:05
Message:

I don't mind not completely different... in fact I'd be delighted if Fujitsu gave us an Efficeon based P2K!

Wish the Sharp info included battery capacity specs along with the battery life so we could do rough comparison with Crusoe and Pentium M notebooks.

Reply author: thetenken
Replied on: 12/08/2003 22:07:45
Message:

This is cool...the only thing that really turned me off from the original Muramasa was the lack of memory and the slowness of the Crusoe processor. If the Efficeon is everything they say it is, this is then a very nice computer. 2.6lbs and 11 hours of battery power, less than 1" thick, and syncs? That's awesome! Especially for someone like me who uses his notebook to supplement his desktop. I'd say this is one badass comp.

Reply author: oion
Replied on: 12/08/2003 22:19:46
Message:

I think the < 1" thickness is with the 3-hr standard battery; according to the photos, the other size batteries are a bit bigger/thicker, but I haven't seen exact battery specifications yet. Hopefully at the end of December or early January there will be more actual real-life reviews.

I'm less concerned about thickness than overall weight with the highest battery. Less than 3 lbs and, assuming the manufacturer spec is overstated, about 7-9 hours of battery life would be just great. Hopefully the PC-MM2 is an indication of what other TM8000 subnotes will be like.

Reply author: blehmann
Replied on: 12/09/2003 00:27:03
Message:

This gives me some hope, a design win at last (although we had indications before that Sharpwould be the first with an efficeon subnotebook. I echo the thought that I would love an efficeon p2k but having a p1, a p2, and a p5 would be slicing the market a bit thin. While it is not the box for me (I finally decided to quit waiting and bought a p5, the idea that there might be a cooler, lighter p2k-like machine out there in the not-too-distant future seems more possible now.

bnl

Reply author: mm42
Replied on: 12/09/2003 17:16:34
Message:

Great find, oion.

The scary thing is they say 1 GHz Efficeon is only 40% faster than 1 GHz Crusoe.

Reply author: oion
Replied on: 12/09/2003 17:32:18
Message:

I'm sure that depends on what benchark(s) they're talking about, as it's folly to depend on a single number for a single task. I didn't pay any heed to that particular blurb myself... Even though the other suite benchmarks are from Transmeta and therefore subject to suspicion, those are much more interesting when comparing to the Pentium-M.

Reply author: oion
Replied on: 12/09/2003 17:50:12
Message:

An actual english article regarding Sharp's Muramasa:
http://www.infoworld.com/article/03/12/09/HNefficeon_1.html

Reply author: thetenken
Replied on: 12/09/2003 19:53:29
Message:

Hmmm...

"Sharp does not have any plans to sell the machine overseas, said Nakayama."

Noooooooooo!

Reply author: oion
Replied on: 12/09/2003 20:48:29
Message:

I read somewhere else that Sharp said the same thing for the MM1 model (Actius) but that came over fine. There are other ways for importing, of course, so I'm more interested in OEM design wins overall.

Reply author: mm42
Replied on: 12/09/2003 21:14:57
Message:

Point taken about benchmarks, if they are randomly selected, but I tend to expect the manufacturer to want to show their new wares in a positive light.

In other words, hearing 40% improvement when I expect marketing hype is not encouraging.

Reply author: oion
Replied on: 12/09/2003 21:32:18
Message:

That's true; a mere 40% over Crusoe 1GHz seems disappointing. But another way to look at it is that a 1GHz Crusoe feels (in theory) more like a 700MHz Pentium III (er, from what others have suggested, I'm conjecturing). That means it's around 30% slower than what we think it should be. Does this mean a 1GHz Efficeon behaves the way we expect it to at the stated MHz? I only have experience with a 800MHz Crusoe, though. Still, a 100% improvement would be too much (1 GHz vs. 1 GHz) and would suggest the Crusoe engineering was so flawed that it had to be stripped and rebuilt for such an improvement. 100% improvement would mean 1GHz Efficeon benches twice as well as 1GHz Crusoe, a bit unrealistic and would speak poorly of Transmeta. So, we should expect something below 100% - 80-90 probably averages too high , so around 50% seems more optimal for an improvement. After all, that would place the expected performance back on the Pentium *hertz map everyone compares TMTA to. Not sure I'm making any sense here... LongRun2 may change that outlook, though.

As a side note, the HP Blade PC press release upped TMTA's stock price a bit, but it's been sliding. The Sharp Muramasa has no official press release from Transmeta right now and the stock has been unaffected (still sliding ). Oh well. I'm sure PR has its reasons. Maybe.

Reply author: christh21
Replied on: 12/10/2003 00:48:43
Message:

Remember that the 40% speed increase in the Efficeon may only represent pure computing power. Remember that the Efficeon has a DDR-400 interface, HyperTransport, ATA100, etc. The Crusoe had a 33mhz PCI, which really limited the speed of peripherals. The Efficeon will compute more, and it will help to speed up the entire system overall.

Reply author: mgglim
Replied on: 12/10/2003 02:29:22
Message:

Regarding speed, and from the standpoint of using an 800 mhz crusoe-based fujitsu p-2040 with 256mb pc133 ram, a 4mb video ram, and a slow drive (5000 rpm?) and a slow PCI bus, and an older CMS, I would very much welcome its replacement that would take care of the ff annoyances:

1. slow boot-up times (maybe hypertransport and faster drive will speed it up)
2. slow launching of applications (ditto)
3. slow draws of web pages (will having more video ram + faster disk I/O and faster RAM suffice?)
4. extremely sluggishness when two java applications are running (is it java or is it the processor?)
5. inability to maintain an interrupted flow of music playing in the background when using another application

I would just like the new laptop to be comparable to my Athlon XP 1700+ -based desktop PC. Just a side-by-side comparison, by "feel," and I know that if I don't have to wait as long as I do with the p2040, and come close to a decent desktop like mine, then I'll be glad to have an Efficeon.

I don't think I would even be contented with a mere 40% increase in performance, from where I'm coming from with my p2040. I want fast I/O's - web pages being refreshed rather quickly, applications launching quickly, plus faster boot-up times (also faster trip to and from standby mode) and better multitasking among applications (especially when java, video, and audio are involved. What good is 40% or even 200% speed increase in processing Photoshop, for example, if the user uses the laptop to surf, to email, to listen to music while at it, and use office applications (Word, Excel, PowerPoint) concurrently and can't work done fast enough because the notebook computer can't keep up because some bottlenecks in the system have not been relieved?

I hope that Sharp (and others) and Transmeta have taken these considerations in mind in the new notebooks. I don't think many people are easily swayed by mere benchmarks these days as we have been burned and consequenntly desensitized to those figures.

Reply author: oion
Replied on: 12/10/2003 11:32:23
Message:

So true about complete system performance expectations; we have to be more holistic about it. As for my P-2040 , my (1) is quite short (skip some POST and few startup apps), (2) I can definitely agree with, (3) Opera has been good to me, (4) I don't run Java, never did like it (and yes, a CPU limitation most likely), (5) never had a problem with that - if you're talking about mp3s, you may want to use a different application with better memory buffering. The optical drive playing a cd is too loud. I don't recall seeing official Transmeta benchmarks for the Crusoe, and I think they weren't even released. But 3rd-party benches are more believable, and I believe those did illustrate the slowness of Crusoe well beforehand (the earliest comments seem to be from 2000 with the first Sony models). So, if people did their homework on the chip and laptop models, they could have known Crusoe was not a speedy chip. These days it would be even sillier for a newcomer to buy a Crusoe machine and then complain about its slowness, given all the information already on the web. So I'm not entirely desensitized to benchmarks myself; just good to take them with a grain of salt.

I don't know how Tablet PCs have been; considering HP is moving on with Blade PCs, I wonder if they'll upgrade their Crusoe tablet line fairly soon.

Reply author: grandall
Replied on: 12/10/2003 12:51:03
Message:

quote:

---

Originally posted by oion
I don't know how Tablet PCs have been; considering HP is moving on with Blade PCs, I wonder if they'll upgrade their Crusoe tablet line fairly soon.

---

Reply author: oion
Replied on: 12/10/2003 13:15:24
Message:

Ah, that would explain why HP didn't have a product announcement for tablets, only Blade PCs. Adding another CPU version would definitely slice that market too thin. Oh well.

Reply author: mdavis
Replied on: 12/10/2003 14:37:44
Message:

I don't know if this is off topic; it is at last tangential to topic, but is it true that the 5000 blows the 2000 out of the water speed wise? I have followed the discussions, but only superficially, partially because I had the impression that the difference wasn't that great. If it is that great, I might want a 5000 as my second laptop. What's the story, if you know?

mdavis

Reply author: oion
Replied on: 12/10/2003 15:01:06
Message:

Er, yes... I don't know where you got that impression, but indeed the P-5k is far faster than the P-2k. A look at applications and other threads would give such an indication, as well as normal benchmarking. Obviously such a comparison hinges on what types of application one uses regularly. For word processing, I'm sure there's little difference. If you're not in a hurry you could wait to see if there are lighter TM8000-based machines on the way with better battery life.

Reply author: christh21
Replied on: 12/11/2003 13:33:15
Message:

But, a lot of P5000 people seem to complain about the heat that the machine generates. One of the great things about Transmeta based computers is the very low heat. Even Intel's mobile processors haven't been able to keep the heat under control.

Reply author: oion
Replied on: 12/11/2003 13:41:23
Message:

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/hotrev237.htm
(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) 

Reply author: jeffro
Replied on: 12/11/2003 14:30:03
Message:

Well, the performance seems pretty good. If it is 60% or so faster than Crusoe, that's in keeping with Transmeta's promises, right?

Battery lifetime sounds somewhat disappointing - about the same as P2120. P2K standard battery is 20.5 Watt*hours, just slightly larger than the Sharp's. At same clock speed, I thought Efficeon was supposed to run much longer than Crusoe and greatly outperform it at the same time. I know processor is only one component drawing power from the batteries so comparing processor efficiency in machines with different displays, HDDs, etc. is not apples to apples. But I get the impression there is not going to be a big boost in battery life in new Efficeon notebooks.

Reply author: christh21
Replied on: 12/11/2003 15:40:49
Message:

You know, this has got to be the longest thread on this forum. I was only a "Starting Member" when this thread first started =). The curious thing is, that after 150 posts, we still don't know much more than when we started it all. I'm starting to get disappointed with my Crusoe, and waiting for an Efficeon to upgrade (a matter of ideology I suppose). I'm getting anxious.

Reply author: mdavis
Replied on: 12/11/2003 22:35:31
Message:

quote:

---

Originally posted by oion

...If you're not in a hurry you could wait to see if there are lighter TM8000-based machines on the way with better battery life.

---

Reply author: oion
Replied on: 12/11/2003 23:01:36
Message:

I think there can never be a bottom line conclusion between the P-2k and P-5k because people using either may have different usage requirements. I really don't want a fan, for instance - and Transmeta notebooks are the way to go with that. The P-5k is heavier, slightly larger (straight specs are best found on Fujitsu's site), battery life is better only because the standard configuration is equivalent to the P-2k's hicap battery (misleading). Keyboard was changed. I really don't want to go *heavier*, however, which is one reason why I'm waiting. That's a step backwards for me.

christh21: Most of this thread relates to business tactics and philosophies, so we didn't get much information at all until just this past week. I'm not sure I should continue to update this thread with news updates related to Efficeon or post to the subnotes forum on specific models. Perhaps I'll just update my site only.

Reply author: oion
Replied on: 12/11/2003 23:17:41
Message:

quote:

---

Originally posted by jeffro

Well, the performance seems pretty good. If it is 60% or so faster than Crusoe, that's in keeping with Transmeta's promises, right?

Battery lifetime sounds somewhat disappointing - about the same as P2120. P2K standard battery is 20.5 Watt*hours, just slightly larger than the Sharp's. At same clock speed, I thought Efficeon was supposed to run much longer than Crusoe and greatly outperform it at the same time. I know processor is only one component drawing power from the batteries so comparing processor efficiency in machines with different displays, HDDs, etc. is not apples to apples. But I get the impression there is not going to be a big boost in battery life in new Efficeon notebooks.

---

Reply author: jeffro
Replied on: 12/12/2003 07:30:33
Message:

My 60% came from a very rough average of the Ubiq startup times. Of course, the speed increase will depend on what you're doing; I think Transmeta claims 50-80%. I hope it's better than 40%. That would really be just a little more than barely noticeable in most cases.

Reply author: thetenken
Replied on: 12/12/2003 13:03:33
Message:

Maybe those claimed numbers are only after they move to 2Ghz and implement LR2 technology? It would seem that LR2 tech would significantly help battery performance in stopping power leaks, and the lower micron process would also cut down on heat and power consumption?

Reply author: christh21
Replied on: 12/12/2003 13:40:16
Message:

LR2 is also the technology that they are looking at licensing to other chip manufactures. I think it's a good business move that will bring in significant revenue to allow them to fund further R&D and chip production, and it will give them a bigger name in the industry. The web cast shows a chip with LR2 running at just 2mw at idle state... that's amazing, probably less than your RAM! The only problem with the battery life cliams, is that the CPU isn't the only power consuming device. Hopefully sometime in the future, motherboards will have some type of LongRun technology that will lower consumption for other devices as well (I know there is ACPI). They need to have a whole package of low-power RAM, low-power HDDs, etc.

I am amazed that we are still using mechanic parts in our computers... hard drives and optical drives? Those consume the most power. Hopefully technology will increase to the point where everything will be convert to non-mechanical, like ultra fast FLASH memory instead of a hard drive, and memory sticks instead of CD drives. That's probably a long way off though.

Reply author: thetenken
Replied on: 12/13/2003 11:38:44
Message:

Yeah, you could make a flash based computer now, but it's kind of cost prohibitive. Having a 60GB hardrive at around 200 dollars, or getting a 4GB flash card at $1,499?

http://www.dpreview.com/news/0308/03080402lexar4gbship.asp

Reply author: oion
Replied on: 12/13/2003 11:57:43
Message:

Maybe only if you're working out on the field; but a computer with no moving parts would certainly be very, very useful in such a setting. Probably much less sensitive to atmospheric pressure, humidity, and other environmental limitations. An Efficeon-based fanless, flash-HDD, memory-slot laptop? How about, considering how small a CompactFlash card is, a system with 4x4GB set up in RAID fashion? Pretty neat idea. But then, the most delicate part would still be the LCD. And the LCD itself tends to be one of the most energy-hogging components in a laptop anyway, unless technology evolves.

Regardless, TMTA's stance on energy efficiency is still the only way to go.

Reply author: mgglim
Replied on: 12/13/2003 18:12:45
Message:

I'll be more surprised if those things you guys are dreaming about don't come out the sooner. Just look at how LCD's and batteries have quickly evolved and production ramped up these past few years. Just look at how cheap LCD lightbubs are turning out to be (even if they're not computer stuff). And some products from nanotech are within reach in the form of drug delivery systems and spill-proof pants. It will be fun to watch display, memory, storage, and power technologies take dramatic twists and turns towards high efficiency and affordability. They have to. Right now the tech world is in a funk, and products aren't ready selling like hotcakes. Companies may be retrenching, but one things for sure they're not holding back on is R&D. They have no choice but to keep trying to have an edge, an enviable intellectual property (IP) portfolio, so that their technologies can be licensed and used on better products.

And out of this will be a slew of killer products and applications. Powerful but extremely mobile is an oxymoron nowadays, but not for that much longer. If Transmeta could slog it out thru its current travails, it will be among those companies with a great product.

For now, I'm wishing for a small thin screen for Christmas. I'll slap it in front of my P's screen and use that transflective screen for surfing out in the backyard, by the pool or by the beach.

Reply author: rgoldman
Replied on: 12/13/2003 19:14:12
Message:

If only Fujitsu would announce an upgrade to their P1000 line with this new chip; has anyone heard anything?

Reply author: oion
Replied on: 12/13/2003 22:07:33
Message:

I haven't heard anything, though the facts that Fujitsu had contracted with TMTA before and were present at the release imply something may be afoot. And don't bother trying to ask Fujitsu directly since all company policies that I'm aware of do not disclose this type of information besides public press releases. Mind you Sony still hasn't peeped a word about the supposed PCG-UX5 so it is still rumour (going rumour is release Dec 15, but if that doesn't happen, I hope it wouldn't mean a production delay from Transmeta ).

As for faster evolving technology - Yes, quite possible more solid-state will come out sooner than we think. It's just a matter of being cost-prohibitive at this point. And I think Hitachi is releasing PDA/handheld-based fuel cells for 2004 or 2005...

Reply author: thetenken
Replied on: 12/14/2003 00:50:09
Message:

As for screens, they've been testing OLED's for some time now with cellphone screens and I think a few PDA's. An OLED screen would cut power use by quite a bit, and there are new flexible OLED's.

http://www.wave-report.com/tutorials/oled.htm
http://www.universaldisplay.com/foled.php

Dream computer? Includes an OLED flexible that can be folded, removed and replaced. Flash RAID, Efficeon processor @ 2Ghz with LR2, and Gigabit ethernet piped through my fiber optic line that costs 40 bucks a month. Ah, dreams...

Reply author: oion
Replied on: 12/14/2003 09:31:58
Message:

I remember talk of OLED screens some time ago, but I don't know why they aren't more popular now these days; either the technology still isn't perfected for large displays or .. what?

Hehe, guess I'll keep my dreams focused on December 2003-January 2004 (more OEMs, c'mon).

Reply author: gellul
Replied on: 12/14/2003 15:32:13
Message:

The energy consumption was supposed to be a lot less with OLED displays. But it turns out the energy consumption is similar to a conventional backlit TFT LCD display. Bottom line is battery won't last any longer with an OLED display.

Since OLED is organic the usable life is not as long as TFT LCD. For now were better off with conventional technology.

With fuel cells just a few years off, no one is going to need a really energy efficient CPU anyways. Efficeon and Centrino products will be on the "Product Roadmap" for just a few more years because fuell cells will give us possibly 10 hours of battery time on a more conventional cpu platform.

quote:

---

Originally posted by oion

I remember talk of OLED screens some time ago, but I don't know why they aren't more popular now these days; either the technology still isn't perfected for large displays or .. what?

Hehe, guess I'll keep my dreams focused on December 2003-January 2004 (more OEMs, c'mon).

---

Reply author: oion
Replied on: 12/14/2003 23:20:35
Message:

I've updated my TM8000 page with a general summary of an article from PC Watch (Japanese) discussing Transmeta and Via's competition with Intel in the small-pc race. The original article is here. My page is getting a little long... I'll likely condense all the benchmark figures later.

Reply author: jeffro
Replied on: 12/15/2003 09:09:06
Message:

I don't think fuel cells are much of a threat to the idea of efficient processors. Even if fuel cells can become cost effective enough to be an attractive technology for small devices like notebooks, recharging batteries is really convenient, just plug in anywhere without worrying about carrying around methanol or whatever. Then there's the heat issue. Finally, even if you want to use a fuel cell, wouldn't you rather be able to go 30 hours than 10?

Reply author: mgglim
Replied on: 12/15/2003 11:15:48
Message:

I'm with you jeffro. Although I'm not sure that's the way a lot of gas-guzzling SUV honchos think, which by the way, is about half of the US population. If today there is a electric hybrid that would make a Hummer H2 go 30 mpg, tomorrow they'll have an H3 the size of a tank in the streets back to guzzling 12 mpg :-)

Reply author: thetenken
Replied on: 12/16/2003 11:32:29
Message:

quote:

---

Originally posted by gellul

The energy consumption was supposed to be a lot less with OLED displays. But it turns out the energy consumption is similar to a conventional backlit TFT LCD display. Bottom line is battery won't last any longer with an OLED display.

Since OLED is organic the usable life is not as long as TFT LCD. For now were better off with conventional technology.

With fuel cells just a few years off, no one is going to need a really energy efficient CPU anyways. Efficeon and Centrino products will be on the "Product Roadmap" for just a few more years because fuell cells will give us possibly 10 hours of battery time on a more conventional cpu platform.

---

Reply author: mm42
Replied on: 12/16/2003 20:56:42
Message:

I keep reading that these media center PC's are poised to take off. It seems like this would be a perfect application for Efficeon. I mean, who wants to try to watch TV with a dang fan running all the time?

Reply author: mgglim
Replied on: 12/17/2003 03:28:44
Message:

quote:

---

Originally posted by mm42

I keep reading that these media center PC's are poised to take off. It seems like this would be a perfect application for Efficeon. I mean, who wants to try to watch TV with a dang fan running all the time?

---

Reply author: oion
Replied on: 12/19/2003 01:45:07
Message:

Benchmarks and efficiency between Sharp MM2 (Efficeon) and ULV Pentium-M laptops.

http://pc.watch.impress.co.jp/docs/2003/1217/ubiq37.htm

I've posted a general summary on my site as well which I'll likely edit later since I wrote it while half-asleep...

Reply author: jeffro
Replied on: 12/19/2003 10:29:45
Message:

With Transmeta giving statements like this regarding (lack of) confidence in manufacturing process, are OEMs going to want to design with Efficeon?

http://www.theregister.co.uk/content/3/34602.html

Reply author: oion
Replied on: 12/19/2003 11:20:23
Message:

quote:

---

Meanwhile, the TM8600, the 130nm version of Efficeon, is now going into production, with units shipping during the current quarter. Alas, "the substantial majority of shipments of this product to date have been of engineering samples and pre-production units", Transmeta admits.

---

Reply author: mika
Replied on: 12/24/2003 05:11:41
Message:

Following Oion's thread, it's interesting to see that on the processing speed normalized to power consumption rating (http://pc.watch.impress.co.jp/docs/2003/1217/graph04.htm) , Efficeon TM8600 1Ghz benched ahead of TR1's Pentium M and roughly on par with the P5000. For Transmeta supporters it's sort of good to know at least Efficeon was quite an improvment to Crusoe.

Reply author: oion
Replied on: 12/25/2003 12:48:34
Message:

To clarify, Sharp's MM2 using a TM8600 1GHz benched ahead of the Sony TR1 system which uses a P-M, but about on pair with the Loox T/P-5k system; the TR1 and Loox T are using the same P-M ULV 900MHz, after all. The MobileMark2002 benchmark does not compare merely the CPUs (even its battery life benchmark is suspicious without knowing battery capacities - note on my site). If they use a crappy HDD and slow RAM but the same high-speed chip, the bottleneck simply moves from one component to the other. OS setup can even affect benchmarks. The best CPU-based comparisons will always be identical system components and OS but different chip (and chipset); these setups are just too rare, however.

Regardless, the TM8000 is mostly competitive against Intel at the moment, but it has to quickly evolve to remain so. Being on par yesterday doesn't cut it today. Meanwhile, AMD isn't even in the subnote picture. For Crusoe users like myself (not speaking in a purely fan-based fashion of course), either P-M/TM8000 chip would be a big improvement anyway. Early in the coming year I'll be more aggressive in searching out a replacement for my P-2040...

Reply author: oion
Replied on: 01/05/2004 11:24:48
Message:

Off-topic from the Efficeon, Transmeta announced in a press release the new TM5700/5900 Crusoe lines for even smaller form factor and embedded designs.

http://investor.transmeta.com/news/20040105-125578.cfm

I'd be interested to see benchmarks; I don't recall the exact TM5800 specifications for a more technical comparison at the moment. Doubtful these would be used in subnotes like the P-series, though.

Reply author: thetenken
Replied on: 01/06/2004 12:04:37
Message:

Maybe we'll finally see a Crusoe PDA?

Reply author: christh21
Replied on: 01/06/2004 23:21:28
Message:

That's an interesting idea... I always thought about that, or at least some type of Transmeta ARM style processor. Imagine, running Windows XP embedded, with 256mb flash memory, being able to run all your favorite WINDOWS apps? (Of course, I wouldn't mind seeing it run Linux embedded, I'm not a big MS fan) I still think that Transmeta chips have opened a whole new world of possibilities, I just hope that companies begin to take advantage of them.

Reply author: gellul
Replied on: 01/07/2004 17:44:47
Message:

I'm not sure the current version of the Efficeon processor manufactured with 130nm process is really going to be competitive with the Pentium M? From what I have been reading the Efficeon is at best 30 to 40 percent faster than the 933 Crusoe. That's not good enough!!!!

I think the Fujitsu manufactured Efficeon using 90nm process is going to be competitive with Pentium M. The Fujitsu manufactured Efficeon will utilize the new Long Run Power Management II software. This will reduce power consumption so higher frequencies can be used. The 90nm process will enable the Efficeon to see 2GHz frequencies. This should be 100+% faster than Crusoe.

If Fujitsu were to use a faster 5400RPM hard drive and a 1.3GHz Efficeon processor I would think this would give the P5000 series a good run for the money. P2000 series owners would see a very dramatic improvement in performance and P2000 series product may see decent sales.

Fujitsu is already assuming a lot of risk manufacturing the Efficeon processor. But if Fujitsu could afford some additional risk, they may want to consider using OLED LCD technology. If OLED is anyhing like I have been reading, the OLED LCD should require less power consumption and possibly make up for the 5400 RPM hard drive and 1.3GHz Efficeon processor. Thus our new Efficeon Processor based P would retain the 4-5 hour battery life of the P2120 series.

I understand Sony will be coming out with new Palm OS based handhelds using the OLED display.

Reply author: thetenken
Replied on: 01/11/2004 19:15:33
Message:

Speaking of embedded...

http://www.oqo.com/

Reply author: christh21
Replied on: 01/12/2004 14:06:51
Message:

Wow... they've actually announced a finish product? Transmeta has been talking about OQO for years now, I think the idea started about 2 years or more ago, but they never got a real product out. Still a year or so away though... I like it. Too bad it'll probably be as expensive as the Antelope that's about $3500. I like the specs, 1ghz, 256ddr ram, sounds like it'll be more snappy than my P2110 at 867mhz. Like I have said many times before, Transmeta is opening up whole new possibilities.

Reply author: oion
Replied on: 01/12/2004 18:58:22
Message:

Well, it's still a bit early for the finished OQO product. :P As for another Efficeon product demonstration at CES:

quote:

---

Secure Communication Systems and chipmaker Transmeta were promoting the new concept of "submersible computing" with the Air Warrior, a petite Windows 2000 touch-screen PC elaborately sealed to be impervious to water and other annoyances. (The model on display at CES was bobbing in an aquarium tank.)

Michael Boice, vice president of sales and marketing for Secure, said initial Air Warrior models are being used by military pilots, who strap a model onto their thigh for navigation, flight planning and other tasks. While pilots are unlikely to go for a swim with an Air Warrior, the unit needs to be able to withstand falls, horrible weather and other military hazards.

Boice credited Transmeta for coming up with a low-power, low-heat processor that can run without any ventilation. "It runs even cooler when it's underwater," he added. Secure is working on making scaled-down versions of the Air Warrior for private pilots, field service technicians and other folks.

---

Reply author: thetenken
Replied on: 01/14/2004 00:20:37
Message:

Some shots of the new Muramasa from the CES Show, courtesy of Anandtech.

http://www.anandtech.com/showdoc.html?i=1945&p=4

Reply author: Alt
Replied on: 01/15/2004 18:45:42
Message:

quote:

---

Originally posted by thetenken

Speaking of embedded...

---

Reply author: jeffro
Replied on: 01/16/2004 08:10:53
Message:

The OQO and Antelope ideas are cool, but I agree that the prices of these things are way too high. I think they need to be $1K or less to catch on. If MPEG4 players with 40G hard drives cost ~$500, I don't see why a Transmeta-based PDA can't be made for <$1K.

I thought Transmeta was talking about some more OEM announcements. If the ones at CES are what they were talking about, that's pretty sad.

Reply author: blehmann
Replied on: 01/16/2004 23:04:00
Message:

Every day that passes without efficeon design wins brings the next version of the pentium m that much closer. The efficeon was going to save TM but I do not see a whole lot of evidence of salvation out there. Hoping that the 90 NM chip or LRM II software is going to save TM seems a bit like grasping at straws, the former because TM has had manufacturing problems in the past and the latter because it is vaporware at this point. I am a lot more confident that Intel is going to stay on its road map than that TM is going to solve problems that have plagued them for years in the next two months. The October announcements were a resounding thud to me. I wish it weren't so.

bnl

Reply author: thetenken
Replied on: 01/17/2004 01:19:30
Message:

quote:

---

Originally posted by Alt

quote:

---

Originally posted by thetenken

Speaking of embedded...

---

Gee thanks Tenken for supporting the topic I attempted to start a discussion on.

http://www.leog.net/fujp_forum/topic.asp?TOPIC_ID=4404
If you nab a link or are even inspired by someone elses topic at least point to the thread if you are not going to give them credit.

|Tab

---

Reply author: thetenken
Replied on: 01/21/2004 17:19:54
Message:

Interesting quote from Transmeta's CEO:

""We are excited that our first Efficeon-powered notebook, Sharp's MM2, began selling in Japan recently. In addition, we have many Efficeon designs currently in development, most of which are planned for shipment beginning in the second quarter of 2004. There is also significant interest in our next-generation 90-nanometer Efficeon processor. Our goal is to produce the 90 nanometer generation of Efficeon processors in the second half of 2004," said Dr. Perry."

This is from Transmeta's press release regarding 4Q/03.

http://investor.transmeta.com/news/20040115-126722.cfm

So there should at least be quite a few Efficeon-based notebooks hitting before June?

Reply author: christh21
Replied on: 01/21/2004 21:17:49
Message:

"Best-selling"? Is that in its class or overall? I wonder if a lot of companies are holding out for the 90nm version and that's why we haven't seen many 130nm models out yet.

Re