Toshiba, Sony close to 65nm sample production

[jvd]

Posted: 18 Dec 2003 09:17 Post subject:

--------------------------------------------------------------------------------

JVD, in fact my post was not referred to MONEY alone, in fact that was one of the weakest elements in the potential success of Cell.
I stated that the combination of:
- years of development
- IBM-Tosh-Sony alliance, and subsequently their own expertise put into one single project
- money

All push this project into the realm of "likelihood of success". Then OF COURSE everything can fail. Nvidia has gone way down when they were at the peak of their success (although that has individual variables into it which are not present in this case, see: NV2A sucked their resources).
Anything can fail, anything can happen, here we're discussing about an average case scenario where meteorites don't fall on Sony's fabs, earthquakes don't kill every MS employee and such

Yes and to give each of your posts that only an act of god can make it fail as you have stated in your posts is just being dumb in my opinion. I do not Look at the cell chip as being able to succeed or fail because of money or time because the other two companys have invested time and money. I think about what moves others can do and can't do to affect each of thier postions in the next gen system race. I do not sit around and keep quoteing that ms has spend countless time and money to make the xbox a sucess or the xbox 2 a sucess . No i don't think about that nor mention it anytime i think that ms can succeed.
That is my problem with all these posts that well sony spent alot of money on this so of course its going to succeed. I do not agree with that line of thinking and it is my right to disagree with it.

 

[Lazy8s]

If a console doesn't try to fully implement and utilize deferred rendering, that's how it can fail.

 

[notAFanB]

I do not Look at the cell chip as being able to succeed or fail because of money or time because the other two companys have invested time and money. I think about what moves others can do and can't do to affect each of thier postions in the next gen system race.

and somehow investmnt by competing firms does not fator into this?

aren't these 'moves' too?

That is my problem with all these posts that well sony spent alot of money on this so of course its going to succeed. I do not agree with that line of thinking and it is my right to disagree with it.

that is not the line of thinking that is primarily being perpetuated here jvd. If anything the focus has been on what may be required for Sony to lose the advantage and momentum they have earned this gen, have to carry this over and what it means for their competitors.

 

[Fafalada]

or Xbox2, I expect some MCM with 4 dual core PPC chips, with massive L2 and L3 cache like greater than 128 MB, dual R500 variant chip, huge memory like 2-4 GB, with bandwidth > 100 GB/s, huge HD at least 200 GB.

You're going to be dissapointed.

 

[notAFanB]

or Xbox2, I expect some MCM with 4 dual core PPC chips, with massive L2 and L3 cache like greater than 128 MB, dual R500 variant chip, huge memory like 2-4 GB, with bandwidth > 100 GB/s, huge HD at least 200 GB.

You're going to be dissapointed.

given info is limited in this point, faf don't you think MS might dump investment on this?

 

[aaaaa00]

or Xbox2, I expect some MCM with 4 dual core PPC chips, with massive L2 and L3 cache like greater than 128 MB, dual R500 variant chip, huge memory like 2-4 GB, with bandwidth > 100 GB/s, huge HD at least 200 GB.

You're going to be dissapointed.

given info is limited in this point, faf don't you think MS might dump investment on this?

I think those specs are dreamland.

Not to imply that xbox2 isn't going to be every bit as good as PS3, and then some.

 

[Dave Baumann]

Question: How does anyone know the level of investment that MS themselves are putting in and how much they a putting in to the companies that are building various other elements of the XBox2?

 

[London Geezer]

Question: How does anyone know the level of investment that MS themselves are putting in and how much they a putting in to the companies that are building various other elements of the XBox2?

We don't. After all we're just speculating about information we have at the moment...

JVD, i think the misunderstanding in this is around the word "succeed". I was talking about Sony and Cell delivering an impressive product (technologically) bacause of the investment, time and resources put into it. Of course PS3 might be the best hardware, still it could "fail", as in "not retain the number 1 spot.

I was never saying that just because of reasons 1, 2 and 3, Sony will "win" the next generation (whatever that means). I was merely commenting about the fact that, given reasons 1, 2 and 3, Cell will be an impressive piece of technology.
Then we have to think that the term "impressive" in itself is not really objective, in which case i wish to clarify that i consider the EE+GS "impressive". I consider "impressive" pretty much everything in current gen consoles, given the tiem they were released, with the only exception of the P3 in the Xbox, which i consider a "lazy" move from MS. The rest in my opinion is top notch, from NV2A to the sound chip, to GCs flipper+gekko...

 

[Vince]

Question: How does anyone know the level of investment that MS themselves are putting in and how much they a putting in to the companies that are building various other elements of the XBox2?

SEC is our friend? Also, I think it can objectivly be stated that the level of cooperation and investment between STI in particular is way, way over that which has been seen this far by Microsoft and it's partners.

Going to back to 1H2001, we've seen SCE, Toshiba and IBM invest $400M in building an independant research and design facility on IBM's Austin campus, hire or shift over 400 engineers to the architecture project aswell as a concurrent investment in process technology (SOI, Low-K, 90 -> 65nm roadmap). Later, around 2002 we've seen further investment in a Cell OS - which is being designed AFAIK as a collaberation between Austin, TJ Watson and the ST partners. In 2003 we've seen upwards of a combined $10Billion invested in ST's fabrication infastructure aswell as investment and a licensing deal with Rambus.

In the same time, we've see Microsoft talk to IBM in 2003, ATI in 2003, those Asian people whose company I forget. Perhaps it's just me, but the level of... devotion (?) put in by Microsoft (and Nintendo) pales in comparason to Sony.

And even if you'd point to Microsoft being more "hands off" and relying on the free-market to capitalize on a distributed investment scheme where each company's stand-alone investment surpasses STI's at a fraction of the direct cost - that's wrong. Hell, that ideology applied itself much, much better on XBox1 when each company basically ran with it independantly. From all indications, yours included, it would appear that Microsoft is taking a much more integral and direct role in defining a set-piece architecture with the partners. If this is so, then how can you question if STI's level of investment is even close when they began looking at their respective set-piece architecture back in 2000?

If you look at who and what has been devoted to STI Cell by companies such as IBM alone, it's clear that the level of devotion to Cell from the three companies is very high. When I see the same people from IBM who defined Cell (basically a who's who list of IBM research) work on Microsoft's project - then I'll perhaps consider they are in the same ballpark.

 

[Dave Baumann]

In the same time, we've see Microsoft talk to IBM in 2003, ATI in 2003, those Asian people whose company I forget. Perhaps it's just me, but the level of... devotion (?) put in by Microsoft (and Nintendo) pales in comparason to Sony.

And you know how long MS have been working on this in the background? You know how long they have been working with their partners prior to the annoucements? (we know that ATI were hiring specifically for the project well before the deal was inked).

Do you also see no distiction between Cell and PS3?

 

[Vince]

And you know how long MS have been working on this in the background? You know how long they have been working with their partners prior to the annoucement?

Something about... the law? Can Microsoft invest $400M in designing a new CPU architecture and break-ground on a cooperative design centre without leting it's investors know? Hmm... I smell borderline illegal/immoral practices Dave.

Besides, didn't the tenders for Microsoft's requirements go out to the companies for voluntary reply in.... 2002? Heh

We know that ATI were hiring specifically for the project well before the deal was inked.

Good call. Didn't you speculate that ATI did it's R400 ->R500 shuffle and other such smart reorganizings that would front-end load the development requirements in late 2002 with an eye for XBox Next?

Do you also see no distiction between Cell and PS3?

Of many here, I think I'm among the ones who've gone out of their way to make the distinction between Cellular Computing, STI Cell and the Sony/Toshiba Broadband Engine. Which, I think I've done a decent job of since finding that Rambus agreements which states this.

Yet, I see that the BE is merely a manifestation of Cell. The microarchitecture design is over, the CellOS work is underway. The BE's timeframe would appear to be bound to the 65nm process, not it's design.

 

[cthellis42]

And you know how long MS have been working on this in the background? You know how long they have been working with their partners prior to the annoucements? (we know that ATI were hiring specifically for the project well before the deal was inked).

Well we certainly know that hundreds of engineers and hundreds of millions of dollars can't move around in this industry without attracting SOME attention. And Microsoft is both watched like a hawk and seems to have plenty of willing leakers. Now if Steve Jobs were in charge of it...

Obviously ATi (and no doubt nVidia as well) had plenty of people at work on preparing propositions (holy alliteration, Batman!) for Xbox 2, considering there are nice meaty dollar signs attached (along with denying one's rival the cash and prestige), but just how much? <shrugs> Meanwhile Microsoft and ATi worked out an IP deal and only recently signed on other companies for the other systems, so whole-system integration couldn't be taken into account from the beginning, and MS itself has to decide how to use the IP and who to fab it. The GPU by all accounts thus far will be similar in nature as NV2A was to nVidia's other PC projects, and if MS is to be at the helm of system integration with all their partners, they couldn't be giving input in that way to ATi while ATi nor anyone else actually WERE brought on as partners.

They could indeed have been working in the background--and I rather imagine Microsoft is "working in the background" with a whole host of companies for a giant stack of future possibilities all over the spectrum, but the same scale was certainly not there--and most likely not the focus or direction.

 

[Vince]

Well we certainly know that hundreds of engineers and hundreds of millions of dollars can't move around in this industry without attracting SOME attention. And Microsoft is both watched like a hawk and seems to have plenty of willing leakers. Now if Steve Jobs were in charge of it...

HA! And if interns are involved, then forget it.. everyone wants something like that on their resumé .

 

[V3]

V3, 1 APU should be definately smaller than a normal PowerPC 970 ( the G5 is a full blown CPU designed for Desktop PCs ), so we cannot really work 32 for Sony, 16 for MS as your 16 are full sized PowerPC cores.

Well I expect 8 PPC cores, 16 maybe too much (but they can suprise me ). PPC by the time it hits 65nm for Xbox2 should be small. Xbox2 can have a cellular architecture of its own. MCM, 4 PPC chips, 128 MB of cache and low power, is what MS should aimed for.

You're going to be dissapointed.

Well, I can have hope, can't I ?

I think those specs are dreamland.

Not to imply that xbox2 isn't going to be every bit as good as PS3, and then some.

Well for 2005 and coming from Microsoft with their record on Xbox (ie they will try to top PS3 anyway they can, and they have high chance of succeed too), I thought that's pretty realistic expectation.

Maybe the HD is not that realistic given rumours that Xbox2 is not going to use them.

 

[Dave Baumann]

And you know how long MS have been working on this in the background? You know how long they have been working with their partners prior to the annoucement?

Something about... the law? Can Microsoft invest $400M in designing a new CPU architecture and break-ground on a cooperative design centre without leting it's investors know? Hmm... I smell borderline illegal/immoral practices Dave.

XBox comes under a specific group – just so long as their investors know what that group is working on they needn’t detail the minutia of what they are working on just so long as it serve the purpose of that group.

That group doesn’t need to be explicitly doing things themselves, they can be investing in other companies and working with those companies to forward their goals. This is what w know they are doing with ATI and IBM now, but we don’t know how long they have been doing it beforehand and with who else they are working with.

We know that ATI were hiring specifically for the project well before the deal was inked.

Good call. Didn't you speculate that ATI did it's R400 ->R500 shuffle and other such smart reorganizings that would front-end load the development requirements in late 2002 with an eye for XBox Next?

I suspect the reshuffling of the roadmap was down to a number of factors in terms of key inflection points (DX9 PS/VS 3.0, DX10, XBox2) but I would certainly guess that one of the key factors was to make sure that the roadmap is set to safely (as possible) deliver what they need to deliver to MS in the timescales MS need it.

However, the point being is that ATI were hiring in March for the XBox 2 project, some 6 months before announcement, and that work was already well underway at that point.

Do you also see no distiction between Cell and PS3?

Of many here, I think I'm among the ones who've gone out of their way to make the distinction between Cellular Computing, STI Cell and the Sony/Toshiba Broadband Engine. Which, I think I've done a decent job of since finding that Rambus agreements which states this.

And yet you cite the levels of investment that Toshiba and IBM have put in as a reason for PS3’s success. Sorry, but it seems to me that doesn’t necessarily follow to PS3. PS3 is but one application of Cell. Are IBM / Toshiba interested in the wider applications of Cell or are they interested in PS3 specifically? On what scale of importance to those companies is the PS3 specifically in relation to Cell?

 

[Entropy]

It seems to me that "CELL" is being used very loosely in the discussions in these forums.

"Cell" is NOT a chip. It can't even be called an architecture properly, it is more of a concept, a concept originally outlined (and the term coined) in an IBM white paper.

It can be safely said that IBM is very interested in the Cell concept and approach to computing. Nor are they alone in this, because the Cell concept has a number of things going for it.

This is quite far removed however from the actual chip that will presumably going into the PS3. I don't think anyone outside IBM or exceedingly well connected can make any statements about IBMs plans for that particular implementation or closely related variations thereof.

 

[Vince]

That group doesn’t need to be explicitly doing things themselves, they can be investing in other companies and working with those companies to forward their goals. This is what w know they are doing with ATI and IBM now, but we don’t know how long they have been doing it beforehand and with who else they are working with.

Exactly and I didn't deny this, in fact I mentioned it specifically under the idea of distributed design. Yet, we still have problems. Namely the question you didn't answer, When did the tenders go out?

Also, you haven't touched upon cthellis42's comments, which raise interesting point wrt joint inter-company development and IP-sharing without legal notice. I mean, we can play these back and forth games all day bud, it still isn't going to chnage the fact that Microsoft had nothing near the level of development that Sony undertook. For if it did, there would have had to be specific SEC disclosures - not to mention I think you know pretty darn well when MS approached the IHVs.

However, the point being is that ATI were hiring in March for the XBox 2 project, some 6 months before announcement, and that work was already well underway at that point.

Ok, and again this comment basically admits that any form of cooperative development on the scale of Cell didn't take place. Sony was hiring in early 2001 for their project, as was IBM and Toshiba as they broke ground on STI-Austin.

And yet you cite the levels of investment that Toshiba and IBM have put in as a reason for PS3’s success. Sorry, but it seems to me that doesn’t necessarily follow to PS3. PS3 is but one application of Cell. Are IBM / Toshiba interested in the wider applications of Cell or are they interested in PS3 specifically? On what scale of importance to those companies is the PS3 specifically in relation to Cell?

PS3 is the application for Cell though. Do you think Sony could move into such an alliance (which perhaps rivals only Intel) without PS3? The fact that the same guys who patented the Emotion Engine patented Cell? That this was pushed by Ken Kutaragi and [formerly] Okamoto? Why would the initial work be initiated by SCE with Sony Group signing on later? For example:

Monday, 12 March, 2001. IBM wins Playstation 3 contract

Or even comments coming from interns working at IBM on Cell who have stated that the first application of it is for Playstation3. Matthew something or another is one such guy who comes to mind - I'm sure someone's seen his webpage. As stated by another intern:

Analyzed the programmability of the Cell processor--a joint IBM/Sony/Toshiba venture to develop a tera-flop
processor

Ahh yes, since we know these companies would design high preformance logic of that scale for tasks other than... hmmm... Common' Dave, you're grasping at straws.

 

[Dave Baumann]

Namely the question you didn't answer, When did the tenders go out?

I don't know exactly when they went out I don't know exactly when they went back, all I do know is one company had indications of one in 2002, but I don't know anything to say of the other companies.

For if it did, there would have had to be specific SEC disclosures - not to mention I think you know pretty darn well when MS approached the IHVs.

Why haven't there been specific SEC disclosures from the companies MS have already employed?

Ok, and again this comment basically admits that any form of cooperative development on the scale of Cell didn't take place. Sony was hiring in early 2001 for their project, as was IBM and Toshiba as they broke ground on STI-Austin.

No, all this confirms is that work was underway before the official public development announcement, it doesn't confirm or deny that work was underway even prior to that.

Do you think Sony could move into such an alliance (which perhaps rivals only Intel) without PS3?

I didn't say that Vince. Obviously one of the primary interests for Sony with "Cell", at least initially, is with PS3 - the question is what importance is the development and R&D from the others your are so happy to quote about is on the PS3 specifically? Again, PS3 is but one potential application of the Cell concept, first or not, but does IBM, for instance, see that as the primary application? The R&D $$$ they are pumping, as far as they are concerned, may not necessarily be for the interests of the PS3 as the goal, but for their further purposes of Cell with PS3 is but one application of Cell.

 

[Vince]

I don't know exactly when they went out I don't know exactly when they went back, all I do know is one company had indications of one in 2002, but I don't know anything to say of the other companies.

Exactly! Hmm... and I wonder if said company is projected to have an integral role in any XBox design. *shrug* Hey, I'm not going to tell you what to believe, but in my mind... the pieces fit.

Why haven't there been specific SEC disclosures from the companies MS have already employed?

Um, there have been:

Microsoft and IBM Announce Technology Agreement. IBM Technology to Power Future Xbox Products and Services. REDMOND, Wash., and EAST FISHKILL, N.Y. -- Nov. 3, 2003 -- Microsoft Corp. today announced that it has entered into a semiconductor technology agreement with IBM Corp

I can find the others, but we've all seen them. Obviously due to SEC disclosure requirements they had to announce this within a given timeframe.

Ok, and again this comment basically admits that any form of cooperative development on the scale of Cell didn't take place. Sony was hiring in early 2001 for their project, as was IBM and Toshiba as they broke ground on STI-Austin.

No, all this confirms is that work was underway before the official public development announcement, it doesn't confirm or deny that work was underway even prior to that.

Ok, lets put it this way Dave since you've basically ignored everything cthellis said. I didn't see Microsoft, IBM and ATI building a joint development complex in 2001 - why is that? Following this logic, exactly how does a company such as Microsoft go about the kind of IP/Roadmap/Technology sharing and tranfer that Sony did in 2001 without SEC disclosures? How do they avoid lawsuits in this case, one which is much worse than the 3dfx/Sega scenario that had legal rammifications?

Point is, you're playing semantic games.

I didn't say that Vince. Obviously one of the primary interests for Sony with "Cell", at least initially, is with PS3 - the question is what importance is the development and R&D from the others your are so happy to quote about is on the PS3 specifically? Again, PS3 is but one potential application of the Cell concept, first or not, but does IBM, for instance, see that as the primary application? The R&D $$$ they are pumping, as far as they are concerned, may not necessarily be for the interests of the PS3 as the goal, but for their further purposes of Cell with PS3 is but one application of Cell.

Ok, even if you ignore everything I posted and your common sence - the fact is IBM, Toshiba and Sony have been developing a scalable microarchitecture since 2001 that will be the basis of the Broadband Engine from what we've seen. I've posted links in the past to several patents from members of IBMs contact team that defined Cell relating to it's microarchitecture. Heck, the architecture speaks for itself - but believe what you want.

 

[Dave Baumann]

Exactly! Hmm... and I wonder if said company is projected to have an integral role in any XBox design. *shrug* Hey, I'm not going to tell you what to believe, but in my mind... the pieces fit.

No.

I can find the others, but we've all seen them. Obviously due to SEC disclosure requirements they had to announce this within a given timeframe.

That’s a press release, not an SEC disclosure. We’ve yet to see any specific SEC disclosures from ATI, for instance, about monetary terms or potential business risks.