Official: ATI in XBox Next

[Joe DeFuria]

Yeilds generally normalize fairly quickly.

Yes, tell that to both ATI and nVidia....who are busy churning out their bread -n- butter, low margin, high volumeparts (5200 and 9200), on 0.15. Why the hell might they have planned to do it that way with 0.13 yields normalizing "fairly quickly" and all?

This is a thought experiment, we are debating technology.

What business does a "thought experiment" have to do with Beyond3D. "We talk about Technology!" Despite that mantra, I did humor you.

It's hardly "imaginary" - the lines are being assembled now. Much more real then, say, the R400.

Vince, I see the point is lost on you. The bleeding edge 65nm process is imaginary, based on your premise of having the same yields as the mature 90nm.

That simply is not reality. By the time 65nm is getting similar yields and volume as 90nm...there's ANOTHER bleeding edge process just starting to come on-line.

Same as above.

Ahh, so unwilling to comment. Typical.

Same as above.

Yup, figured.

I have to admine that what I didn't figure on is you missing th obvious point. Try reading up again.

Their not debunked, you just refuse to give into my origional argument - which if you'd care to go back and read - was concerning the theoretical approach to IC design and the importance of lithography. Which you refuse to comment on... which is obvious why.

Vince, I give in to your argument freely. "All else being equal, a more advanced lithography is preferred."

No comments needed. I AGREE with that. Always have,

Now all you have to do comment on the fact that all else, in reality, is not equal. Capisce?

 

[cthellis42]

It's hardly "imaginary" - the lines are being assembled now. Much more real then, say, the R400.

Hasn't the R400 been shifted so that there is no more R400 anyway?

And the lines ARE being assembled. TSMC is assembling them, as they have done for years and will continue to do for years. They're not much in the habit of telling people which lines are dedicated to what purposes, and they have even fierce competitors running their top-end chips through them...

Or does TSMC construct imaginary chips too?

Why do you insists on restating the same old "all things being equal" argument, when it has been dubunked by everyone that's cared to comment on it in this thread?

Their not debunked, you just refuse to give into my origional argument - which if you'd care to go back and read - was concerning the theoretical approach to IC design and the importance of lithography. Which you refuse to comment on... which is obvious why.

True, not so much "debunked" as "none of us agree with you"... At least not to the level you seem to be taking it.

You also keep bringing back up the "theoretical approach" theme, but it was hinging on the "all things being equal" part as well. As far as I've seen we'd ALL agree with you--in theory--but find the application of said theory useless because all things are NEVER equal. (Plus, Joe--and most of us--have INDEED already commented on it. But you seem unswerving in your opinion that this is all that matters.)

Theories are nice. Theories look amazing on paper and are great to sigh at and long for. Theories can fail like nobody's business.

 

[Vince]

Yes, tell that to both ATI and nVidia....who are busy churning out their bread -n- butter, low margin, high volumeparts (5200 and 9200), on 0.15. Why the hell might they have planned to do it that way with 0.13 yields normalizing "fairly quickly" and all?

-Looks at Forum name- Pretends he cares.

Vince, I see the point is lost on you. The bleeding edge 65nm process is imaginary, based on your premise of having the same yields as the mature 90nm.

That simply is not reality. By the time 65nm is getting similar yields and volume as 90nm...there's ANOTHER bleeding edge process just starting to come on-line.

We don't know this. And it's irrelevent if you look at the precedent set by SCE and OTSS in prodicing the PS2 components. We already covered yeilds and they're lack of meaning in consoles. In fact, I did in my origional post:

http://www.beyond3d.com/forum/viewtopic.php?t=7406&postdays=0&postorder=asc&start=80[/url]]
To cause a 'borderline performance revolution' with type of architecture relies on bleeding-edge lithography and it requires the design team to push the process to the edge and beyond into the realm of poor-yields with the understanding that future lithography will bring the yields and costs under control. It requires massive investment like STI is doing ($8Billion in total) and it requires technologies like SOI/SS, Low-K, 65/45nm and lower lithography and other such advancements that are pushed hard.

You've also talked of "Imaginary lithography" several times dispite the fact that they presented it to the International Electron Devices Conference and there is volume of evidence that it will exist:

• http://www.reed-electronics.com/ele...=article&articleid=CA263017&rid=0&rme=0&cfd=1
• http://www.docmemory.com/page/news/shownews.asp?num=5080
• http://www.eetimes.com/semi/news/OEG20021205S0049
• http://www.sony.net/SonyInfo/News/Press/200212/02-1203E/
• http://www2.ccnmatthews.com/scripts/ccn-release.pl?/2002/12/02/1202176n.html
• http://www.electronicstalk.com/news/tos/tos165.html
• http://pcweb.mycom.co.jp/news/2003/06/10/18.html

Maybe we can have Dave edit your posts for posting the "same thing" and "objectivly wrong" comment like above. I'm got 5 saying he doesn't have the balls.

No comments needed. I AGREE with that. Always have,

Now all you have to do comment on the fact that all else, in reality, is not equal. Capisce?

I want to say Yes. But, I can't imagine a custom IC built (that's >250mm^2) on a true-lith revision that's come anywhere near the preformance of an inferior part. You just would never see it in for a console enviroment where the parts are custom ICs (and not superscalars or such).

 

[AzBat]

Maybe we can have Dave edit your posts for posting the "same thing." I'm got 5 saying he doesn't have the balls.

Think Vince. Maybe there's a reason why.

Tommy McClain

 

[cthellis42]

That simply is not reality. By the time 65nm is getting similar yields and volume as 90nm...there's ANOTHER bleeding edge process just starting to come on-line.

We don't know this.

I suppose technically we don't, but please state the historical precident almost ANYWHERE in modern times that bears this out.

We'll stretch silicon to the limit, to be sure, but we already have different materials and even whole different approaches to what we think of as the computer coming down the pipe, so while we may hit some lags between technology shifts, I can't imagine it being too great.

And if we DO get stranded on, say, 0.045 for a while, doesn't that mean more power to the developer who works out how to use it BEST, rather than who gets there FIRST?

No comments needed. I AGREE with that. Always have,

Now all you have to do comment on the fact that all else, in reality, is not equal. Capisce?

I want to say Yes. But, I can't imagine a custom IC built (that's >250mm^2) on a true-lith revision that's come anywhere near the preformance of an inferior part. You just would never see it in for a console enviroment where the parts are custom ICs (and not superscalars or such).

This would be, of course, a very long and involved discussion all on its own. And so far as I can tell from general impression, people lean in favor of a fully customized IC with tons invested into R&D as well. But Microsoft chooses not this model, nor does it seem will Nintendo too fiercely. Why seek to blame ATI or nVidia because of it?

Microsoft BRINGS IN PC parallels BECAUSE of its chosen design. Nintendo not as much, but neither does it really break any molds or push R&D to massive scale, which I would think puts it far behind S/I/T as well. And until we know more about the future of the Xbox2 and GC2, we are basically left with what we have--console analogies and PC analogies alike.

 

[Vince]

Think Vince. Maybe there's a reason why.

Then don't edit a comment that someone typed for being "repititions." Thats wrong. How many times in this forum do we have people like "Chap" (good guy, but) post the same thing literally 5 times. And you delete my post, which was rewritten because he feels like it?

I'm just curious, how many pro-ATI posts do you cull?

 

[AzBat]

Think Vince. Maybe there's a reason why.

Then don't edit a comment that someone typed for being "repititions." Thats wrong. How many times in this forum do we have people like "Chap" (good guy, but) post the same thing literally 5 times. And you delete my post, which was rewritten because he feels like it?

I'm just curious, how many pro-ATI posts do you cull?

Umm, one, I'm not Dave and two, I don't moderate any of the forums. If you don't like how Dave runs the forums, there's always the door...

Tommy McClain

 

[Dave Baumann]

Vince - its was edited because people have complained here and to me privately about the use of your language in this thread - I publically warned you that your posted would be editted, which you then chose not to heed. Not only that, you then proceeded to reinsert the exact same profanity

Maybe we can have Dave edit your posts for posting the "same thing" and "objectivly wrong" comment like above. I'm got 5 saying he doesn't have the ....

And it comes back to the same old thing Vince - methink you doth protest too much...

 

[Vince]

I suppose technically we don't, but please state the historical precident almost ANYWHERE in modern times that bears this out.

Intel moving to Cu, IIRC AMD is yeilding in the 90% range on their 130nm CU Palmino's. The move to 180nm was relativly stable.

Doesn't that mean more power to the developer who works out how to use it BEST, rather than who gets there FIRST?

Not in a closed box. Which is kinda fundimental to this issue. Yet everyone overlooks it.

This would be, of course, a very long and involved discussion all on its own. And so far as I can tell from general impression, people lean in favor of a fully customized IC with tons invested into R&D as well. But Microsoft chooses not this model, nor does it seem will Nintendo too fiercely. Why seek to blame ATI or nVidia because of it?

Which is basically what I've been saying regardling this discussion. Read my initial posts - they were all based on lithography and the advantages the bring and where the XBox's downfalls in this area are. Yet, people turn it into a ATI-nVidia debate and then you get this.

PS. I appologize for the language. I don't understand how you can let a thread on how many times a day you masterbate carry on while getting offended and censoring a word we've all heard - but to each his own.

 

[Dave Baumann]

Yet, people turn it into a ATI-nVidia debate and then you get this

Go back and read your first few posts in this thread Vince, you were and are just as much of an instigator in that as anyone else.

 

[Vince]

Yet, people turn it into a ATI-nVidia debate and then you get this

Go back and read your first few posts in this thread Vince, you were and are just as much of an instigator in that as anyone else.

And how many times since did I see that it wasn't necessary and out of hand and tried to stop it?

I find this whole thing funny that my core argument - the only thing I was truely fighting for concerning lithography - Joe just said, I AGREE with thatafter 12 off pages of off-topic, not related BS. I mean, think about it. How many times (even his, we'll I'll agree to this if you agree to something your not proposing) did we have utter denials, avoidences and fights over this?

And then you wonder why I get all... mad and repititious when people don't address the right issue for over 10 pages! It's so fundimental, it's not even a question. Give me a break. This was an excersise in hair-triggered damage control based on a certain IHV and nothing more.

 

[cthellis42]

I suppose technically we don't, but please state the historical precident almost ANYWHERE in modern times that bears this out.

Intel moving to Cu, IIRC AMD is yeilding in the 90% range on their 130nm CU Palmino's. The move to 180nm was relativly stable.

No, erm... You see that, uh... This was in your statement that we don't KNOW that there won't be a new process already in the wings while the current one matures. This is pretty much borne out so far...

That Moore is one harsh mistress! Er... master!

Doesn't that mean more power to the developer who works out how to use it BEST, rather than who gets there FIRST?

Not in a closed box. Which is kinda fundimental to this issue. Yet everyone overlooks it.

Let us then pretend that 0.045 will not be improved upon for 5 years. Is this not then a "locked box" from a litho standpoint?

Now Company A started developing to hit the cutting edge of it and drive to market before their competitors, but had to make some quick design decisions that impacted their long range outlook and yields and profit margins to get there, claiming perhaps a 6-12 month lead. Company B develops for said process keeping in mind the long term knowledge that they will be stuck on it for 5 years and design to make the best use of the process, and are not forced to make any concessions simply to be the first one there. Company A and B overlap by some 4-4.5 years in this cycle, but B takes better advantage of the technology, made fewer concessions, and garners more profit--having only yielded some time to market. In the intervening 4+ years, which stance would logically have more advantages?

Regardless, this is never the issue. If a company could always drive the process further, get the faster, and make the best use of it ANYWAY, there would be no competitors.

This would be, of course, a very long and involved discussion all on its own. And so far as I can tell from general impression, people lean in favor of a fully customized IC with tons invested into R&D as well. But Microsoft chooses not this model, nor does it seem will Nintendo too fiercely. Why seek to blame ATI or nVidia because of it?

Which is basically what I've been saying regardling this discussion. Read my initial posts - they were all based on lithography and the advantages the bring and where the XBox's downfalls in this area are. Yet, people turn it into a ATI-nVidia debate and then you get this.

I believe this primarily comes from the maligning comments you've made since the beginning of this thread in speculating about ATI, both using their PC track record against them and ignoring its importance, borrowing comparisons from one side while ignoring it from the other, and refusing to yield how any of us FOOLISH people could possibly make comparisons and draw consolusions from much physical evidence on how the next consoles JUST MIGHT HAPPEN, instead of bowing under the unrelenting pressure of pure theory.

PS. I appologize for the language. I don't understand how you can let a thread on how many times a day you masterbate carry on while getting offended and censoring a word we've all heard - but to each his own.

I'm going to go with "To tone things down before they become complete useless" for $200, Alex.

I find this whole thing funny that my core argument - the only thing I was truely fighting for concerning lithography - Joe just said, I AGREE with thatafter 12 off pages of off-topic, not related BS. I mean, think about it. How many times (even his, we'll I'll agree to this if you agree to something your not proposing) did we have utter denials, avoidences and fights over this?

We've never much argued about this in the FIRST place. I don't think anyone. Ever. But you certainly haven't limited your comments to lithography theory, and branch out to make certain assumptions and conclusions and connections which MANY people have found fault with, and you haven't sufficiently defended.

The "core arguement" isn't even an ARGUEMENT. And though you're trying to attach relevance, that does not MAKE it relevant. Everyone has basically been arguing all the next steps out you've taken, because you've basically been applying factors we do not agree with, and making claims that many of us do not find logically consistant.

 

[Dave Baumann]

I find this whole thing funny that my core argument - the only thing I was truely fighting for concerning lithography - Joe just said, I AGREE with thatafter 12 off pages of off-topic, not related BS.

Where did he, or anyone else disagree in the first place? Everyone here knows the benefits that advanded processes can bring, but then most people here seem to say that we don't consider that to be the only factor.

Quite frankly I'm at a loss to see what you are still arguing about. When your argument didn't hold much water in the PC space you then seemed to shift the tangent to "but we're talking about consoles here" and then we all see two instances in the most recent rounds of consoles where lithography that was in reach wasn't the be all and end all and other market pressures played a bigger hand.

As has been pointed out there are still too many unknowns about what is happening WRT to the XBox2 and the process used. If ATI is to produce the chip then as I have explained there is room for this to be on a more advanced process within than R500 their current framework for process operations. If ATI aren't to produce the chip then we currently have no idea who MS will select and hence what process will be used.

 

[AzBat]

I find this whole thing funny that my core argument - the only thing I was truely fighting for concerning lithography - Joe just said, I AGREE with thatafter 12 off pages of off-topic, not related BS.

First, here's exactly what Joe said:

Vince, I give in to your argument freely. "All else being equal, a more advanced lithography is preferred."

No comments needed. I AGREE with that. Always have,

If I understand Joe correctly he's still not agreeing with your stance...

Lithography is everything

With your reasoning it seems you believe it is required while Joe believes it's preferred, which is why he still seems to believe...

Lithography is NOT everything

Personally I would have to agree with Joe on this and I don't think I'm the only one either.

Tommy McClain

 

[nonamer]

Dave: You've just said some interesting things. Please elaborate.

PS: You don't seriously think that after 20 pages their would not be some major confusion going on?

 

[cthellis42]

As has been pointed out there are still too many unknowns about what is happening WRT to the XBox2 and the process used. If ATI is to produce the chip then as I have explained there is room for this to be on a more advanced process within than R500 their current framework for process operations. If ATI aren't to produce the chip then we currently have no idea who MS will select and hence what process will be used.

I rather think MS, as they did before with nVidia, will have to exist primarily withing ATI's PC gameplan.

If, however, MS were to come out with Intel (perhaps TSMC) firmly in toe NOW, slap $2 billion+ in cash on the table, toss some extra at ATI, and say "we're really looking to fab up. What can you give us?" then I can certainly see ATI drolling and setting their engineers loose like nobody's business.

 

[Vince]

With your reasoning it seems you believe it is required while Joe believes it's preferred, which is why he still seems to believe...

It will be required to compete against a competitor which does take the chance. For example, when the Graphic Synthesizer was first in silicon back in 1998, the best PC part was a TNT2 or Avenger.

XBox1 skewed the perspective since they [Microsoft] launched approaching 2 years after Sony first did. This time, for economic reasons (which I was hoping to avoid) they will launch in similar windows (as per MS comments) - thus process technology becomes vital. And when a competitor, (eg. STI) is working hard on an advanced process and have a history of fully utilizing it to make enormous IC's - you start to see why lithography is very important.

When your argument didn't hold much water in the PC space you then seemed to shift the tangent to "but we're talking about consoles here"

Quite frankly it's comments like this that's what caused this problem. I was never talking about a PC product. This was abundently clear from the beginning and only echoed again and again.

It was stated many times the differences between a PC and Console IC and the development that... dare I say... should go into each 5 year iteration.

Of course it didn't hold the same water in the PC arena... thats why we're debating in the console forum and not 3D architecture. Only part of the problem.

And then we all see two instances in the most recent rounds of consoles where lithography that was in reach wasn't the be all and end all and other market pressures played a bigger hand.

The only true instince is the NV2A, but that just shows the downfall of adapting a PC part for the console platform. Something I addressed as did Mfa and his posts on the CPU/dedicated IC's.

The GS and EE are shining examples of this given their advanced process, eDRAM, and physical size is the best attribute to show.

As has been pointed out there are still too many unknowns about what is happening WRT to the XBox2 and the process used.

Exactly which is why I said this to you:

This is the best comment you've made this far. It is a crapshoot, and all the special cases scenarios surrounding NV3x and R3x0 are utterly irrelevent from the standpoint of this discussion or any that is forward looking. Special cases, like what Joe and WHQL love to propogate, are nothing more than fanb0y excuses and arguing-points when you attempt to look at the underlying dynamics of a future situation. You must come to the realization that special cases are utterly irrelevent to any case but that - and as such aren't a good metric for projected observation.

But, what is a good indicator are general trends based on precedent and the fundimental and governing laws of design. You forgo all of these in your persuit of a half-truth based around a single case scenario - the R300. What happened before? What happened after? What you're doing is just as bad as Joe or Whql - you're focusing on a single case and not the more fundimental picture. It's like the Dow, there will be special cases that allow extrapolation for negative growth going forward, but if your attempting to look forward you must do what Laffer, Kadlec and Acampora do by looking at long-term growth averages and percentage based growth.

But appearently nobody saw it as everyone started pulling specific one-time situations, that should have no place in a lithography discussion, out and using them as arguing-points. Instead of people looking back on lithography and saying, "Yes, having the most advanced process that will allow for upwards or beyond a billion tranistsors would be adventageous to a IC designer for a closed box" or even a, "With an increase like that seen in lith of logic gates, it's definatly the pre-eminent decider in computation preformance - which happens to be gaining in precedence."

But no... people fight on stupid stuff that aren't even related. They avoid questions that could have ended this long before and bring in topics not even related... mades sence to me. Like I said, this was just an excersise in IHV damage control for a situation that doesn't need it.

If ATI is to produce the chip then as I have explained there is room for this to be on a more advanced process within than R500 their current framework for process operations. If ATI aren't to produce the chip then we currently have no idea who MS will select and hence what process will be used.

We shall see.

 

[V3]

Personally I would have to agree with Joe on this and I don't think I'm the only one either.

In consoles, designs are locked for 5 years, Improvements in lithography are the popular way to cut cost, thus I think it is everything in consoles space.

I am sure, after the Xbox experience, MS is also aiming to reduce cost fast for Xbox2.

 

[TheMightyPuck]

I'm not offended by the language but rather by the lack of rigor in the argument. Back off the weed kids.

*runs into bomb shelter

 

[jvd]

One last warning , Stop with the bad words and name calling or this thread is locked. Way to many complaints are coming in.