Official: ATI in XBox Next

[Megadrive1988]

zurich, IMO this isn't the final deal. Assuming that the ATI offer is a r5XX derivative then I really don't think MS is interested. MS wants something that match or exceed the PS3 and it's Cell computing, which I doubt can be achieved with a 90nm part. Perhaps ATI is going to make a 65nm r6XX derivative or whatever, but in that case one would have to seriously consider the enormous development on ATI and whether they can handle 2 massive console GPU projects and still make something in the PC graphics market.

The part about a r5XX or r6XX derivative is most likely true. although I dont see why it would be such an enourmous problem for ATi since they would be developing r500 and r600 anyway, for the PC. i listened to the ATi presentation http://tinyurl.com/k345 the last few min deals with this issue to some degree. i believe it is simply a matter of adapting, modifying and tweaking an r500 or r600 core for the next Xbox. from what i understand, it the deal might even be that MS is simply licensing some IP from ATi, and someone else will design and manufacture the graphics processor (dont hold me to that, its only something ive read on B3D and elsewhere) MS holds some IP too and they probably want to get that worked into hardware. In all likelyhood ATi will do all the development but it will be not at all like what Nvidia did on Xbox1 where they diverted 350+ engineers to the XGPU.

 

[Josiah]

IIRC the team that's working on R500 contains most of the ArtX people. I doubt work on R600 has even started.

 

[Brimstone]

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.

When I saw IBM and nVidia team up and basically gain access to STI's advancements combined with some comments I heard a while back from a little bird - I thought it was over. nVidia has the balls to push and stick with it. When 3dfx was in the corner touching itself with .25um, nVidia was on Cu utilizing 180nm and utterly destroying 3dfx in everyway, performance, features, per IC cost, yields.. it goes on and on.

And I see the same now. While nVidia is testing with 130nm Low-K dielectrics, ATI is off pissing in the wind on a 150nm process. Sure, nVidia had problems this time, but it's the exception. What's ATI going to do when nVidia is utilizing a derivative of STI/AMD's 10S or 11S (11S is Cell's 65nm process slated for 2H 2004/1H 2005 production) process at IBM? Have you fanpeople tell us that SOI isn't necessary? That the thermal or 30% performance increase seen on Power4 isn't that big of a deal? That TSMC's sub-par roadmap and execution of <100nm is adequate? Don't even get me started on UMC, are they serious in going alone for 90nm and below then everyone is concentrating their R&D? HA! Give me a break.

Today is the first day I can say that Sony will be alright, that if I was Okamoto or Ken, I'd be happy as a pig in shit.

that was one helluva killer post

It would seem a little premature to declare Sony is out of the woods yet. So far the only thing revealed is ATI is designing a VPU for Microsoft. What isn't known is who will be fabbing it. In the case of an ATI design for Nintendo the partner for them would be NEC most likely. Who will it be for Microsoft?

Micron a while back showed some intrest in graphics chips by purchasing Rendition and then fabbing a chip with eDRAM. It never went into production but it showed an intrest in that area of manufacturing. Don't know if it would make economic sense for Micron and Microsoft to get involved in some deal like this, but this would be an intresting combination. I do think Micron would be more likely than Intel though, just because of the margins. RAM is a low margin buisness and certainly working on the manufacture of a console VPU isn't going to be high margin.

There is no evidence to suggest Micron will be involved with the VPU. Just some speculation on what Sony wouldn't want to see. If an announcement comes out stating TSMC will be manufacturing the VPU, I could see Sony breathing a sigh of relief then. It's just too early to say for sure what the XB2 will look like.

 

[nonamer]

The 350 engineers nVidia had to use probably came mostly from the development of the audio chip, motherboard, and custom features in the NV2a not found in the NV2x series, hence they had to design the Xbox chip pretty much in parallel to their PC GPUs. It looks like ATI will simply design a R500 or R600 with the PC in mind and then simply upgrade/modify for the Xbox2 a la R350 and/or R420 afterwards. This could save a lot of money, but it could also be weak on power. It will take time for the modifications to be done, whereas a mostly custom designed chip could use that extra time to make something really impressive. That's what nVidia is did for the Xbox 1, and the costs where so great that they're now reluctant to do it again.

So I suppose for Microsoft choosing ATI is choosing the cheaper one but less powerful, and choosing nVidia is the more expensive but more powerful. I suspect that MS wants the more powerful, thus I call the idea that ATI has won the GPU contract for Xbox"2" into question. However, this is all speculation so it all really pointless to write this. Just wait and see is the only thing we can do.

EDIT: There's one thing I don't get about this whole Xbox2 thing: http://translate.google.com/transla...om/ar/2328-1.html&hl=en&lr=&ie=UTF-8&oe=UTF-8

Clubic.com: Recent rumours and persistentes indicate that you dropped the project XBox 2. Why?

J-h H.: Normally one does not make comments on the rumours resulting from the Web, but for this one I will make an exception. We did not give up the project XBox 2. Our relation with Microsoft and the XBox group is good. We continue to work narrowly with Microsoft on a certain number of initiatives and we are always under discussion with them about XBox 2.

This was two days before the ATI announcement. Was J-h H making up stuff or was he blindsided by the announcement?

 

[V3]

Wow Been away for several days and read this ?

The rumour was true. Both Nintendo and MS to ATI, Well Done ATI !

NV looks like taking the 3DFX route to destruction.

 

[Dave Baumann]

This was two days before the ATI announcement. Was J-h H making up stuff or was he blindsided by the announcement?

Or could it be that the deal was just very, very fresh when the announcement was made...?

So I suppose for Microsoft choosing ATI is choosing the cheaper one but less powerful, and choosing nVidia is the more expensive but more powerful.

What on earth is your continual instence for this? Just it just not enter you realm of thinking that ATI could possibley have better tech for the direction that MS want to take for the XBox? Its not as though they haven't had good technology for the past year or so.

 

[Megadrive1988]

It will take time for the modifications to be done, whereas a mostly custom designed chip could use that extra time to make something really impressive. That's what nVidia is did for the Xbox 1, and the costs where so great that they're now reluctant to do it again.

I have to disagree here. the Nv2A was NOT all that impressive compared to Nv20/Nv25 NV2A is less power / performance than Nv25. the Xbox was originally supposed to have an Nv25 or slightly better. not worse. now although the Nv2A has *some* features not present in either the Nv20 or Nv25, its nothing that extrodanary or mindblowing. in fact, the Nv2A is quite a bit less powerful than Nv25.

The original Xbox spec was for Nv25 or slightly better. what we got was somewhat weaker than Nv25, even though Nv2A has a couple features not found until Nv30. recall the original fillrate and texel rate for the Nv25- based Xbox was 4.8 Gpixels and 4.8 Gtexels. what we got was 932 Mpixels and 1864 Mtexels. with geometry it was going to be 150m T&L'd polys/verts but we only got 116m T&L'd polys/vertices. Honestly, whatever major effort Nvidia expended themselves on, to improve NV2x so much, into Nv2A, did not mount to all that much. basicly, the biggest difference was the 2nd vertex shader. the rest were just tweaks and minor additions/extensions. the Nv2A does not even have the decent FSAA unit that the Nv25 has. remember how great those pre-rendered tech demos were, that were ment to show off what Xbox would be capable of? if Nvidia had delivered an Nv25 or Nv25+ for Xbox with full geometry & fillrate spec and MS agreed to put 128 MB memory in it, I'll bet Xbox could do R-T graphics extremely close to those demos. i have all the consoles so i can say these things.

So I suppose for Microsoft choosing ATI is choosing the cheaper one but less powerful, and choosing nVidia is the more expensive but more powerful. I suspect that MS wants the more powerful, thus I call the idea that ATI has won the GPU contract for Xbox"2" into question. However, this is all speculation so it all really pointless to write this. Just wait and see is the only thing we can do.

Again, I gotta pause here, and say that ATi is *More Powerful* and *ahead* of Nvidia here and now. even with Nvidia having the the Nv35. MS selected ATi since, not only is ATi ahead of Nvidia right now, but MS looked at all the graphics tech in development, and almost certainly felt that ATi had the best performance, features, inovations, effiency/etc for the cost, compared to what Nvidia or anyone else was developing.

 

[zurich]

*ahem*, Xbox was said to do 4.8 ANTIALIASED gpixel/gtexel per second 8)

 

[RaolinDarksbane]

I already posted this on another thread, but I am doing it again for this one.

I have a conspiracy theory about MS, Nvidia, Nintendo, and ATI. What if MS is giving Nvidia a chance to make a better chipset than ATI(then they will switch).. ATI will know that Nvidia has a chance so they would focus a lot of attention into making a better chip than Nvidia, which is presumably take some resource away from Nintendo's chip. In the end, MS will win either(choosing ATI or Nvidia) way by getting a superior chipset than their competitor.

If you think about it, if MS had sticked with Nvidia, they wouldn't know how powerful Nintendo chip would be(ATI could possibly developed a superior chip than Nvidia). The XBOX2 technological advantage wouldn't have been so sure. By doing this, they are more likely to have a more superior chip than Nintendo.

 

[RaolinDarksbane]

Technology - PC World

Future XBoxes Will Use ATI Graphics
Thu Aug 14, 3:00 PM ET

Tom Krazit, IDG News Service

ATI Technologies has secured a "future technology agreement" for upcoming versions of Microsoft's XBox (news - web sites), taking over the business formerly held by rival graphics vendor Nvidia, ATI has announced.






ATI will license graphics technology to Microsoft for the next version of the XBox, says Chris Evenden, an ATI spokesperson. Microsoft will attempt to surpass gaming console leader Sony Computer Entertainment's Playstation 2 (news - web sites) with the next generation of the XBox. The Xbox has overtaken longtime gaming company Nintendo (news - web sites)'s GameCube, but trails Sony among console gamers.


Financial terms of the deal were not disclosed, Evenden says.

Nvidia Backs Out

Nvidia had recently signaled the company wouldn't be interested in pursuing the next-generation XBox business, says Jon Peddie, principal analyst for Jon Peddie Research. The company recently said it was looking to cut costs, and a number of industry observers have indicated that Microsoft was looking for another partner, Peddie says.


Microsoft and Nvidia have sparred over pricing issues since their initial agreement was signed in 2000. The companies brought in an arbitrator in 2002 to hear a dispute over the price Microsoft pays for Nvidia's chips. But they agreed to settle their differences in February, and talked of a future partnership to reduce XBox costs.


At one point in 2002, Nvidia was also left with a sizable amount of unusable inventory after Microsoft changed the security settings for the XBox, forcing Nvidia to absorb that cost. ATI will not carry inventory for Microsoft, Evenden says.

Graphics Roulette

Sony and Nintendo have both designed their own graphics controllers for their consoles using intellectual property from other companies, according to Peddie. In fact, Nintendo uses ATI's technology in the GameCube, an arrangement Microsoft was aware of as it evaluated ATI's technology, Evenden says.


Those companies have greatly lowered the cost of their graphics engine with the strategy, but the approach can be tricky, Peddie says.


"By going for an intellectual property deal, Microsoft now has to get involved with the fabs and integrated circuit design. Although the company has had experience doing things like that with its WebTV box and its set-top box, they may or may not have a team in place ready to go quickly. Building and testing high-performance integrated circuits at 0.13-microns or better is really tricky stuff," he says.


Microsoft would cut their overall costs, but would "take on quite a burden of technology management," Peddie adds.


Nvidia will continue to supply XBox graphics chips for the current generation of the console, the Microsoft spokesperson says. An Nvidia representative did not immediately return calls seeking comment.

http://www.pcworld.com/news/article/0,aid,112014,00.asp

 

[nonamer]

This was two days before the ATI announcement. Was J-h H making up stuff or was he blindsided by the announcement?

Or could it be that the deal was just very, very fresh when the announcement was made...?

That's the same as being blindsided. Surely nVidia must have some inkling of the progress of the ATI deal unless Microsoft intentionally kept them in the dark.

So I suppose for Microsoft choosing ATI is choosing the cheaper one but less powerful, and choosing nVidia is the more expensive but more powerful.

What on earth is your continual instence for this? Just it just not enter you realm of thinking that ATI could possibley have better tech for the direction that MS want to take for the XBox? Its not as though they haven't had good technology for the past year or so.

What's with all this fanATIcism on this board? Not trying to sound offensive, but it seems that I can't make a single statement that sounds bad for ATI without getting criticized (not just you Dave). The assumption I made was that ATI was simply going to modify an already existing chip whereas nVidia was planning on a more custom chip, more suited to the. Thus, ATI would be cheaper but overall less powerful.

I have to disagree here. the Nv2A was NOT all that impressive compared to Nv20/Nv25 NV2A is less power / performance than Nv25. the Xbox was originally supposed to have an Nv25 or slightly better. not worse. now although the Nv2A has *some* features not present in either the Nv20 or Nv25, its nothing that extrodanary or mindblowing. in fact, the Nv2A is quite a bit less powerful than Nv25.

The original Xbox spec was for Nv25 or slightly better. what we got was somewhat weaker than Nv25, even though Nv2A has a couple features not found until Nv30. recall the original fillrate and texel rate for the Nv25- based Xbox was 4.8 Gpixels and 4.8 Gtexels. what we got was 932 Mpixels and 1864 Mtexels. with geometry it was going to be 150m T&L'd polys/verts but we only got 116m T&L'd polys/vertices. Honestly, whatever major effort Nvidia expended themselves on, to improve NV2x so much, into Nv2A, did not mount to all that much. basicly, the biggest difference was the 2nd vertex shader. the rest were just tweaks and minor additions/extensions. the Nv2A does not even have the decent FSAA unit that the Nv25 has. remember how great those pre-rendered tech demos were, that were ment to show off what Xbox would be capable of? if Nvidia had delivered an Nv25 or Nv25+ for Xbox with full geometry & fillrate spec and MS agreed to put 128 MB memory in it, I'll bet Xbox could do R-T graphics extremely close to those demos. i have all the consoles so i can say these things.

It should noted that the NV2A came out before the NV25. Being less powerful is somewhat expected. Also, it is not *much* less powerful. About GF4 Ti 4200 or so in power from what I've heard, and the clock speed was downclocked from the original goal of 300Mhz to 233Mhz. NV2A could have been about as powerful as the GF4 Ti 4600 had things gone better. You're exaggerating the problems of the NV2A and your excuse at the bottom does not make you right.

 

[Brimstone]

I already posted this on another thread, but I am doing it again for this one.

I have a conspiracy theory about MS, Nvidia, Nintendo, and ATI. What if MS is giving Nvidia a chance to make a better chipset than ATI(then they will switch).. ATI will know that Nvidia has a chance so they would focus a lot of attention into making a better chip than Nvidia, which is presumably take some resource away from Nintendo's chip. In the end, MS will win either(choosing ATI or Nvidia) way by getting a superior chipset than their competitor.

If you think about it, if MS had sticked with Nvidia, they wouldn't know how powerful Nintendo chip would be(ATI could possibly developed a superior chip than Nvidia). The XBOX2 technological advantage wouldn't have been so sure. By doing this, they are more likely to have a more superior chip than Nintendo.

The only thing that really limits the power of chip designs is the technology the creates the chip. ATI could design a chip with twice the amount of transistors than anyone else for Nintendo, but if NEC doesn't have the fab power to pull of a design it wouldn't work.

My opinion is that the designs of a Nintendo and Microsoft VPUs will reflect the technology level of the foundaries making them.

Sony/IBM/Toshiba will be at .065 nm with eDRAM

Nintendo/NEC will be at .065nm??? with eDRAM (MOSYS)

Microsoft ?????

 

[Josiah]

My opinion is that the designs of a Nintendo and Microsoft VPUs will reflect the technology level of the foundaries making them.

that is not neccessarily true. for example, the Radeon 9800 is 150nm while the GeforceFX 5900 is 130 nm. but the 9800 is superior.

 

[Joe DeFuria]

What's with all this fanATIcism on this board? Not trying to sound offensive, but it seems that I can't make a single statement that sounds bad for ATI without getting criticized (not just you Dave). The assumption I made was that ATI was simply going to modify an already existing chip whereas nVidia was planning on a more custom chip, more suited to the. Thus, ATI would be cheaper but overall less powerful.

Problem is, you have zero basis for making that guess. If you had some rationale behind it, we could take it seriously. That's why the criticism. (That, plus all your previous talk about ATI being a "bluff" for what MS really wants...again with no basis.)

In other words...why would you make some assumption that ATI was going to simply "modify" an already existing chip? Why would nVidia do something more "custom?" If anything, I'd say it'd be the other way around...considering that it doesn't appear ATi is going to actually produce the chip.

NV2A could have been about as powerful as the GF4 Ti 4600 had things gone better.

But they didn't go better.

 

[PC-Engine]

http://www.extremetech.com/article2/0,3973,1220430,00.asp

Don't expect the graphics capabilities of future Nintendo and Microsoft products to be exactly the same, however, the ATI spokesman said. "Yes, we have different design teams working on them, with different requirements and different timetables," the spokesman said.

That pretty much confirms ATI is working on the next Nintendo console for those who doubted it.

The part about different time tables is interesting although it could refer to ATI's internal production schedule (different teams, different cores, different levels of completion).

Cool thx Cybamerc. Now we know it won't be identical chips.

What on earth is your continual instence for this? Just it just not enter you realm of thinking that ATI could possibley have better tech for the direction that MS want to take for the XBox? Its not as though they haven't had good technology for the past year or so.

I guess for some people a more powerful chip means it needs to suck more POWER and require a snow blower for cooling

I have a conspiracy theory about MS, Nvidia, Nintendo, and ATI. What if MS is giving Nvidia a chance to make a better chipset than ATI(then they will switch).. ATI will know that Nvidia has a chance so they would focus a lot of attention into making a better chip than Nvidia, which is presumably take some resource away from Nintendo's chip. In the end, MS will win either(choosing ATI or Nvidia) way by getting a superior chipset than their competitor.

If you think about it, if MS had sticked with Nvidia, they wouldn't know how powerful Nintendo chip would be(ATI could possibly developed a superior chip than Nvidia). The XBOX2 technological advantage wouldn't have been so sure. By doing this, they are more likely to have a more superior chip than Nintendo.

MS and N are buddies. The threat is SONY.

The only thing that really limits the power of chip designs is the technology the creates the chip. ATI could design a chip with twice the amount of transistors than anyone else for Nintendo, but if NEC doesn't have the fab power to pull of a design it wouldn't work.

My opinion is that the designs of a Nintendo and Microsoft VPUs will reflect the technology level of the foundaries making them.

The question isn't whether they can or cannot manufacture a huge chip. The question is cost and volume.

 

[Saem]

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.

Yes, except those have yet to really shine over well designed custom logic on a bulk CMOS process with copper interconnects. I cite Intel, Alpha, AMD (K7) and so on. Much of the stuff you're mentioning doesn't say much, thus far much of those technologies have been used in to keep up with Intel, Alpha, K7 ... in a reasonable amount of development time. The Power 4 doesn't use much custom logic compared to it's "peers". I'm not sure where you're getting the 30% figure from, IIRC in an IBM report that was a 25-30% increase with a fair bit of work in the 0.25um with diminishing returns in later processes.

When I saw IBM and nVidia team up and basically gain access to STI's advancements combined with some comments I heard a while back from a little bird - I thought it was over. nVidia has the balls to push and stick with it. When 3dfx was in the corner touching itself with .25um, nVidia was on Cu utilizing 180nm and utterly destroying 3dfx in everyway, performance, features, per IC cost, yields.. it goes on and on.

Maybe you're not paying attention but ATI and Intel have been trading their pokemon cards as well. And ATI's pikachu has got some new attacks. In anycase, AFAIK ATI did used a few "new" -- for them at least -- tricks. Mainly doing a bit of custom work to push the R3xx faster. It seems to have worked well.

And I see the same now. While nVidia is testing with 130nm Low-K dielectrics, ATI is off pissing in the wind on a 150nm process. Sure, nVidia had problems this time, but it's the exception. What's ATI going to do when nVidia is utilizing a derivative of STI/AMD's 10S or 11S (11S is Cell's 65nm process slated for 2H 2004/1H 2005 production) process at IBM? Have you fanpeople tell us that SOI isn't necessary? That the thermal or 30% performance increase seen on Power4 isn't that big of a deal? That TSMC's sub-par roadmap and execution of <100nm is adequate? Don't even get me started on UMC, are they serious in going alone for 90nm and below then everyone is concentrating their R&D? HA! Give me a break.

I see ATI doing more with less. Nvidia has always had one disadvantage against ATI, on a parity process, Nvidia chips always ran hotter. This seems to be the case to this day. Except things might have gotten worse we have ATI on a lesser process running faster. Hrm... your argumentation seems spun. Taking out the bias, you'd have my above statement, to respin it in ATI's favour... Nvidia needs the bleeding edge processes to keep pace with ATI. Oh, did I forget to mention, they delivered well before Nvidia? Funny, how that works.

Today is the first day I can say that Sony will be alright, that if I was Okamoto or Ken, I'd be happy as a pig in shit.

Icky, but I guess that's what they do. There are jokes in there about some cultural fetishes, but I'll refrain.

 

[Dave Baumann]

Nonamer – because your statements don't make much sense given the environment presently, and the deal announced. Lets have a look what you’ve said and pick it apart:

The 350 engineers nVidia had to use probably came mostly from the development of the audio chip, motherboard, and custom features in the NV2a not found in the NV2x series, hence they had to design the Xbox chip pretty much in parallel to their PC GPUs. It looks like ATI will simply design a R500 or R600 with the PC in mind and then simply upgrade/modify for the Xbox2 a la R350 and/or R420 afterwards. This could save a lot of money, but it could also be weak on power. It will take time for the modifications to be done, whereas a mostly custom designed chip could use that extra time to make something really impressive. That's what nVidia is did for the Xbox 1, and the costs where so great that they're now reluctant to do it again.

So I suppose for Microsoft choosing ATI is choosing the cheaper one but less powerful, and choosing nVidia is the more expensive but more powerful. I suspect that MS wants the more powerful, thus I call the idea that ATI has won the GPU contract for Xbox"2" into question.

a.) What makes NV2A so “special†in comparison to the rest of the NV2x series. I really don’t see that as being much beyond what they already had. The VS was already being beefed up in NV25, and the stencil rendering is a relatively easy tweak from MSAA.

b.) You say that ATI might take R500/600 and “simply upgrade or modify†for the XBox – how is this any different from the path NVIDIA took with NV2A?

c.) Why do you assume that it will just be simple modifications? The majority of the time spent on NV2A was not designing the IP (since all of it was already in-house at NVIDIA), but doing the chip layout, fabrication and debugging – being a licensing deal this is probably not going to be ATI’s concern since Microsoft are the ones choosing and dealing with how that is done. It takes far less time to alter the IP than it does to fabricate a chip.

d.) Even if this was “only†an R500/600 who’s to say that it wasn’t already sufficiently inline with MS’s needs in the first place?

e.) Why assume that ATI was “cheaper but less powerful†perhaps it was more powerful but at a higher cost and they were negotiating with NVIDIA for a cheaper solution.

f.) You mention that MS played other companies off each other for the XBox, which they did, and its very probable that they did the same for the XBox2 – however with xb1 they didn’t announce until they had struck the deal. NVIDIA said recently that they had switched their minds right up until the last day – it sounds as though the same thing has happened here, given JHH’s fresh comments. MS were playing ATI and NVIDIA off against each other until the last days but then went with ATI.

 

[bbot]

If MS chooses Intel to make the Xbox2 gpu, then it wouldn't be possible to make a gpu based on 65nm process? Intel will have fabs using 65nm lithography by 2005. One more thing. Didn't Intel announce a deal to license gpu designs from ATI earlier this year? If MS/ATi uses Intel to make their gpus the difficulty making chips using 90nm or less process disappears.

 

[Dave Baumann]

ATI and Intel announced an extension of the cross licensing deal. Terms of the deal were not announced, other than ATI got access to the 800MHz FSB.

 

[bbot]

Thanks for the correction, Dave. However Microsoft can still choose Intel to make the Xbox2 gpu.