ATI engineering must be under alot of strain. MS funding?

[Joe DeFuria]

But it makes no sense either for nVidia to insist they still have the development relationship with Microsoft... if they don't.

But I don't recall nVidia making any such insistence after ATI announced their technolgy development deal with MS.

See, same logic Hehe.

Actually, it's not. If nVidia comes out on Monday and says "we're still in talks and have a development relationship of our own for future x-box technology", that would be something different.

 

[KimB]

Actually, it's not. If nVidia comes out on Monday and says "we're still in talks and have a development relationship of our own for future x-box technology", that would be something different.

Which may be possible. From what I've seen so far, nVidia has the superior motherboard chipset technology... That would be an interesting turn of events

More seriously, though, I doubt nVidia will have anything to do with the second x-box. I expect nVidia wants to focus on their core graphics technology, to attempt to re-cement their leadership in the PC graphics arena.

 

[Joe DeFuria]

Actually, it's not. If nVidia comes out on Monday and says "we're still in talks and have a development relationship of our own for future x-box technology", that would be something different.

Which may be possible.

Or not.

http://www.ebnonline.com/showArticle.jhtml?articleID=13100350

Nvidia Corp. has confirmed that it will not supply the graphics processor for the next-generation Microsoft Xbox, which instead will feature a graphics engine designed by Nvidia rival ATI Technologies Inc...

....

Still, Nvidia said it was disappointed to have lost a chance to support the Xbox 2, which is expected to be released in 2005 or 2006. "As far as our own business and desires are concerned, we don't like losing design wins," said a spokesman, who noted that Nvidia will continue to supply the original Xbox and work with Microsoft to develop drivers, the DirectX API, and the Microsoft Media Center PC platform.

 

[WaltC]

Which may be possible. From what I've seen so far, nVidia has the superior motherboard chipset technology... That would be an interesting turn of events

More seriously, though, I doubt nVidia will have anything to do with the second x-box. I expect nVidia wants to focus on their core graphics technology, to attempt to re-cement their leadership in the PC graphics arena.

How much good does the nf2 do for the PIII in the xBox? I agree nVidia has a very good AMD chipset (as I use one at home.) Works very well with ATi R3x0 graphics cards--which is a compliment to nVidia in how well the chipset supports the Intel specifications (better than Intel-chipset boards it would appear, in some cases.) Heh... Wouldn't it be a hoot if M$ also switched to AMD for the xBox2 even as it replaces nVidia with ATi?

I think people are kind of missing the obvious here: M$ has made its decision relative to what it knows of both companies' upcoming technologies, which presumably is much more than we do. Also, despite whatever quibbles M$ has had with nVidia over the last year, M$ has been able to see as well as anybody how stellar ATi's product execution has been and how shabby nVidia's has been--and the combination of these factors alone could have had them leaning toward ATi for several months. I think it's important to remember that the problems nVidia's had over the last year date back to key decisions the company made well over a year ago. The same thing could be said about ATi's successes in the same time frame. I can easily see how M$ might conclude it was looking at a trend.

 

[KimB]

Actually, it's not. If nVidia comes out on Monday and says "we're still in talks and have a development relationship of our own for future x-box technology", that would be something different.

Which may be possible.

Or not.

http://www.ebnonline.com/showArticle.jhtml?articleID=13100350

Hehe, I guess I didn't specify what I meant. Remember that nVidia supplied the motherboard chipset for the original x-box, not just the graphics processor (well, the processor was part of the chipset, but there are two chips...)

 

[KimB]

I think people are kind of missing the obvious here: M$ has made its decision relative to what it knows of both companies' upcoming technologies, which presumably is much more than we do.

And you're also seeming to assume that this decision is entirely pragmatic, and entirely motivated by performance/features.

It could be due to a number of things:

1. Possible friction between Microsoft and nVidia over DX9 specification.
2. The price mediation on the original X-Box chipset.
3. nVidia not wanting to do the chip.
4. Other possibilities I have yet to think about...

 

[Joe DeFuria]

Hehe, I guess I didn't specify what I meant. Remember that nVidia supplied the motherboard chipset for the original x-box, not just the graphics processor (well, the processor was part of the chipset, but there are two chips...)

Well lemme put it this way:

Assuming ATI does the graphics chip, (fairly safe, no?) I will eat my hat if there is any nVidia tech in the box. Certainly not in core chipset tech, for which ATI already has products...

 

[Ailuros]

I expect nVidia wants to focus on their core graphics technology, to attempt to re-cement their leadership in the PC graphics arena.

That would make more sense if NVIDIA wouldn´t research other alternative markets at the moment, to shift resources to other projects, than to concentrate them more on the PC graphics market.

MediaQ should ring the first bell there.

1. Possible friction between Microsoft and nVidia over DX9 specification.
2. The price mediation on the original X-Box chipset.
3. nVidia not wanting to do the chip.
4. Other possibilities I have yet to think about...

1. Or the competition just delivered a better sollution while defining DX9.0 specs.
2. Could be; but then again it´s probably the same reason why MS decided to just lisence a graphics core.
3. I doubt that.
4. See first paragraph above.

 

[KimB]

Well lemme put it this way:

Assuming ATI does the graphics chip, (fairly safe, no?) I will eat my hat if there is any nVidia tech in the box. Certainly not in core chipset tech, for which ATI already has products...

Very true...but I don't think ATI has anywhere close to the same sound quality as nVidia has right now.

Anyway, yes, it is highly unlikely.

 

[Joe DeFuria]

Very true...but I don't think ATI has anywhere close to the same sound quality as nVidia has right now.

ATI has as good or better sound quality now than nVidia did when it won the x-box 1 contract.

 

[KimB]

1. Possible friction between Microsoft and nVidia over DX9 specification.
2. The price mediation on the original X-Box chipset.
3. nVidia not wanting to do the chip.
4. Other possibilities I have yet to think about...

1. Or the competition just delivered a better sollution while defining DX9.0 specs.
2. Could be; but then again it´s probably the same reason why MS decided to just lisence a graphics core.
3. I doubt that.
4. See first paragraph above.

1. Which leads us right back to my #1. It's all a matter of perspective. I contend that a large part of the reason that the NV3x performs poorly is because of how the DX9 spec was defined.
3. nVidia already hinted previously that they may not want to produce the next X-Box chip (can't remember the exact quote, but it had something to do with everything being right for the first X-Box deal, and that everything would have to be right again for the next...). It is possible that at least some people at nVidia feel that the work on the X-Box is part of the reason for the delays and flaws of the NV3x architecture.

 

[Joe DeFuria]

1. Which leads us right back to my #1. It's all a matter of perspective. I contend that a large part of the reason that the NV3x performs poorly is because of how the DX9 spec was defined.

Ignoring the fact that even if that's true...ATI had a DX9 part out well in advance of nVidia's part, which can be argued is the reason why DX9 is defined the way it is.

3. nVidia already hinted previously that they may not want to produce the next X-Box chip

I thought I just quoted a few posts above how nVidia is disappointed at losing any contract?

 

[FUDie]

It's all a matter of perspective. I contend that a large part of the reason that the NV3x performs poorly is because of how the DX9 spec was defined.

What an odd way to look at it. I contend that a large part of the reason that the NV3x performs poorly is because of how the chip was designed.

There, that makes more sense.

-FUDie

 

[KimB]

1. Which leads us right back to my #1. It's all a matter of perspective. I contend that a large part of the reason that the NV3x performs poorly is because of how the DX9 spec was defined.

Ignoring the fact that even if that's true...ATI had a DX9 part out well in advance of nVidia's part, which can be argued is the reason why DX9 is defined the way it is.

It doesn't matter why the DX9 spec was defined the way it is... (in the context within which I was speaking).

3. nVidia already hinted previously that they may not want to produce the next X-Box chip

I thought I just quoted a few posts above how nVidia is disappointed at losing any contract?

Note the wording: disappointed to have lost the chance. Perhaps they felt that there was still another 6-12 months before that decision had to be made.

 

[KimB]

What an odd way to look at it. I contend that a large part of the reason that the NV3x performs poorly is because of how the chip was designed.

There, that makes more sense.

-FUDie

It's all a matter of perspective. Keep in mind that one big reason that the NV3x performs poorly in DirectX 9 is the floating-point performance.

Who says that FP performance was needed at this stage of 3D graphics? Why isn't there the ability to drop to integer calculations in DirectX 9? This isn't about reducing quality: FP calculations can still be used where they're needed. But how often are they needed? That's the main question.

Anyway, my point is that the NV3x is designed to do both integer and FP performance in the same shaders. Whether or not that design decision was bad should have depended upon games. But DirectX 9 ensures that the NV3x will either perform poorly, or will not operate to spec (well, the NV35 should do okay...).

 

[WaltC]

And you're also seeming to assume that this decision is entirely pragmatic, and entirely motivated by performance/features.

It could be due to a number of things:

1. Possible friction between Microsoft and nVidia over DX9 specification.
2. The price mediation on the original X-Box chipset.
3. nVidia not wanting to do the chip.
4. Other possibilities I have yet to think about...

Yes, but don't you think that anything other than a pragmatic decision would be very irresponsible and reckless of M$...?... The simplest explanation based on M$'s comments to date would be that of the two companies M$ felt it would do better overall for xBox2 with ATi...Some of the things you mention may have been a part of that calculation, but I doubt any one of them was the proverbial straw. But, based on the fact that nVidia has never said of xBox2 that "We don't want it," and quotes as recounted in the article Joe links above by nVidia which lament their loss of the contract, I think we can conclude that #3 above was not a possibility. After all, the linked article below states that nVidia's revenue from xBox for 2003 is $445M. That's a huge chunk of nVidia's revenue stream. I doubt that nVidia would part with that kind of potential lightly.

Brian Alger, an analyst with Pacific Growth Equities Inc., said Thursday the deal between ATI and Microsoft is a negative turn of events for Nvidia.

"The current generation of Xbox contributed $445-million to Nvidia's top line in 2003 (its highest year) and in 2004 and 2005 we believe it would contribute $292-million and $200-million, respectively."

http://www.globeinvestor.com/servlet/ArticleNews/story/RTGAM/20030814/wbati0814

Indeed, keeping the xBox contract would seem very pragmatic for nVidia...

 

[Joe DeFuria]

It doesn't matter why the DX9 spec was defined the way it is...

It doesn't matter why NV3x performs as poorly as it does wrt shaders.

Note the wording: disappointed to have lost the chance. Perhaps they felt that there was still another 6-12 months before that decision had to be made.

:? Perhaps they're not too bright then?

 

[Ailuros]

I contend that a large part of the reason that the NV3x performs poorly is because of how the DX9 spec was defined.

So there isn´t a chance in a million that the design was actually impressive but several mistakes happened while transitioning it from paper to silicon?

I´d personally call the design on paper actually brilliant....

nVidia already hinted previously that they may not want to produce the next X-Box chip (can't remember the exact quote, but it had something to do with everything being right for the first X-Box deal, and that everything would have to be right again for the next...).

Or they just knew while making those statements (I did listen to almost every live broadcast in the past months) that someone else already had won the XBox2 design. How long ago did rumours appear about ATI winning said deal? Those statements you mention where NOT made before those rumours appeared.

It is possible that at least some people at nVidia feel that the work on the X-Box is part of the reason for the delays and flaws of the NV3x architecture.

Coming from official NV lips it was stated a couple of months ago that if they don´t win the XBox2 deal, that they´ll shift to alternative markets, the PDA/mobile market included.

Of course does your sentence above make perfectly sense, but I don´t see why resources then get shifted elsewhere and more specifically (as it sounded) those resources that were dedicated to Xbox1.

 

[WaltC]

It's all a matter of perspective. Keep in mind that one big reason that the NV3x performs poorly in DirectX 9 is the floating-point performance.

Which of course has nothing to do with M$...or ATi, that I can see. Agreed that nVidia has its own narrow perspective here.

Who says that FP performance was needed at this stage of 3D graphics?

Everyone except nVidia?

Why isn't there the ability to drop to integer calculations in DirectX 9? This isn't about reducing quality: FP calculations can still be used where they're needed. But how often are they needed? That's the main question.

Here's a simpler question: Why couldn't nVidia design a better fp chip?--The "market leader" and all of that...That's the thing about leadership. Fall behind and you lose the title.

Anyway, my point is that the NV3x is designed to do both integer and FP performance in the same shaders. Whether or not that design decision was bad should have depended upon games. But DirectX 9 ensures that the NV3x will either perform poorly, or will not operate to spec (well, the NV35 should do okay...).

Games usually follow the hardware--not the other way around. It's a good thing, too, because if hardware followed the games we'd all still be using GLIDE....

Suggestion Chally: stop defending them--you'll just get stuck in a loop...

 

[Megadrive1988]

Still, Nvidia said it was disappointed to have lost a chance to support the Xbox 2, which is expected to be released in 2005 or 2006. "As far as our own business and desires are concerned, we don't like losing design wins,"

Heh, this isn't the first design win for a game machine that Nvidia has lost.

Sometime in 1995-1996 Nvidia lost a design win (the disaster NV2 chip) for either a new Sega console, or an upgrade for the 32-Bit Saturn. And the NV2 chip was not for Dreamcast, as Dreamcast was a specific technology (PowerVR2, Hitachi SH4, MS WinCE)

Actually, you can make that 3 lost game system design wins for Nvidia as a whole, since the Sega Black Belt console prototype was rejected in 1997, which had a 3Dfx chip, and 3Dfx is part of Nvidia now!

Though most people will only remember Nvidia losing one console contract, (loss #3) the one for Xbox 2, since the NV2-SEGA thing (loss #1) is so obscure, and loss #2 (BlackBelt) 3Dfx was a seperate entity in 1997.