The LAST R600 Rumours & Speculation Thread

[Aerows]

I think they've had silicon working under Vista for a long time. How many spins have we heard about?

My reasoning behind that statement is that while many many vendors have had applications and hardware working under Vista, how long have they had it under released Vista? Microsoft has never been shy about changing code at the last minute.

 

[Tim Murray]

My reasoning behind that statement is that while many many vendors have had applications and hardware working under Vista, how long have they had it under released Vista? Microsoft has never been shy about changing code at the last minute.

Since November? That's when Vista came out, you know.

 

[Aerows]

Since November? That's when Vista came out, you know.

Right. And so far, it seems that many companies are struggling to get their products to function correctly/ fully supported state. While three months (let's say 4 for actual gold/finalized status) is an eternity in the technology business, simply not getting all of the correct information hampers even the mightiest of humans .

Is it really such a far-fetched notion?

 

[Natoma]

Great. My upgrade plans have gone straight out the window. Thanks for suggesting that I wait for R600 to make my choice, geo.

 

[3dilettante]

At this rate, we're going to need a LAST LAST R600 Rumours & Speculation Thread.

I'm not trying to be negative, but I'd be interested in a Pre-Post Mortem on R600, where we could bounce ideas off each other as to reasons why a chip like R600 would be delayed.

I suppose there are a few non-technical and non-compelling reasons.
1) AMD's strategy is to have some kind of unified launch of Barcelona and R600.
I don't find this particularly compelling, since it would be paper-launching a delayed server CPU with a delayed GPU lineup.
In the worst case (Barcelona is not significantly better than Kentsfield, R600 no better than G80 or G81), AMD gets to announce that it's an also-ran in two markets at the same time.

2) AMD hates to make money.

3) The late-stage development process was disrupted by the buyout.
I think the corporate side would be able to insulate the engineers from most of this.

4) Any other excuses people can think of?

Technical:

1) Yields are low.
Possible, this is a new design on a new process. This is possibly a factor in the following reasons.

2) Bin splits are bad (possibly due to new process). The chip comes out functional, but the clocks for the majority are too low for the high end.
Perhaps the delay for a new high-end respin would make sense for a family launch. The stockpile of low-clocked R600s could make up a good portion of the lower performance parts.

3) Clocks of the design are lower than desired, which leads to:

3a) The chip is too hot because it is being pushed to the upper edge of its clock envelope.

3b) It is clock-limited by something (if not by heat from 3a).

3c) It needs to be clocked higher.

Reason 3 is consistent with some rumors.
The rumored cooling solutions and the need for more power seems to show silicon at the edge of acceptable power consumption. This may be in part due to the adjustment to a new process, or due to clock and voltage bumps made for the sake of hitting higher clocks.

There isn't much hard evidence R600 isn going scalar like G80. This may indicate that utilization for the vector units is low enough to impact any peak numbers it can reach, so the clock must rise to compensate.

There hasn't been much substantive talk of clock domains. This means ATI is likely trying to push the higher levels scheduling logic along with the ALUs. Even should the ALUs be capable of being pushed faster, the scheduling logic might not be. It should be noted that Nvidia doesn't try to push the full core to high clocks, and keeps things in the slow domain several hundred MHz less than ATI's entire-chip clock.

There are other possible problems. Perhaps the way R600 handles branches is less efficient than G80, which devotes units to keep them from tying up the ALUs. Perhaps the data paths are more complex for R600, which would increase heat and suppress clock rates.
Perhaps the wide memory bus leads to clock issues, though one would think ATI could muster a workable ring bus.
Tweaks to the memory controllers and ring bus design would need a respin.

Maybe a utilization problem is worsened by difficulties with driver development, though it wouldn't necessarily need a hardware respin (if that is what is happening). The time frame of the delay seems to be long enough to allow one more go at the fab.

Wild cards:
Let's just think of what else could go wrong, since we're getting nothing else worth talking about.

 

[Geo]

At this rate, we're going to need a LAST LAST R600 Rumours & Speculation Thread.

Too late! I said that two pages ago. :smile:

Great. My upgrade plans have gone straight out the window. Thanks for suggesting that I wait for R600 to make my choice, geo.

Hah! I bought an 8800GTX the week after it was released, and everyone knew it.

 

[Tim Murray]

Right. And so far, it seems that many companies are struggling to get their products to function correctly/ fully supported state. While three months (let's say 4 for actual gold/finalized status) is an eternity in the technology business, simply not getting all of the correct information hampers even the mightiest of humans .

Is it really such a far-fetched notion?

Yes.

 

[flippin_waffles]

I have an opinion on this situation. In the back of my mind, I keep getting nudged by the possibility that perhaps AMD is underestimating the importance of the enthusiast crowd. I think ATI fully understood what the impact of such a crowd can be, and perhaps AMD sees things from a different point of view, as it has focused on the server market more than anything else in the last 3 years, and rightly so. The opteron line of processors took that market by storm. However, when it comes to graphics, the industry is almost entirely driven by enthusiasts and performance mongers. If this is true, I think they need to reassess how important the top end is to their graphics division and release this card, even if it is in limited quantities, or not a full line up.

If they've taken all that into consideration already, then the only thing left is subpar performance, which I won't let myself believe quite yet.

The only other possibility I can see, is that this is all FUD. But there are alot of repudable people claiming that it is NOT FUD, so who knows.

 

[3dilettante]

Too late! I said that two pages ago. :smile:

Drats, I was too late to seize the opportunity.
Looks like I sure AMD'ed myself.

 

[Aerows]

Okay, well I threw my speculation out there - a zero hour code change by MS screwed up software/driver development.

It could also be DRM related - I'll blame DRM anyway because it's the devil .

The first is less likely than the fact that it is a performance per (watt+$$+heat) issue, but I think that it is possible. I think that is more possible than the "delayed for AMD's CPU solutions".

Maybe testing revealed that the increase in clock speed caused early failure of the chips. No company would want to release a product only to have to RMA a large percentage of them a month or two down the line.

 

[Geo]

The only other possibility I can see, is that this is all FUD. But there are alot of repudable people claiming that it is NOT FUD, so who knows.

I would deeply hesitate to call xbit FUD-artists when they say they've got a direct quote from AMD. And that quote says what it says. Plus there are enough reports that they cancelled an already scheduled editors day to make that credible. What it means, of course, can be open to some interpretation, but it seems to me unless you've got rose colored glasses on none of those interpretations are positive ones, in the sense that, say, you were thinking yesterday that R600 was going to drive NVIDIA out of business in the first month after it's release, and now that you've heard this news you realize that it will actually drive NVIDIA out of business in the first week after its release instead.

Could it be "neutral" news, meaning nothing one way or another? I doubt it, really, not when you're cancelling an already scheduled editors' day. Occam's razor is pretty heavily in favor of bad news here, but that doesn't necessarily mean disastrously bad news. Unfortunately, much of the community has that ZOMG thing going generally, where it's a binary triumph/disaster reaction to everything. Real life is usually more complex than that.

 

[SugarCoat]

Okay, well I threw my speculation out there - a zero hour code change by MS screwed up software/driver development.

It could also be DRM related - I'll blame DRM anyway because it's the devil .

The first is less likely than the fact that it is a performance per (watt+$$+heat) issue, but I think that it is possible. I think that is more possible than the "delayed for AMD's CPU solutions".

Maybe testing revealed that the increase in clock speed caused early failure of the chips. No company would want to release a product only to have to RMA a large percentage of them a month or two down the line.

Still doesnt make any sense, there are plenty of non-vista/vista users out there, like myself, who would happily fork over the cash for the fastest DX9 card and would wait for DX10 drivers to come while at least having the hardware. Thats why this smells of hardware problems, at least to me. If it needs another respin i think that would be a good thing, but at this point i'm worried about performance in certain sections of gaming, much like how the FX series looked fine until it tried to do DX9 shader 2.0. If the problem is indeed as deep rooted as it was in the case of the NV3x chips then they'd still be better off blitzing the market with parts as fast as they could.

I think they just lost my sale, i'll think on it a day or two but a GTX OC looks mighty tempting right now. Which ever PR person from AMD who said its "a marketing strategy" is an idiot. Sorry. I'm an enthusiest, i want the best/fastest now, i dont want to wait for them to look good when i should/could have it now. Did they forget they do infact have competition or what? Adding fuel to the fire in my personal situation is this X800XT is painful and i'm graphics limited in everything since i upgraded my platform.

I wouldnt mind hearing a non candy coated reason for them doing this because the current one is worth about as much as white noise to me.

 

[Thorburn]

In the back of my mind, I keep getting nudged by the possibility that perhaps AMD is underestimating the importance of the enthusiast crowd.

I'd hope AMD more than almost anybody would understand the importance of this better than anyone, back in the K6 days and even early in the Athlon's life this always seemed to be the one market that really gave a damn.

 

[Corwin_B]

If it needs another respin i think that would be a good thing, but at this point i'm worried about performance in certain sections of gaming, much like how the FX series looked fine until it tried to do DX9 shader 2.0.

Didn't the R300 inflict a sad beating on the FX in about every situation including DX7/8 (apart from some very specific OpenGL benches) at decent resolutions with AF/AA ? Anyway, they could be waiting for some DX10-specific benches to show them in the best light, but the 512-bits bus should be enough to win DX9 in high-res, high-AA/AF scenarios too ? And if R600 is ROP-limited, then why this huge bus ?

I think they just lost my sale, i'll think on it a day or two but a GTX OC looks mighty tempting right now.

I'm in the same boat, provided Nvidia releases good Vista drivers in the next weeks I'll make the jump. I've been a fidel ATI customer since the 9700 Pro (bought a 9800XT, a X800XL and a 1900XT since), but there's only so much time you can delay your products before your customers start looking at alternatives.

Which ever PR person from AMD who said its "a marketing strategy" is an idiot. Sorry. I'm an enthusiest, i want the best/fastest now, i dont want to wait for them to look good when i should/could have it now.

Well, sometimes the job of PR persons makes them look like idiots. I mean, they can't come and say "Well, we screwed up and it looks like we won't beat the competition, so it's back to the drawing board". So they come up with some good-sounding nonsense aimed primarily at investors.

 

[Aerows]

I'd hope AMD more than almost anybody would understand the importance of this better than anyone, back in the K6 days and even early in the Athlon's life this always seemed to be the one market that really gave a damn.

No doubt. I was one of those intrepid die-hards that survived the super socket 7/VIA chipset horrors in the K62 days.

 

[SugarCoat]

Didn't the R300 inflict a sad beating on the FX in about every situation including DX7/8 (apart from some very specific OpenGL benches) at decent resolutions with AF/AA ? Anyway, they could be waiting for some DX10-specific benches to show them in the best light, but the 512-bits bus should be enough to win DX9 in high-res, high-AA/AF scenarios too ? And if R600 is ROP-limited, then why this huge bus ?

Yes it wasnt the best, but i think many of us are expecting the R600 to be a feature laden card, much like the R520. Even if it lost by 5-10% i'd still take it now with more features and simply hope that driver maturity helps performance down the road. When people bring up the NV3x cards though that waves the flag of critical hardware failure which is in my opinion what made the FX line-up so infamous. Releasing a slower part later is one thing, releasing a slower part later that totally flops in what it should support (eg. FX cards and DX9) is something else entirely. No amount of respins or increase in clocks is going to fix that.

Well, sometimes the job of PR persons makes them look like idiots. I mean, they can't come and say "Well, we screwed up and it looks like we won't beat the competition, so it's back to the drawing board". So they come up with some good-sounding nonsense aimed primarily at investors.

I'm not expecting them to say that either, but i would prefer they launch what they have when they have it rather then letting themselves get walked on. Having some product is absolutly better then nothing at all. Surely there has to be some things the R600 does well in its current form, if it was that bad it wouldnt of taken 6 months after the competitions launch for them to realize it.

For me personally though it does get to the point where i dont even want to bother anymore. Cut my losses (losses being wasted time waiting) and run, because by the time they do actually release it we will infact be talking about nextgen nVidia parts, especially since they realize the value of a reliable launch schedule.

 

[Arty]

I'm dissappointed but not to the extreme extent as most of you guys here. :smile: I didnt buy availability in April for a second considering their schedule.

It doesnt smell like yeild, spin problems. Like Geo, I'd bet on a full family launch; R600, RV610, RV630, Mobility series, FireGL etc.

 

[lopri]

It could just be bad yields. (i.e. not too many chips hitting targeted clocks) Some folks say that "some products out on the market is better than no products", but I'd think otherwise especially when AMD itself talked a "Hard Launch". Sites like AnandTech as well as [H]ardOCP has been extremely vocal about availability of launched products, and they can and will make a product/company look bad for that reason regardless of actual potential of the product. (And they have a good reason to do so)

Until more concrete news comes, that's where I'm inclined at the moment: not good enough yields for a hard launch - therefore yet another delay.

 

[SugarCoat]

I'm dissappointed but not to the extreme extent as most of you guys here. :smile: I didnt buy availability in April for a second considering their schedule.

It doesnt smell like yeild, spin problems. Like Geo, I'd bet on a full family launch; R600, RV610, RV630, Mobility series, FireGL etc.

how many chips do you think they need to satisfy the market that goes after the $650 parts? If anything i think the flagship parts are the least stressful in terms of product output and demand.

 

[lopri]

how many chips do you think they need to satisfy the market that goes after the $650 parts? If anything i think the flagship parts are the least stressful in terms of product output and demand.

How many G80 chips do you think NV has shipped since the launch?

Edit: Well, maybe I misunderstood your comment. My point is that NV has sold lots of G80 boards already. (I've heard somewhere along the line of 400,000~600,000 units so far. By the time R600 ships, I bet NV will have marked 1M.)