The LAST R600 Rumours & Speculation Thread

[Kaotik]

Supposedly R600 has been pushed back to Q2 according to the same source that came with previous GPU news. This news was told to him today by contacts in Taiwan. He forgot to ask if this was for R600 in general or for the GDDR3 or GDDR4 version. A couple of weeks ago ATI told him that R600 would be launched in February.

So this kinda contradicts the newsreports of a January launch from that Level505 site that leaked 'R600 info'. Or AMD (or his Taiwan contacts) might be feeding him with misinformation this time around.

On a sidenote, RS690 and RS690c are sampling now, and MP will start in February, but due to some productionproblems (reason not stated) the real volume production will be in H2. This chip is faster than nV's C61.

If RS690 is only sampling now, it seems rather strange we get pics of Epox mobo with it?
http://www.beyond3d.com/forum/showthread.php?t=36818

 

[kemosabe]

Q2 would be a monumental bust....Vista will have come and gone and the G80 refresh will likely be shipping in volume. Gotta wonder how long Orton will last at AMD if R600 misses the proverbial DX10 inflection point.

 

[Jawed]

another 3 months +

Jawed

 

[Geo]

Supposedly R600 has been pushed back to Q2 according to the same source that came with previous GPU news. This news was told to him today by contacts in Taiwan. He forgot to ask if this was for R600 in general or for the GDDR3 or GDDR4 version. A couple of weeks ago ATI told him that R600 would be launched in February.

So this kinda contradicts the newsreports of a January launch from that Level505 site that leaked 'R600 info'. Or AMD (or his Taiwan contacts) might be feeding him with misinformation this time around.

Well, CJ, you've certainly earned some credibility points in my book, so I'm not about to dismiss this out of hand. But the timing feels wrong to me to make a decision that far out. Assuming they made it last week, and we're aiming *at best* at early April (which is the earliest Q2 could be), then we're talking about 3.5 months from decision to presumed launch date (again, at best), and that doesn't seem to follow typical patterns as we know them. If they were pushing to March that would fit the pattern, but "April at best" doesn't feel right for a decision made in late December. Not this late in the process where they've already got multiple spins behind them anyway.

 

[Kanyamagufa]

Damn, and we thought they had left delays behind with R520...

It would be very unfortunate to see R600 suffer the same fate.

 

[INKster]

Well, CJ, you've certainly earned some credibility points in my book, so I'm not about to dismiss this out of hand. But the timing feels wrong to me to make a decision that far out. Assuming they made it last week, and we're aiming *at best* at early April (which is the earliest Q2 could be), then we're talking about 3.5 months from decision to presumed launch date (again, at best), and that doesn't seem to follow typical patterns as we know them. If they were pushing to March that would fit the pattern, but "April at best" doesn't feel right for a decision made in late December. Not this late in the process where they've already got multiple spins behind them anyway.

The hardware may be more than ready, but who knows about the driver development effort ?
We shouldn't underestimate the task of such an extensive reworking of the software "plumbing" required for WVDDM + Direct3D 10.

 

[Ailuros]

The hardware may be more than ready, but who knows about the driver development effort ?
We shouldn't underestimate the task of such an extensive reworking of the software "plumbing" required for WVDDM + Direct3D 10.

I don't see why this could be in any way related to the hardware release; especially since there's still no G80 Vista driver officially available even today.

 

[Unknown Soldier]

The hardware may be more than ready, but who knows about the driver development effort ?
We shouldn't underestimate the task of such an extensive reworking of the software "plumbing" required for WVDDM + Direct3D 10.

Oh the stories that could be told.

US

 

[nutball]

I don't see why this could be in any way related to the hardware release; especially since there's still no G80 Vista driver officially available even today.

Well other than G80 has been out for a while now, whereas Vista hasn't. A few people are pressuring NV for Vista G80 drivers, but the real world isn't. Yet another advantage of getting your hardware out on time^H^H^H^H^H^H early. To launch a next-gen graphics architecture before Vista without Vista drivers may be regarded a misfortune, to launch a next-gen graphics architecture after Vista without Vista drivers looks like carelessness.

Having said that though, a hardware-ready-but-drivers-not release would surely be some sort of semi-paper-wink-nudge release. It wouldn't necessarily change the headline date, and in practice most people would be entirely unaffected. It would cause some ructions around here and other places though

 

[INKster]

I don't see why this could be in any way related to the hardware release; especially since there's still no G80 Vista driver officially available even today.

Exactly.
But there's one significant difference:
There is no Windows Vista yet, at least not for end users.
So, writing a standard DirectX9 driver for Windows XP is a lot easier, which is exactly what Nvidia did.
They too are having difficulties with Direct3D 10 drivers, or they would be out by now.

Since there's no OS supporting it (in practice), both IHV's must feel little appeal towards releasing potentially bug-laden software right now.
First impressions do count in this line of business.

 

[Kaotik]

Exactly.
But there's one significant difference:
There is no Windows Vista yet, at least not for end users.
So, writing a standard DirectX9 driver for Windows XP is a lot easier, which is exactly what Nvidia did.
They too are having difficulties with Direct3D 10 drivers, or they would be out by now.

Since there's no OS supporting it (in practice), both IHV's must feel little appeal towards releasing potentially bug-laden software right now.
First impressions do count in this line of business.

If 'end user' is defined as the average joes buying stuff from the local walmart, true, but I count many of the MSDN subscribers (especially those with access to MSDN via work/university/etc) as end users, too, and Vista has been available for them for quite a some time already.

 

[Geo]

And are you presuming then that R600 sucks in DX9? R300 was released without a DX9 driver. G80 was released without a DX10 driver. Maybe if we hear something about waiting for a new rev of DX10 from MS then I might buy that explanation. It seems to me that ATI has generally been suggesting that they are considerably ahead of NV in preparing Vista drivers, so that'd be a bit of a puzzler for a 3.5 month push.

 

[INKster]

If 'end user' is defined as the average joes buying stuff from the local walmart, true, but I count many of the MSDN subscribers (especially those with access to MSDN via work/university/etc) as end users, too, and Vista has been available for them for quite a some time already.

How many of these buy high-end graphics cards ?
And, of those with the hardware, how many are Direct3D 10-ready ?
It's still a very small minority right now, no matter how you look at it.

 

[Ailuros]

Exactly.
But there's one significant difference:
There is no Windows Vista yet, at least not for end users.
So, writing a standard DirectX9 driver for Windows XP is a lot easier, which is exactly what Nvidia did.
They too are having difficulties with Direct3D 10 drivers, or they would be out by now.

Since there's no OS supporting it (in practice), both IHV's must feel little appeal towards releasing potentially bug-laden software right now.
First impressions do count in this line of business.

I still don't see that as a significant reason to hold back any hardware, as long as it's usable with current OSs.

There must be something I am missing; or my memory is obviously stuck selectively around the R300 launch. Pardon me but I have a tad of a hard time believing anything close to that. If there is a certain amount of delay or further delays down the line, then it suggests in my book hardware and not software problems.

Unless IHVs need Einstein clones to write a darned Vista driver these days....

 

[Kaotik]

And are you presuming then that R600 sucks in DX9? R300 was released without a DX9 driver. G80 was released without a DX10 driver. Maybe if we hear something about waiting for a new rev of DX10 from MS then I might buy that explanation. It seems to me that ATI has generally been suggesting that they are considerably ahead of NV in preparing Vista drivers, so that'd be a bit of a puzzler for a 3.5 month push.

Considering Vista has been now final for what, 2 months or so, I doubt anyone is waiting for new revision of DX10

 

[Geo]

Considering Vista has been now final for what, 2 months or so, I doubt anyone is waiting for new revision of DX10

Yeah? Then why did MS split the release between business and consumer by two (edited) months, if it wasn't to give them more time to polish consumer features (like, potentially, DX10)?

 

[Kaotik]

Yeah? Then why did MS split the release between business and consumer by three months, if it wasn't to give them more time to polish consumer features (like, potentially, DX10)?

The differences in Business and Consumer versions are in the media department (media center and such, not directx), also, they're both the very same build (artificially lifted from 59xx to 6000), and last but not least, all Vista editions are shipped (at least as far as I know) on the same DVD, and it installs the correct version based on your key.

 

[Geo]

But the users are different. How many business license users would have DX10-class graphics cards? If the versions are the same, that's great. Doesn't necessarily mean MS didn't build in the opportunity to make them different if they wanted to, given when that decision was made to make the release dates different. Also, how many people here loaded DX9.0C off their install CDs? There are other ways for MS to deliver an updated feature for new users to get it quickly after installing the base OS off their new CD.

But we're going tangential, given I don't think DX10 is the reason for pushing R600 to April anyway (nor am I convinced they have, as noted upstream).

 

[Kaotik]

Yes, this is getting offtopic, but just to get the last word razz, everyone who bought their copy sometime after SP2 was released have DX9.0c on their install disc (and those who slipstreamed their original discs with SP2)
The consumer and business versions either way are identical in terms of build, and thus, also with directx version, and any possible updates would be handled via Windows Update. The reasonings behind different release dates, well, I can only guess, but I'd be betting on giving the business versions via MSDN & Connect would be just to get it out and available asap, while the consumer versions need some time to be produced and shipped for OEM's to be customized with their own logos etc.

 

[silent_guy]

If there is a certain amount of delay or further delays down the line, then it suggests in my book hardware and not software problems.

I agree. Unless your project managment process is completely out of whack, I don't see how a SW project can slip multiple months just 1 or 2 month before the earlier reported target. It's much more understandable for hardware. Say they discovered some really critical bug somewhere in the design that needs a base spin: fix layout, fab, qualification, production... All these add up quickly. (Though 3.5 months is maybe a bit much indeed.)