DX9 Delayed until Q1 '03?

[Joe DeFuria]

Hard to tell how reputable (or who exactly) the source is, but anyway:

http://www.nvplanet.com/modules.php?name=News&file=article&sid=747

UPDATEue to unexpected problems with the DirectX 9 Technology we have decided to release DirectX 9 some time within the first quarter of 2003. .

If true, It sounds to me like Microsoft and nVidia are ironing out issues between DX9 and NV30. I have a few reasons for this theory:

1) nVidia has stated in the past that "DX9 and NV30 are 'aligned'".
2) MS seems to want the DX9 release to be the "last" DX release until Longhorn.
3) Delays with NV30 and Delays with DX9 just appear to be more than purely coincidental, IMO.

I'm NOT saying or implying that NV30 is somehow "broken", won't be DX9 compliant, etc.

But I have to believe that since now there should be "close to final" if not final silicon for NV30 floating around for testing, it's only at this time can any 'quirks' or unexpected behaviour between the NV30 and DX9 interface be definitively worked out. (I'm sure there were also quirks with DX9 and R-300, which shuold have obviously been worked out by now, however.) MS probably wants to release a DX9 run-time that has no known issues with R-300 and NV30 architecures, since they will undoubtedly be the two most popular DX9 products. Can't really fault them for that strategy, since it was probably originally anticipated that both architectures would ship at a similar time frame.

However, the way things worked out, it's actually quite unfortunate for ATI, IMO. ATI could have really benefitted if MS kept to a DX9.0 release at the time of the R-300 launch, and thought about a possible DX 9.0a patch, if needed, whenever NV30 actually ships. I'm sure nvidia wouldn't like that all too much though...imagine a 3D Mark DX9 having been relased already....the GeForce4 line would look even worse in comparison to the R-300 series.

My suspicions will be more or less confirmed if we start to see NV30 p/reviews almost immediately after DX9 goes public. Because I'm guessing that one of the things nVidia will do, is supply some DX9 benchmark to reviewers, since certain DX9 shader performance is probably the biggest strength of NV30 relative to R-300.

 

[RussSchultz]

1st quarter '03 is about 2 weeks away.

I know I'm not productive until Jan2.

 

[Joe DeFuria]

Jan 2? My...you get back on track pretty quickly.

 

[RussSchultz]

I pre-load all my ineffeciencies. I've been worthless since about Thanksgiving. Or my wife has been telling me that since at least then...

 

[Babel-17]

I've noticed this coincidence also. In that same vein of reasoning I think the latest patch for UT2k3 (and the bonus pack) is being delayed from going final for similar reasons. Benchmark runs for that game are starting to become part of the standard suite reviewers run.

 

[John Reynolds]

I pre-load all my ineffeciencies. I've been worthless since about Thanksgiving. Or my wife has been telling me that since at least then...

Your wife actually limits the timeframe during which she tells you that? Wow.

 

[Psikotiko]

I think this is very good for Nvidia.
Remenber when they said that GFFX would come out at the same time that directx9. They were right .

Anyone knwos what were the problems?

 

[jjayb]

Speaking of the ut2k3 patch. Anyone remember the whole "going to release a patch to allow ultra high details when the nv30 comes out" thing? Didn't Epic say they weren't releasing the patch even though the 9700 is here now because it takes a card with 256mb of memory? Is the gforcefx even going to have 256mb of memory? I've only seen one place state the amount of memory it would have on board and they said it was 128mb. Sorry, don't remember what site that was. If it does have 128 mb of memory then someones got some splainin' to do.

 

[antlers]

I remember an nVidia statement (in one of their conference calls) that the NV30 would be out in November, and would be tied to DX9. If it was delayed beyond November, it would be because of problems with DX9 rather than problems with the NV30

I guess this means they were telling the truth...

 

[Doomtrooper]

I think this is very good for Nvidia.

There more to PC graphics than Nvidia, I think it sucks a$$ for all the 9700 owners. I've stated all along that DX9 would release when Nv30 releases..Nvidia and Microsoft are making it easy to see the 'partnership' they have formed with the X-box deals (including Xbox 2).
You can count on 'Futuremark' ready for the the Nv30 release too

Anyone knwos what were the problems?

Ya...

M$ -->

<--Nvidia

 

[Tahir2]

Its better to have on release that has no further updates and causes headaches for developers than having to patch it when they figure out something that could have been fixed in the first release was not just to get it out.

Patches = Evil

 

[ben6]

Hrm I had a conference call with Nvidia on Cg . They said that DX9 was going to be released yesterday . LOL . On a lighter note, Philip Taylor said yesterday in response to The Inquirer story on DX9 :

Q: Where did The Inquirer get the idea that DX9 would be released on the 19th?

A I-Ching?

Philip Taylor

 

[Joe DeFuria]

Its better to have on release that has no further updates and causes headaches for developers than having to patch it when they figure out something that could have been fixed in the first release was not just to get it out.

I agree....to an extent. Though in this case, I don't think it is worth it when one DX9 product will have been on the market for probably 5-6 months before the DX9 release.

Why didn't MS hold off on Releasing DX8, for the arrival of the Radeon 8500 after the GeForce3? Did "patching" to DX 8.1 really cause developers headaches?

In this specific case, it would have been better for MS to ship DX9 in line with the R-300 release, allowing developers and consumers to play with actual hardware and officially supported drivers, and then deal with any potential patches after NV30 is released.

Again, I can't fault MS too much because I'm sure the original timing was for both NV30 and R-300 to be released at a similar time frame. That doesn't mean ATI didn't end up getting screwed out of the deal though.

 

[Joe DeFuria]

Ben....

I liked your new "prediction" for DX9 release over at NVNews:

I think it'll be the end of the year or next year before we'll see it.

I bet you'll turn out to be right!

 

[ben6]

A buddy of mine works for Microsoft, though not on DirectX 9.0 . He said he hadn't received the internal email saying it would be released (normally they send one out a week or so prior (he got one on DX8.1b and 8.1) , so it's not going to happen.

My earlier "prediction" was based on a conversation I had with Nvidia . Shrug , they were wrong.

 

[MDolenc]

1. DirectX 9.0 will ship when it's ready. All those "official" dates that ended up on the net are just speculation from (high ranking) people that don't work at MS.
2. The only problem that R300 or NV30 could have with DX9 is bad DX9 driver. And as far as I know DX9 driver from ATI is far from perfect, while NVidia doesn't have DX9 driver at all.
3. If there are critical problems in DX9 they are not related to Direct3D, so all your ideas about "MS waiting for NV30 to show up" go right through the window.

Of course you can always say that Longhorn will be released next week and after a week, when Longhon still isn't released you say "Longhorn delayed due to unkown issues". Sounds bizzare? That's just the kind of bullshit that's happening with those "official DX9 release dates".

 

[ben6]

Yeah it is funny the conspiracy theories that brew around the net. I'm wondering what excuse The Inquirer will have this time?

 

[KimB]

But I have to believe that since now there should be "close to final" if not final silicon for NV30 floating around for testing, it's only at this time can any 'quirks' or unexpected behaviour between the NV30 and DX9 interface be definitively worked out.

I think that's kind of silly. Hardware-level quirks should be worked out entirely within the drivers. The API should be a level above that, independent of the hardware.

The most obvious reason that I can see is that Microsoft is simply having a hard time getting DX9 finished (read: bugs/performance problems in current builds...).

Of course, nVidia and Microsoft having an agreement is possible, but it does sound rather odd. I don't really see what Microsoft has to gain by holding off on DX9. Microsoft is not known to be terribly loyal...

 

[antlers]

Of course, nVidia and Microsoft having an agreement is possible, but it does sound rather odd. I don't really see what Microsoft has to gain by holding off on DX9. Microsoft is not known to be terribly loyal...

On the other hand, the statement from the Nvidia guy earlier did make it sound like there was an agreement; perhaps DX9 contains some technology licensed from Nvidia and this was a secret part of the licensing agreement.

 

[Doomtrooper]

These conspiracy theories never happen in corporate business do they ....if people think that way....well too put it bluntly, it happens every day...that is reality.
Corporations make all kinds of alliances, and there is lots of good reasons for M$ to wait for its favorite graphic IHV..X-box is one of them..possible further cut backs on Xbox chips..supposedly CG and Dx9 HLSL is very close.....who knows.
The Beta's have been out forever and some of these 'so called release rumors' came from people in the know (and I'm not talking the Inquirer)...

Its very ironic Nv30 is late and now is DX9...we shall see who's more accurate when the Nv30 boards are released to reviewers...again a pity for the current DX9 card owners.