DX 9.1 True?

[Unit01]

I just browsed the nvnews forum and almost every post in this thread http://www.nvnews.net/vbulletin/showthread.php?s=&threadid=19613

Keeps saying "when DX 9.1 arrive etc etc"
And from what i remember from this board DX 9.1 was just some empty air.

So i ask now, has DX 9.1 become reality or is it still some unconfirmed rumour?

 

[elroy]

Read the last page of that thread - it pretty much explains it all.

 

[Dave Baumann]

FYI - If you flick through the Microsoft press DirectX9 programming book you'll see that PS/VS3.0 currently only has software render targets for it supplies in the API - DX9 runtime and SDK will require updates to provide HAL layers for the PS/VS3.0 models. Reading the GDC mailer adverts that I'm getting, MS are also talking about "New shader model support for HLSL". Its possible that when MS does the SDK and runtime updates this will be labelled DX9.1.

 

[Demirug]

FYI - If you flick through the Microsoft press DirectX9 programming book you'll see that PS/VS3.0 currently only has software render targets for it supplies in the API - DX9 runtime and SDK will require updates to provide HAL layers for the PS/VS3.0 models. Reading the GDC mailer adverts that I'm getting, MS are also talking about "New shader model support for HLSL". Its possible that when MS does the SDK and runtime updates this will be labelled DX9.1.

You do not need a runtime update for PS/VS 3.0. The DDK and the HAL contains anything you need to support PS/VS 3.0. Even the SDK is able to support PS und VS 3.0. The only thing that is not in the SDK is a HLSL compiler with PS and VS 3.0 targets. But this only need a SDK Update.

 

[T2k]

Don't be surprised on urban legends and misunderstandings like this: as usually even Anand doesn't have a flying frog what he's talking about - as usually...

I've read the latest rumors on FP32 support in R420, but I don't see how that's any major news. As far as I know, the DirectX 9.1 spec (with pixel shader 3.0 support) calls for full precision to be 32-bit floating point, so if ATI wants to be DX9.1 compliant they will have fp32 support. [/i]

http://www.anandtech.com/weblog/index.html - roll down to Jan 16.

EDIT: misformatted

 

[THe_KELRaTH]

I understood that another DX update is required to expose ATI's Centroid AA.

 

[991060]

I asked the same problem in the DirectX developer mailing list, folks there said only a SDK update is required to support 3.0 shader mode in HAL.

 

[Dave Baumann]

There probably will be a runtime update to 9.0c sooner or later.

 

[digitalwanderer]

"AAAAAAIIIIEEEEEEEEEEEEEEEE!!!!!!!!"

"Thud-thud-thud-thud-thud-thud-thud!", sayeth the desk and the Dig's forehead in unison!

That is absolutely the most brilliant piece of FUD I have ever seen, it just keeps coming back!

 

[Ailuros]

It keeps coming back for a reason. Blame IHVs and how they plan to market future products.

 

[KimB]

"AAAAAAIIIIEEEEEEEEEEEEEEEE!!!!!!!!"

"Thud-thud-thud-thud-thud-thud-thud!", sayeth the desk and the Dig's forehead in unison!

That is absolutely the most brilliant piece of FUD I have ever seen, it just keeps coming back!

Untrue as it may be, I don't think FUD describes it

 

[akira888]

I understood that another DX update is required to expose ATI's Centroid AA.

No. Centroid texture sampling (which guarantees that the texture sample in multisampling is taken from a point in screen space which intersects the primitive IIRC) is exposed in PS3.0 which is of course already in DirectX9.0b.

Oh, here it is - http://msdn.microsoft.com/library/e...reference/Shaders/PS_InstructionModifiers.asp

 

[KimB]

I understood that another DX update is required to expose ATI's Centroid AA.

No. Centroid texture sampling (which guarantees that the texture sample in multisampling is taken from a point in screen space which intersects the primitive IIRC) is exposed in PS3.0 which is of course already in DirectX9.0b.

Which doesn't help if the R3xx doesn't support PS 3.0...

 

[Clootie]

Actually ATI already exposed centroid sampling via "hack"

 

[KimB]

Actually ATI already exposed centroid sampling via "hack"

According to OpenGL Guy that's not the case.

 

[Clootie]

According to OpenGL Guy that's not the case.

So, why it still works on Cat3.10 (have not tested 4.1 though)?

 

[Lezmaka]

Yep, for ATI R3xx chips in PS_2_0: 'dcl_pp t0' == 'dcl_centroid t0'

I'm afraid you are mistaken.

He never said centriod sampling wasn't exposed. He just said that was in correct. Doesn't mean it's not exposed in some way.

 

[THe_KELRaTH]

Yep, for ATI R3xx chips in PS_2_0: 'dcl_pp t0' == 'dcl_centroid t0'

I'm afraid you are mistaken.

He never said centriod sampling wasn't exposed. He just said that was in correct. Doesn't mean it's not exposed in some way.

I'm convinced that OpenGL Guy was a politition in a previous life

 

[guste]

They were talking about this very thing in MS' developers messageboard for DirectX, last week.

If you go to #22 on this page, you might be able to glean some helpful info.

Hopefully this will aid in lowering digitalwanderer's blood pressure.

 

[Mariner]

Interesting link - perhaps we should point it out to all of the web sites which are blathering on about "PS/VS 3.0 becoming available with DirectX 9.1"? 8)