WINE 1.1.21 Starts On Shader Model 4 Support

[phoronix]

Phoronix: WINE 1.1.21 Starts On Shader Model 4 Support

The release two weeks ago of WINE 1.1.20 brought some cleaner Direct3D code, but there really was not much to get overly excited about. The release of WINE 1.1.21, however, is different...

http://www.phoronix.com/vr.php?view=NzI1Mg

[L33F3R]

decent. Unfortunately DX10 games would run admirably slow on even the best of hardware. SM4 is a big deal but for the long run they should be focusing on actually speeding up the d3d-ogl translations.

[Remco]

As far as I know, DX10 games are entirely based on SM4 shaders. And those run directly on the GPU. That means there is no D3D > OGL translation, and no performance-loss.

[bulletxt]

Wow, what an interesting news!!

[L33F3R]

As far as I know, DX10 games are entirely based on SM4 shaders. And those run directly on the GPU. That means there is no D3D > OGL translation, and no performance-loss.

they sure as hell use sm4 but entirely based on sm4?

[Remco]

they sure as hell use sm4 but entirely based on sm4?

Well, what else should they be doing? Everything is done with shaders these days.

[NeoBrain]

As far as I know, DX10 games are entirely based on SM4 shaders. And those run directly on the GPU. That means there is no D3D > OGL translation, and no performance-loss.

AFAIK they need to convert Direct3D shaders to OpenGL shaders which kind of gives the same bottleneck.
Anyways, things will be playable (after we approach feature completeness) with at the very least 50% of original performance since there isn't anything that fancy about DX10 apart from the switch to a programmable-only-environment.

[SavageX]

Well, what else should they be doing? Everything is done with shaders these days.

No. You need to upload and manage textures, you actually want *geometry* to be rendered, you need to track the state of your graphics pipeline...

There is *much* to do before even thinking of rendering stuff with shaders.

[remm]

I agree it does not make sense from a user perspective. Actually, the whole wine does not make sense, there are glitches all over the place, that seemingly never get fixed.

For DirectX, they should probably redo DX 9 (or 10) support on top of Gallium, rather than dealing with an intermediate layer on top of OpenGL, but looking at their resistance for integrating submitted code like the Pulseaudio support, this will probably never happen

[NeoBrain]

I agree it does not make sense from a user perspective. Actually, the whole wine does not make sense, there are glitches all over the place, that seemingly never get fixed.

For DirectX, they should probably redo DX 9 (or 10) support on top of Gallium, rather than dealing with an intermediate layer on top of OpenGL, but looking at their resistance for integrating submitted code like the Pulseaudio support, this will probably never happen

1) Most things actually work pretty fine for me, you're playing the wrong games On a more serious side though, I can play many games, for which I'd otherwise need to boot up Windows, with acceptable to good FPS within Wine, so it DOES make sense (and that is not even about at least 90% of the "normal" windows apps I can run with it).
2) There are several reasons NOT to use Gallium3D for that, see http://www.phoronix.com/forums/showp...61&postcount=7 and the corresponding thread
3) We did not want to include that support as it would've meant yet one MORE sound backend to maintain. It took several years for the ALSA backend to catch up the OSS one, and all the other backends were simply crap (not working, nowhere near feature completeness, ugly code, etc...)