Why The Radeon Gallium3D Performance Is Down
Written by Michael Larabel in Display Drivers on 23 December 2011. Page 3 of 4. 16 Comments

What's one of the other problems in Mesa 7.12-devel? Documented here on OpenBenchmarking.org, the other issue was tracked down to: 363ff844753c46ac9c13866627e096b091ea81f8.

winsys/radeon: move GEM domains out of the drivers into winsys

The drivers do not need to care about the domains. All they need to set are the bind and usage flags. This simplifies the winsys too.

This also fixes on r600g:
- fbo-depth-GL_DEPTH_COMPONENT32F-copypixels
- fbo-depth-GL_DEPTH_COMPONENT16-copypixels
- fbo-depth-GL_DEPTH_COMPONENT24-copypixels
- fbo-depth-GL_DEPTH_COMPONENT32-copypixels
- fbo-depth-GL_DEPTH24_STENCIL8-copypixels
I can't explain it.

This other commit to the Gallium3D winsys for the Radeon driver and also touches the R300 and R600 Gallium3D drivers directly magically fixes a few FBO regressions, which Marek Olšák says in the commit message he can't even explain why those regressions get fixed on R600g (note: not R300g). But this seems to cause a performance issue on at least R300 in certain cases.

After the Phoronix Test Suite reported these two commits, I have also manually verified they affected the performance in the particular configuration under test.

Related Articles
Featured Articles
Trending Linux News