On Monday 06 June 2011 17:55:43 Stefan Dösinger wrote:
Feedback is welcome. I'm also interested in test results, e.g. if the changes break a game, or the performance impact. If those patches cause a 5% performance increase I am happy.
Here are some results from my Radeon test machine: http://bit.ly/mi0i5J . It's about a 1% increase in the real Windows apps, much less than I hoped for. I'll test on Nvidia tomorrow, and manually test a few more (newer) apps for which I don't have PTS tests yet.
For the curious, these are the URLs to my nightly trackers: Radeon 5750: http://openbenchmarking.org/result/1106056-GR-WINE1321145&track_view=20 Geforce 7600: http://openbenchmarking.org/result/1106064-GR-WINE1321175&track_view=20
Watch out for the result ordering. Basically the results are ordered from newest to oldest(left to right), but the Radeon test is a bit more confusing. Openbenchmarking.org orders it in the order I uploaded the results. The Nvidia box boots automatically, the Radeon box is booted manually. So I forgot about the radeon box a few days and tested the older Wine versions a few days later. This is something Michael Larabel and I still have to work on.
Also the uptick in the Radeon tests is the test I ran with my performance patches applied. I forgot to change the result identifier, so OB.org picks it up like a normal nightly test.
Note that the Unigine Heaven tests are using the opengl backend right now. The other tests are D3D-based.
The source code for the d3d_clear, d3d_streamsrc and d3d_drawprim tests is available at https://84.112.174.163/~git/perftest (This is a git repo). I hope the server is stable finally.
Stefan