System Environment: -------------------------- --Platform: 945gm --Architecture: 32bit --2D driver: git commit:ea7c7d088e0a3c3b7ab2bdd0475aa13379764bf4 --3D driver: Mesa master commit:0da03c0aa9ed120fe5ff781a365b503b8ed70e0c --DRM: git commit:fd595fa4dc6f788a8a1e1b56178e15f411706cb9 --Xserver: git commit:4632ea22580c31d44b0786321668d9e78f02900e --Kernel: 2.6.23 using the latest git code on my 945GM machine, I seperately tested the Xtest under NoAccel(adding Option "NoAccel" "true" in Xorg.conf),VESA,EXA environments and did a compare to show the differentiations.
Created attachment 14775 [details] Xtest in NoAccel-VESA-EXA mode report log
Could you describe your testing method and environ? Those x11 core protocol function tests shouldn't have difference in either case.
(In reply to comment #2) > Could you describe your testing method and environ? > > Those x11 core protocol function tests shouldn't have difference in either > case. > NoAccel test: adding Option "NoAccel" "true" in xorg.conf VESA test: using Driver "vesa" in Device Section of xorg.conf EXA test: using Driver "intel" in Device Section of xorg.conf running Xtest under xts5 directory and run "tcc -e" to get the journal.
Carl, would you please take this one? thanks.
I'll be glad to take a look. I'm quite surprised like Zhenyu that there is any difference here. One of the huge shortcoming of the xtest test suite is that it doesn't test anything of the things that really matter for 2D performance today, (such as Render extension requests). So basically everything in xtest should hit identical code paths in the X server whether using NoAccel or EXA. But at some point soon I'll run the test suite on a 945 machine and see if I can replicate these results at all. -Carl
Close as wontfix, since EXA has been removed from master code, and also none has interest in xtest now.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.