On Tue, Jan 13, 2004 at 10:35:01PM +0100, manu wrote: > Hi all, >I just made a x11perfcomp of 4.4rc2 (actually CVS of 2004/01/09 IIRC). >Here are the results (I also put the logs). But it seems that 4.4rc2 is >not really better (performances speaking) than 4.3.0.
That is what I would expect for most cases: 2D performance should be about the same for most drivers.
>Test has been done on a Compaq Presario 713EA (laptop), gfx is >TwisterK (proSavage). No WM, crude X server and x11perf, and hopefully >nothing else to disturb the tests (no cron, network...). >Bye >Manu > >PS : I plan to do the same for a radeon 9200, please tell me if you >really need all tests, I'd prefer only to do the necessary ones ;-)
Running xtest would be useful, both as a correctness test and a robustness test. I have already found and fixed several problems with the i810 support by doing this, and I'm still looking into some possible issues with a couple of other drivers. It is a good idea to do test runs at different depths and root window sizes. Also, a test run with a software-only driver (like vesa, or against Xvfb, or with the "NoAccel" option enabled) can provide a good reference run to compare against.
The xtest source can be checked out from the XFree86 CVS repository (the module is "test"), or a source tarball can be downloaded from <ftp://ftp.xfree86.org/pub/XFree86/xtest/>. The easiest way to build and run it is to use the build.sh and run.sh scripts in the test/xsuite directory.
OK it makes sense, I will do that.
People who have hardware that supports DRI could also look at GL performance and correct testing. Maybe someone else can post links to tools that are useful for that.
glxgears ;-) Bye Manu _______________________________________________ XFree86 mailing list XFree86@xxxxxxxxxxx http://XFree86.Org/mailman/listinfo/xfree86
