On Sun, 1 Feb 2004, Pieter Hulshoff wrote: > On Saturday 31 January 2004 22:22, Mark Vojkovich wrote: > > This is a perfectly respectable score for 32bpp. It's actually > > exemplary if you're doing it in software. That's about AGP 2x > > performance if you're doing it in hardware. > > *nod*, but that doesn't explain the problems I've been experiencing (taking > you back to the beginning of this thread): > > On my SuSe 8.2 system (radeon 9000), performance of xmame.x11, full screen DGA > is very good. If I switch to using a graphics enhancing function (in > software), the performance is incredibly bad (never above 20 fps, nomatter > which game). On my Morphix cd however, performance is still good when I > enhance the graphics. I couldn't find any major differences between the > config files, and the version of XFree86 is the same. I posted the scores > (which you say are not too bad), because someone asked me for them (hoping > they might show some insight into this problem). > > Anyone have an idea as to why performance under SuSE is so bad when using > software graphic enhancing functions while under Morphix it works fine? If I recall correctly, SuSE made and ill-advised change to some memory mapping code in their XFree86. I don't remember which SuSE versions had that problem, but it might be your version. That problem doesn't exist in official XFree86 releases. This might have been limited to Athlon processors though. Also, if you have an Athlon processors, Linux kernel has an MTRR bug in some older kernels that effects them. If I recall correctly, /proc/mtrr reads nonsense in that case. Mark. _______________________________________________ XFree86 mailing list XFree86@xxxxxxxxxxx http://XFree86.Org/mailman/listinfo/xfree86
