On Tue, 2001-11-27 at 16:43, Didier Moens wrote: > Running an A30p, 1.2GHz-PIIIM, i830MP, ATI Radeon Mobility LY, > resolution 1600x1200x16. > > After spending the last few days on linux-kernel sorting out the agpgart > support for the Intel 830MP (= i830 with external graphics : patch > posted for inclusion in next kernel release), I went through some > benchmarking. > > In order to support the 830MP/ATI-LY/1600x1200 combo, I downloaded, > compiled and installed the latest X CVS'es (stock RH 4.1.0-6 returns the > dreaded "panel size not detected" error). > > Considering CVS : as of today, DGA 2.0 seems to play nicely together > with VMWare in full-screen mode : very nice indeed. > > > Unfortunately, 3D performance on the ATI Radeon Mob. LY seems to be a > bit "sub-optimal". > > Running 'gears' in default window mode renders me some 480~500 FPS, > whereas my 'old' A21p (PIII-850,440BX,ATI Mobility 128) gives me ~700 FPS. > > "agpgart.o" and "radeon.o" (kernel 2.4.15/16-version) are both loaded. > > 'glxinfo' reports "direct rendering: yes", and xdpyinfo reports "GLX", > "DGA" and "DRI" as being present. > > No "(EE)" or "(!!)" in XFree86.0.log. > > If I rmmod either agpgart or radeon, performance in 'gears' drops to > ~250 FPS, so something certainly is accelerated, but the performance > doesn't seem up to par, being at 60% of an aged Mobility-128.
Set LIBGL_DEBUG=verbose when running a DRI client to get some debug output. > I noticed a radeon_drv.o in the CVS tree. Would this one replace the > radeon.o from the kernel distribution ? No, that's the 2D driver. You can build a radeon.o kernel module in xc/programs/Xserver/hw/xfree86/os-support/drm/kernel, I suggest trying that instead of the one that comes with the kernel. -- Earthling Michel D�nzer (MrCooper)/ Debian GNU/Linux (powerpc) developer XFree86 and DRI project member / CS student, Free Software enthusiast _______________________________________________ Xpert mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/xpert
