On Thu, 2010-05-27 at 23:32 +0200, Jerome Glisse wrote: > I merged r600g, it's no where from being usefull but people wanted to > have it on master. I changed a little bit the backend generation for > the shader (i am still not happy the way the backend is but it's better > that what was before). > > To support instruction (like lit or other instruction which doesn't match > natively the hw) tweak r600_shader_alu_translate in r600_compiler_r600.c > Right now there is issue with trans operation & litteral. It leads to > bogus shader. In order to avoid suffering GPU reset over and over i > disabled rendering. I am using radeondb and fp shader sample to debug the > shader compiler : > > radeondb -c gallium.bof > radeondb -s gallium.json > > This will print the r600 assembly with some human readability (assuming > i am still classified as human). > > To get gears working RSQ,MAX,SLT,DP3 need to be fixed (i get gears > working by cheating on those opcode). > > I haven't applied the depth buffer patch as it segfault for me and > haven't time to debug this before next week. > > Happy hacking everyone. > > Cheers, > Jerome Glisse
I have fixed a bug with my depthbuffer patch that could cause a segfault: It assumed that there was always a depthbuffer available, resulting in a segfault if there wasn't. Here is my new depthbuffer patch: http://www.basnieuwenhuizen.nl/r600g-added-code-for-depthtesting-v3.patch - Bas Nieuwenhuizen _______________________________________________ mesa-dev mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/mesa-dev
