On Thu, Oct 16, 2014 at 1:25 PM, Pohjolainen, Topi <[email protected]> wrote: > On Thu, Oct 16, 2014 at 09:50:44AM -0400, Ilia Mirkin wrote: >> On Thu, Oct 16, 2014 at 8:24 AM, Topi Pohjolainen >> <[email protected]> wrote: >> > Here is some basis for supporting double precision floats on i965 hw. >> > On IVB this gives (details below): >> > >> > piglit-run.py --include-tests "ARB_gpu_shader_fp64" tests/all.py /tmp/foo >> > [32/32] crash: 2, fail: 5, pass: 24, skip: 1 >> >> This should be more like 1K tests. Looks like somehow the generated >> tests aren't getting picked up? You might try with "-t fp64". >> >> $ ./piglit-run.py -t fp64 -d tests/all.py asdf >> [2179/2179] dry-run: 2179 >> >> But only like half of those are execution tests... I usually just run >> it with tests/gpu.py: >> >> $ ./piglit-run.py -t fp64 -d tests/gpu.py asdf >> [1102/1102] dry-run: 1102 > > Thanks a lot for the tip! That gives: > > pass: 172 > fail: 15 > crash: 909 > skip: 1 > timeout: 0 > warn: 0 > dmesg-warn: 0 > dmesg-fail: 0 > total: 1097 > > > which correlates well with the fact that there are a lot of things simply > still missing. I've been mostly playing with uniform support so far. The > reason I sent the work to the list is to check if people agree on the > basic design choices.
Sure. Just wanted to make sure you had the full set of piglits available to you :) BTW, if you've only done fs but not vs/gs, you can add things like -x vs -x gs to exclude a lot of the piglits that you _know_ will fail. There are some good tests in there... things like dmat4*dmat4 will give your RA a nice stress... _______________________________________________ mesa-dev mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/mesa-dev
