On Wednesday, March 13, 2019 4:25:24 PM PDT Jason Ekstrand wrote: > We don't set it on HSW and earlier in i965 and disabling it appears to > make derivatives somewhat more reliable. > > Cc: Kenneth Graunke <kenn...@whitecape.org> > --- > src/intel/vulkan/genX_pipeline.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/src/intel/vulkan/genX_pipeline.c > b/src/intel/vulkan/genX_pipeline.c > index 975052deb79..3767e5033c5 100644 > --- a/src/intel/vulkan/genX_pipeline.c > +++ b/src/intel/vulkan/genX_pipeline.c > @@ -1761,7 +1761,7 @@ emit_3dstate_ps(struct anv_pipeline *pipeline, > brw_wm_prog_data_prog_offset(wm_prog_data, > ps, 2); > > ps.SingleProgramFlow = false; > - ps.VectorMaskEnable = true; > + ps.VectorMaskEnable = GEN_GEN >= 8; > /* WA_1606682166 */ > ps.SamplerCount = GEN_GEN == 11 ? 0 : > get_sampler_count(fs_bin); > /* Gen 11 workarounds table #2056 WABTPPrefetchDisable */ >
I have no idea why it would behave any differently, but something definitely changed with Broadwell. If we don't set it on Gen8+, derivatives go horribly wrong - basic Piglit tests fail. It seems like setting it should be the right thing. But I never observed it to matter at all on pre-Gen8. Not sure why. Acked-by: Kenneth Graunke <kenn...@whitecape.org>
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
