Skip to content

intel/compiler: Allow simd32 in some more cases with no8 and/or no16

Jordan Justen requested to merge jljusten/mesa:cs-override-simd-size into master

If no16 was specified, and the shader can't run in simd8 due to the local_size, then we need to generate a simd32 program.

One example I encountered was piglit's generated_tests/spec/glsl-4.30/execution/built-in-functions/cs-op-selection-bool-bvec4-bvec4.shader_test.

If both no8 and no16 are specified, then we need to generate a simd32 program.

I found that no8,no16 was also preventing an FS program from being generated, so I added a 2nd patch to this MR.

Cc: @cmarcelo

Ref: !4098 (closed)

Edited by Jordan Justen

Merge request reports