projects / platform / upstream / mesa.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 5814436) | patch
anv: disable SIMD16 for RT shaders
authorLionel Landwerlin <lionel.g.landwerlin@intel.com>
Tue, 3 May 2022 19:12:57 +0000 (22:12 +0300)
committerMarge Bot <emma+marge@anholt.net>
Wed, 28 Sep 2022 05:38:37 +0000 (05:38 +0000)
commit23c7142cd670d9634d53e3a9c03fc42ce6b32c51
tree39c15c4dee6c332b2aacc12214f47f5291f8a982tree | snapshot
parent58144361592c4d6f002d4963950b21fbdac84ba2commit | diff
anv: disable SIMD16 for RT shaders

Since divergence is a lot more likely in RT than compute, it makes
sense to limit ourselves to SIMD8.

The trampoline shader defaults to SIMD16 since this one is uniform.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16970>
src/intel/compiler/brw_fs.cpp diff | blob | history
src/intel/compiler/brw_nir.c diff | blob | history
src/intel/vulkan/anv_pipeline.c diff | blob | history
Domain: Graphics System / GL;