Skip to content

Draft: radv: lower subgroups after first optimizations

Georg Lehmann requested to merge DadSchoorse/mesa:radv-cts-shuffle-xor-pattern into main

Without constant folding the cts tests for shuffleXor(a, const) don't actually trigger our optimizations. We need another optimization run between subgroup and explicit io lowering though to avoid LDS size regressions.

Merge request reports