Skip to content

intel/compiler: Xe2+ SIMD width adjustments

Rohan Garg requested to merge shadeslayer/mesa:gargr/lnl/math_simd16_hf into main

What does this MR do and why?

Xe2 can do SIMD32 on UGM/TGM load/stores as well as SIMD16 on Math HF ops.

Edited by Rohan Garg

Merge request reports