Pravin Jagtap 5e007afa9d [AMDGPU] Handle hazard in v_scalef32_sr_fp4_* conversions (#118589) ...

Presently, compiler selectivelly adds nop when opsel != 0 i.e. only when
partially writing to high bytes.
Experiments in SWDEV-499733 and SWDEV-501347 suggest that we need nop
for above cases irrespective of opsel values.

Note: We might need to add few others into the same table.

2024-12-11 18:38:10 +05:30

115 KiB

Raw Blame History

View Raw