Mac 727e7d1930 shader: lower wave64 subgroup broadcast workgroup-wide ...

OpGroupNonUniformShuffle with Subgroup scope only reaches 32 lanes on this
hardware, but GCN wave64 code expects all 64. A decoupled-lookback prefix scan
broadcasts its ticket across the whole wave, so the subgroup-scoped lowering
silently read the wrong lane. Lower the broadcast across the workgroup instead.

DS_BPERMUTE and DS_SWIZZLE keep the raw subgroup id: they are defined on
independent 32-lane halves and must not be widened.

2026-08-31 23:40:09 +05:30

..

spirv

shader: lower wave64 subgroup broadcast workgroup-wide

2026-08-31 23:40:09 +05:30