Merge commit 'ae9f29b7d514b0829256a0a3ca9ab4511e7a1e04' into develop

This commit is contained in:
assistant-librarian[bot]
2025-10-07 19:11:30 +00:00
parent 98c3368cb6
commit d12bd51265

View File

@@ -1134,6 +1134,7 @@ struct UniversalGemmKernel
while(block_id < num_work)
{
s_waitcnt_barrier();
// Get the tile index for this block
const auto tile_idx = amd_wave_read_first_lane(block_id % num_tiles);
const auto [iM, iN] = TilePartitioner{kargs.M, kargs.N}.GetOutputTileIndex(tile_idx);