Merge commit 'e135dd518d19a36466ce7c61bb9d3203ec18c8af' into develop

This commit is contained in:
assistant-librarian[bot]
2025-10-31 03:32:13 +00:00
parent c41df57bad
commit a0dd3fc932
13 changed files with 2953 additions and 6 deletions

View File

@@ -902,8 +902,8 @@ struct FlatmmKernel
{
const auto [iM, iN] =
TilePartitioner{kargs.M, kargs.N}.GetOutputTileIndex(partition_idx);
const index_t i_m = __builtin_amdgcn_readfirstlane(iM * TilePartitioner::MPerBlock);
const index_t i_n = __builtin_amdgcn_readfirstlane(iN * TilePartitioner::NPerBlock);
const index_t i_m = amd_wave_read_first_lane(iM * TilePartitioner::MPerBlock);
const index_t i_n = amd_wave_read_first_lane(iN * TilePartitioner::NPerBlock);
const SplitKBatchOffset splitk_batch_offset(kargs);
// options