mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-19 12:30:16 +00:00
Merge commit 'f161b5b738781c71bd5f2c191561b81f679ba9ed' into develop
This commit is contained in:
@@ -598,8 +598,8 @@ struct FlatmmKernel
|
||||
CK_TILE_DEVICE void operator()(KernelArgs kargs) const
|
||||
{
|
||||
const auto [iM, iN] = TilePartitioner{kargs.M, kargs.N}.GetOutputTileIndex(blockIdx.x);
|
||||
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 index_t i_m = __builtin_amdgcn_readfirstlane(iM * TilePartitioner::MPerBlock);
|
||||
const index_t i_n = __builtin_amdgcn_readfirstlane(iN * TilePartitioner::NPerBlock);
|
||||
|
||||
const SplitKBatchOffset splitk_batch_offset(kargs);
|
||||
// options
|
||||
|
||||
Reference in New Issue
Block a user