ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-06-07 00:04:37 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update chunk size

Browse Source
This commit is contained in:
Rostyslav Geyyer
2025-03-19 20:37:05 +00:00
parent 221862c912
commit 5adf19ccb3
2 changed files with 10 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
test/mx_mfma_op/mx_mfma_op.hpp
View File

@@ -397,7 +397,8 @@ __device__ BFragT load_B_col_major(BType const* input_ptr)
static constexpr int32_t WAVE_SIZE = 64;
// Here we want to load from cols of B in chunks of 16 elements each.
static constexpr uint32_t chunk_size = 16;
static constexpr uint32_t chunk_size =
16 / (ck::is_same_v<ck::remove_cvref_t<BType>, ck::f4x2_pk_t> ? 2 : 1);
// each chunk is separated by an offset
static constexpr uint32_t chunk_offset = chunk_size * WAVE_SIZE / BLOCK_N; // 32 or 64