ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-20 04:49:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix splitk preshuffle (#3137)

Browse Source 
* Fix splitK multiply_multiply_wp

* Add tests for gemm_multiply_multiply_wp

* Add tests for gemm_universal_preshuffle (KBatch = 1)

* Add tests gemm_blockscale_wp

* Fix splitk gemm universal preshuffle

* Run new tests on arch supporting fp8

* Restore example

* Fix strides profiler

* Fix tests

* Fix clang format

* Finalize profiler preshuffle with tolerances

* Minor improvements to splitk related changes

* Address review comments: clang format and ckProfiler typo

* Remove b_k_split_offset from SplitKBatchOffset struct

[ROCm/composable_kernel commit: 507d81c3af]
This commit is contained in:
Enrico Degregori
2025-11-03 20:59:01 +01:00
committed by GitHub
parent bf0dc8ce56
commit 9575bcd099
19 changed files with 777 additions and 172 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
profiler/src/profile_gemm_blockscale_wp.cpp
View File

@@ -126,19 +126,19 @@ int profile_gemm_blockscale_weighpreshuffle(int argc, char* argv[])
const int DefaultStrideB = ck::is_same_v<BLayout, Row> ? N : K;
const int DefaultStrideE = ck::is_same_v<ELayout, Row> ? N : M;
bool pass = ck::profiler::profile_gemm_blockscale_weighpreshuffle_impl<A0DataType,
A1DataType,
B0DataType,
B1DataType,
ComputeDataType,
AccDataType,
EDataType,
scale_block_m,
scale_block_n,
scale_block_k,
ALayout,
BLayout,
ELayout>(
bool pass = ck::profiler::profile_gemm_blockscale_weightpreshuffle_impl<A0DataType,
A1DataType,
B0DataType,
B1DataType,
ComputeDataType,
AccDataType,
EDataType,
scale_block_m,
scale_block_n,
scale_block_k,
ALayout,
BLayout,
ELayout>(
do_verification,
init_method,
do_log,