ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-07-01 12:17:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

best perf pw gemm config 0.11ms

Browse Source
This commit is contained in:
huizzhan
2025-08-26 09:55:50 +00:00
parent d43228fbca
commit 6ab1de129f
2 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
example/ck_tile/03_gemm/gemm_utils.hpp Normal file → Executable file
View File

@@ -280,8 +280,8 @@ struct GemmConfigPreshuffleDecode : public GemmConfigBase
template <typename PrecType>
struct GemmConfigPreshufflePrefill : public GemmConfigBase
{
static constexpr ck_tile::index_t M_Tile = 128;
static constexpr ck_tile::index_t N_Tile = 128;
static constexpr ck_tile::index_t M_Tile = 64;
static constexpr ck_tile::index_t N_Tile = 256;
static constexpr ck_tile::index_t K_Tile = 128 / sizeof(PrecType);
static constexpr ck_tile::index_t M_Warp = 1;

2
example/ck_tile/03_gemm/gemm_weight_preshuffle.cpp Normal file → Executable file
View File

@@ -277,7 +277,7 @@ int main(int argc, char* argv[])
try
{
return !run_gemm_example<GemmConfigPreshuffleDecode>(arg_parser);
return !run_gemm_example<GemmConfigPreshufflePrefill>(arg_parser);
}
catch(const std::runtime_error& e)
{