ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-14 18:17:44 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge commit 'b0ee317d83b77741022997265d4125697e7f7804' into develop

Browse Source
This commit is contained in:
assistant-librarian[bot]
2025-09-12 20:11:58 +00:00
parent facbc883fa
commit 302aa809ea
65 changed files with 2301 additions and 232 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
example/ck_tile/01_fmha/CMakeLists.txt
View File

@@ -26,7 +26,7 @@ endforeach()
# "fwd" is a must-have api for the fmha_fwd example, add it if not specified
if(NOT "fwd" IN_LIST FMHA_FWD_ENABLE_APIS)
list(APPEND FMHA_FWD_ENABLE_APIS "fwd")
list(PREPEND FMHA_FWD_ENABLE_APIS "fwd")
endif()
file(GLOB_RECURSE CODE_GEN_SCRIPTS CONFIGURE_DEPENDS
@@ -51,6 +51,15 @@ set(FMHA_BWD_CODE_GEN_COMMON_ARGS
# --filter fmha_bwd_dot...@fmha_bwd_convert...@fmha_bwd...
)
# Reduce building time by disabling instances that are not currently used in the gtests
# TODO: Consider to use a special receipt for testing only, or even two receipts: a small subset of
# instances for quick CI runs and a larger subset for scheduled runs (the tests skip tests when
# there is no corresponding instance for parameters).
if(BUILD_TESTING)
# Filters are in the order of FMHA_FWD_KNOWN_APIS: fwd,fwd_splitkv_combine@fwd_splitkv,fwd_appendkv,pagedkv_prefill
list(APPEND FMHA_FWD_CODE_GEN_COMMON_ARGS --filter *_nlogits*_nskip*,*@*_nlogits*_nbias*,*,*_nlogits*_nskip*_pagedkv)
endif()
# generate a list of kernels, but not actually emit files at config sta
execute_process(
COMMAND ${Python3_EXECUTABLE} ${FMHA_FWD_CODE_GEN_COMMON_ARGS}

8
example/ck_tile/03_gemm/run_gemm_example.inc
View File

@@ -181,15 +181,15 @@ auto shuffle_b(const ck_tile::HostTensor<T>& t)
if(ck_tile::is_gfx12_supported())
{
// TODO: Please modify it once kABK0PerLane is changed in WmmaTraitsBase<gfx12>
constexpr int divisor = 2;
constexpr int kABK0PerLane = 2;
constexpr int kABK1PerLane = 8;
constexpr int kABK0PerLane = GemmConfig::K_Warp_Tile / divisor / kABK1PerLane;
ck_tile::HostTensor<T> t_view({n_ / GemmConfig::N_Warp_Tile,
GemmConfig::N_Warp_Tile,
k_ / GemmConfig::K_Warp_Tile,
divisor,
kABK0PerLane,
GemmConfig::K_Warp_Tile / divisor / kABK0PerLane});
divisor,
kABK1PerLane});
std::copy(t.begin(), t.end(), t_view.begin());
return ck_tile::reference_permute(t_view, {0, 2, 4, 1, 3, 5});
}

8
example/ck_tile/17_grouped_gemm/grouped_gemm.hpp
View File

@@ -314,15 +314,15 @@ auto shuffle_b(const ck_tile::HostTensor<T>& t)
if(ck_tile::is_gfx12_supported())
{
// TODO: Please modify it once kABK0PerLane is changed in WmmaTraitsBase<gfx12>
constexpr int divisor = 2;
constexpr int kABK0PerLane = 2;
constexpr int kABK1PerLane = 8;
constexpr int kABK0PerLane = GemmConfig::K_Warp_Tile / divisor / kABK1PerLane;
ck_tile::HostTensor<T> t_view({n_ / GemmConfig::N_Warp_Tile,
GemmConfig::N_Warp_Tile,
k_ / GemmConfig::K_Warp_Tile,
divisor,
kABK0PerLane,
GemmConfig::K_Warp_Tile / divisor / kABK0PerLane});
divisor,
kABK1PerLane});
std::copy(t.begin(), t.end(), t_view.begin());
return ck_tile::reference_permute(t_view, {0, 2, 4, 1, 3, 5});
}

8
example/ck_tile/18_flatmm/run_flatmm_example.inc
View File

@@ -45,15 +45,15 @@ auto shuffle_b(const ck_tile::HostTensor<T>& t)
if(ck_tile::is_gfx12_supported())
{
// TODO: Please modify it once kABK0PerLane is changed in WmmaTraitsBase<gfx12>
constexpr int divisor = 2;
constexpr int kABK0PerLane = 2;
constexpr int kABK1PerLane = 8;
constexpr int kABK0PerLane = FlatmmConfig::K_Warp_Tile / divisor / kABK1PerLane;
ck_tile::HostTensor<T> t_view({n_ / FlatmmConfig::N_Warp_Tile,
FlatmmConfig::N_Warp_Tile,
k_ / FlatmmConfig::K_Warp_Tile,
divisor,
kABK0PerLane,
FlatmmConfig::K_Warp_Tile / divisor / kABK0PerLane});
divisor,
kABK1PerLane});
std::copy(t.begin(), t.end(), t_view.begin());
return ck_tile::reference_permute(t_view, {0, 2, 4, 1, 3, 5});
}