ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-20 06:49:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

upgrade from clang-format-12 to clang-format-18 (#2568)

Browse Source 
* upgrade to clang-format-18

* update to clang-format-18 in pre-commit-config
This commit is contained in:
Illia Silin
2025-07-28 11:34:07 -07:00
committed by GitHub
parent 9786087010
commit 504b101da3
373 changed files with 3351 additions and 3760 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
include/ck/tensor_operation/gpu/block/blockwise_gemm_mx_pipeline_xdlops_base.hpp
View File

@@ -152,7 +152,7 @@ struct BlockwiseGemmXdlops_mx_pipeline_base
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();

4
include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_wmmaops_base.hpp
View File

@@ -93,7 +93,7 @@ struct BlockwiseGemmWmmaops_pipeline_base
struct Empty
{
__device__ Empty(){};
__device__ Empty() {};
template <index_t NBuffer>
__device__ void GlobalLoad(bool cond)
{
@@ -119,7 +119,7 @@ struct BlockwiseGemmWmmaops_pipeline_base
GridBuffer b_scale_grid_buf_)
: b_scale_thread_copy(b_scale_thread_copy_),
b_scale_grid_desc(b_scale_grid_desc_),
b_scale_grid_buf(b_scale_grid_buf_){};
b_scale_grid_buf(b_scale_grid_buf_) {};
static constexpr index_t num_scale_k_block = BScaleThreadDesc{}.GetLength(Number<1>{});
static constexpr index_t num_scale_krepeat = KRepeat / num_scale_k_block;

8
include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops.hpp
View File

@@ -96,9 +96,9 @@ template <
index_t KPack,
bool TransposeC = false,
index_t AMmaKStride =
KPack* XdlopsGemm<FloatAB, MPerXDL, NPerXDL, KPack, FloatAB, TransposeC>{}.K0PerXdlops,
KPack * XdlopsGemm<FloatAB, MPerXDL, NPerXDL, KPack, FloatAB, TransposeC>{}.K0PerXdlops,
index_t BMmaKStride =
KPack* XdlopsGemm<FloatAB, MPerXDL, NPerXDL, KPack, FloatAB, TransposeC>{}.K0PerXdlops>
KPack * XdlopsGemm<FloatAB, MPerXDL, NPerXDL, KPack, FloatAB, TransposeC>{}.K0PerXdlops>
struct BlockwiseGemmXdlops_pipeline_v4
{
static constexpr auto I0 = Number<0>{};
@@ -188,7 +188,7 @@ struct BlockwiseGemmXdlops_pipeline_v4
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();
@@ -217,7 +217,7 @@ struct BlockwiseGemmXdlops_pipeline_v4
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex8D(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex8D(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();

4
include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_base.hpp
View File

@@ -153,7 +153,7 @@ struct BlockwiseGemmXdlops_pipeline_base
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();
@@ -182,7 +182,7 @@ struct BlockwiseGemmXdlops_pipeline_base
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex8D(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex8D(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();

4
include/ck/tensor_operation/gpu/block/blockwise_gemm_smfmac_xdlops.hpp
View File

@@ -110,7 +110,7 @@ struct BlockwiseGemmXdlops_k0mk1_k0nk1_m0n0m1n1m2m3m4n2_v1
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();
const auto waveId_m = wave_idx[I0];
@@ -138,7 +138,7 @@ struct BlockwiseGemmXdlops_k0mk1_k0nk1_m0n0m1n1m2m3m4n2_v1
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex8D(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex8D(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();
const auto waveId_m = wave_idx[I0];

12
include/ck/tensor_operation/gpu/block/blockwise_gemm_xdlops.hpp
View File

@@ -114,7 +114,7 @@ struct BlockwiseGemmXdlops_k0mk1_k0nk1_m0n0m1n1m2m3m4n2_v1
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();
@@ -143,7 +143,7 @@ struct BlockwiseGemmXdlops_k0mk1_k0nk1_m0n0m1n1m2m3m4n2_v1
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex8D(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex8D(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();
@@ -667,9 +667,9 @@ template <
index_t KPack,
bool TransposeC = false,
index_t AMmaKStride =
KPack* XdlopsGemm<FloatAB, MPerXDL, NPerXDL, KPack, FloatAB, TransposeC>{}.K0PerXdlops,
KPack * XdlopsGemm<FloatAB, MPerXDL, NPerXDL, KPack, FloatAB, TransposeC>{}.K0PerXdlops,
index_t BMmaKStride =
KPack* XdlopsGemm<FloatAB, MPerXDL, NPerXDL, KPack, FloatAB, TransposeC>{}.K0PerXdlops>
KPack * XdlopsGemm<FloatAB, MPerXDL, NPerXDL, KPack, FloatAB, TransposeC>{}.K0PerXdlops>
struct BlockwiseGemmXdlops_v2
{
static constexpr auto I0 = Number<0>{};
@@ -742,7 +742,7 @@ struct BlockwiseGemmXdlops_v2
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();
@@ -771,7 +771,7 @@ struct BlockwiseGemmXdlops_v2
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex8D(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex8D(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();

2
include/ck/tensor_operation/gpu/block/blockwise_gemm_xdlops_skip_b_lds.hpp
View File

@@ -90,7 +90,7 @@ struct BlockwiseGemmXdlops_k0mk1_k0nk1_m0n0m1n1m2m3m4n2_v1r1
template <index_t m0, index_t n0, index_t xdlops_i, index_t blk_i>
__device__ static auto
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
CalculateCThreadOriginDataIndex(Number<m0>, Number<n0>, Number<xdlops_i>, Number<blk_i>)
{
const auto wave_idx = GetWaveIdx();

6
include/ck/tensor_operation/gpu/block/thread_group_tensor_slice_transfer_direct_load.hpp
View File

@@ -258,8 +258,7 @@ struct ThreadGroupTensorSliceTransfer_DirectLoad
src_buf.template DirectCopyToLds<remove_cvref_t<decltype(dst_buf)>, ScalarPerVector>(
dst_buf, src_offset, dst_offset, is_src_valid);
constexpr auto move_on_dim = [&]() constexpr
{
constexpr auto move_on_dim = [&]() constexpr {
StaticallyIndexedArray<bool, nDim> move_on_dim_;
static_for<0, nDim, 1>{}([&](auto i) {
@@ -271,8 +270,7 @@ struct ThreadGroupTensorSliceTransfer_DirectLoad
});
return move_on_dim_;
}
();
}();
// Decide whether to move forward or backward.
constexpr auto forward_sweep = [&]() {

12
include/ck/tensor_operation/gpu/block/thread_group_tensor_slice_transfer_gather_direct_load.hpp
View File

@@ -281,8 +281,7 @@ struct ThreadGroupTensorSliceTransfer_Gather_DirectLoad
src_buf.template DirectCopyToLds<remove_cvref_t<decltype(dst_buf)>, ScalarPerVector>(
dst_buf, src_offset, dst_offset, true);
constexpr auto move_src_on_dim = [&]() constexpr
{
constexpr auto move_src_on_dim = [&]() constexpr {
StaticallyIndexedArray<bool, nDim> move_on_dim_;
static_for<0, nDim, 1>{}([&](auto i) {
@@ -295,11 +294,9 @@ struct ThreadGroupTensorSliceTransfer_Gather_DirectLoad
});
return move_on_dim_;
}
();
}();
constexpr auto move_dst_on_dim = [&]() constexpr
{
constexpr auto move_dst_on_dim = [&]() constexpr {
StaticallyIndexedArray<bool, nDim> move_on_dim_;
static_for<0, nDim, 1>{}([&](auto i) {
@@ -311,8 +308,7 @@ struct ThreadGroupTensorSliceTransfer_Gather_DirectLoad
});
return move_on_dim_;
}
();
}();
// Decide whether to move forward or backward.
constexpr auto forward_sweep = [&]() {