ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-12 09:16:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

tidy

Browse Source
This commit is contained in:
Chao Liu
2021-08-09 17:33:32 +00:00
parent e62bae7a4a
commit 54fba515b3
8 changed files with 0 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
composable_kernel/include/problem_transform/transform_forward_convolution_into_gemm_v4r4_nchw_kcyx_nkhw.hpp
View File

@@ -126,9 +126,6 @@ transform_forward_convolution_into_gemm_v4r4_nchw_kcyx_nkhw_no_pad(
const auto C = in_n_c_hi_wi_global_desc.GetLength(I1);
const auto K = out_n_k_ho_wo_global_desc.GetLength(I1);
const auto Hi = in_n_c_hi_wi_global_desc.GetLength(I2);
const auto Wi = in_n_c_hi_wi_global_desc.GetLength(I3);
const auto Ho = out_n_k_ho_wo_global_desc.GetLength(I2);
const auto Wo = out_n_k_ho_wo_global_desc.GetLength(I3);
@@ -209,9 +206,6 @@ __host__ __device__ constexpr auto transform_forward_convolution_into_gemm_v4r4_
const auto C = in_n_c_hi_wi_global_desc.GetLength(I1);
const auto K = out_n_k_ho_wo_global_desc.GetLength(I1);
const auto Hi = in_n_c_hi_wi_global_desc.GetLength(I2);
const auto Wi = in_n_c_hi_wi_global_desc.GetLength(I3);
const auto Ho = out_n_k_ho_wo_global_desc.GetLength(I2);
const auto Wo = out_n_k_ho_wo_global_desc.GetLength(I3);

3
composable_kernel/include/problem_transform/transform_forward_convolution_into_gemm_v4r4_nhwc_kyxc_nhwk.hpp
View File

@@ -125,9 +125,6 @@ __host__ __device__ constexpr auto transform_forward_convolution_into_gemm_v4r4_
const auto C = in_n_hi_wi_c_grid_desc.GetLength(I3);
const auto K = out_n_ho_wo_k_grid_desc.GetLength(I3);
const auto Hi = in_n_hi_wi_c_grid_desc.GetLength(I1);
const auto Wi = in_n_hi_wi_c_grid_desc.GetLength(I2);
const auto Ho = out_n_ho_wo_k_grid_desc.GetLength(I1);
const auto Wo = out_n_ho_wo_k_grid_desc.GetLength(I2);

1
composable_kernel/include/tensor_operation/blockwise_gemm_dlops_v3.hpp
View File

@@ -69,7 +69,6 @@ struct BlockwiseGemmDlops_km_kn_m0m1n0n1_v3
"wrong! K dimension not consistent\n");
constexpr index_t K = BlockMatrixA{}.GetLength(I1); // A is transposed
constexpr index_t N = BlockMatrixB{}.GetLength(I1);
constexpr index_t H = BlockMatrixB{}.GetLength(I2);
constexpr index_t W = BlockMatrixB{}.GetLength(I3);

3
composable_kernel/include/tensor_operation/blockwise_gemm_xdlops.hpp
View File

@@ -52,7 +52,6 @@ struct BlockwiseGemmXdlops_km_kn_m0m1m2n_v1
const index_t waveId = thread_id / WaveSize;
const index_t laneId = thread_id % WaveSize;
const index_t waveId_m = waveId / NWaves;
const index_t waveId_n = waveId % NWaves;
if constexpr(xdlops_gemm.IsKReduction)
{
@@ -73,7 +72,6 @@ struct BlockwiseGemmXdlops_km_kn_m0m1m2n_v1
const index_t thread_id = get_thread_local_1d_id();
const index_t waveId = thread_id / WaveSize;
const index_t laneId = thread_id % WaveSize;
const index_t waveId_m = waveId / NWaves;
const index_t waveId_n = waveId % NWaves;
if constexpr(xdlops_gemm.IsKReduction)
@@ -293,7 +291,6 @@ struct BlockwiseGemmXdlops_km_kn_m0m1m2n_v1_2x2pipeline
const index_t thread_id = get_thread_local_1d_id();
const index_t waveId = thread_id / WaveSize;
const index_t laneId = thread_id % WaveSize;
const index_t waveId_m = waveId / NWaves;
const index_t waveId_n = waveId % NWaves;
if constexpr(xdlops_gemm.IsKReduction)

2
composable_kernel/include/utility/sequence.hpp
View File

@@ -685,8 +685,6 @@ __host__ __device__ constexpr auto operator+(Number<Y>, Sequence<Xs...>)
template <index_t Y, index_t... Xs>
__host__ __device__ constexpr auto operator-(Number<Y>, Sequence<Xs...>)
{
constexpr auto seq_x = Sequence<Xs...>{};
return Sequence<(Y - Xs)...>{};
}

1
composable_kernel/src/kernel_wrapper/dynamic_convolution_forward_implicit_gemm_v4r4_xdlops_nhwc_kyxc_nhwk.cpp
View File

@@ -225,7 +225,6 @@ extern "C" __global__ void
constexpr auto I0 = Number<0>{};
constexpr auto I1 = Number<1>{};
constexpr auto I2 = Number<2>{};
constexpr auto I3 = Number<3>{};
constexpr auto in_n_hi_wi_c_desc =
make_dynamic_naive_tensor_descriptor_packed_v2(make_tuple(256, 28, 28, 256));