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 18:20:02 -05:00
parent f885c131d8
commit d18428901e
8 changed files with 21 additions and 44 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
composable_kernel/include/tensor_operation/gridwise_dynamic_gemm_xdlops_v2r3.hpp
View File

@@ -203,9 +203,6 @@ struct GridwiseDynamicGemm_k0mk1_k0nk1_mn_xdlops_v2r3
__host__ __device__ static constexpr auto
MakeCM0M1M2NGridDescriptor(const CMNGridDesc& c_m_n_grid_desc)
{
const auto M = c_m_n_grid_desc.GetLength(I0);
const auto N = c_m_n_grid_desc.GetLength(I1);
constexpr auto xdlops_gemm = XdlopsGemm<FloatAB, MPerWave, NPerWave, K1>{};
constexpr auto CLayout = xdlops_gemm.GetCLayout();
@@ -217,7 +214,6 @@ struct GridwiseDynamicGemm_k0mk1_k0nk1_mn_xdlops_v2r3
constexpr index_t MWaves = MPerBlock / (MPerWave * MRepeat);
constexpr index_t NWaves = NPerBlock / (NPerWave * NRepeat);
constexpr auto N0 = Number<CLayout.N1()>{};
constexpr auto N1 = Number<CLayout.N0()>{};
const auto c_m0_m1_m2_n_grid_desc = transform_dynamic_tensor_descriptor(
@@ -277,8 +273,6 @@ struct GridwiseDynamicGemm_k0mk1_k0nk1_mn_xdlops_v2r3
p_c_grid, c_m0_m1_m2_n_grid_desc.GetElementSpaceSize());
const auto K0 = a_k0_m_k1_grid_desc.GetLength(I0);
const auto M = a_k0_m_k1_grid_desc.GetLength(I1);
const auto N = b_k0_n_k1_grid_desc.GetLength(I1);
// divide block work by [M, N]
const auto block_work_idx =