Skip to content

GitLab

Commit b62926dc authored by aska-0096's avatar aska-0096
Browse files

Remove tensor layout limitation to LDS usage in tesnor contraction

parent 2ab0d8f5
Hide whitespace changes
Inline Side-by-side
Showing with 2 additions and 4 deletions
include/ck/tensor_operation/gpu/device/impl/device_batched_contraction_multiple_d_wmma_cshuffle.hpp
View file @ b62926dc
...@@ -133,10 +133,8 @@ struct DeviceBatchedContractionMultipleD_Wmma_CShuffle ...@@ -133,10 +133,8 @@ struct DeviceBatchedContractionMultipleD_Wmma_CShuffle
static constexpr auto NWaves = NPerBlock / (NRepeat * NPerWmma); static constexpr auto NWaves = NPerBlock / (NRepeat * NPerWmma);
static constexpr auto WmmaK = K1 == 16 ? 32 : 16; static constexpr auto WmmaK = K1 == 16 ? 32 : 16;
static constexpr auto AEnableLds_auto = static constexpr auto AEnableLds_auto = NWaves == 1 ? false : true;
(NWaves == 1 && is_same<tensor_layout::gemm::RowMajor, ALayout>::value) ? false : true; static constexpr auto BEnableLds_auto = MWaves == 1 ? false : true;
static constexpr auto BEnableLds_auto =
(MWaves == 1 && is_same<tensor_layout::gemm::ColumnMajor, BLayout>::value) ? false : true;
// If true, LDS is used unconditionally // If true, LDS is used unconditionally
static constexpr auto AEnableLds_manu = false; static constexpr auto AEnableLds_manu = false;
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment