Skip to content

GitLab

Commit 205e0365 authored by Jing Zhang's avatar Jing Zhang
Browse files

fix

parent c13366af
Hide whitespace changes
Inline Side-by-side
Showing with 2 additions and 2 deletions
include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_v3.hpp
View file @ 205e0365
...@@ -653,7 +653,7 @@ struct GridwiseGemm_xdl_cshuffle_v3 ...@@ -653,7 +653,7 @@ struct GridwiseGemm_xdl_cshuffle_v3
// in some cases. // in some cases.
else if constexpr(is_same<tensor_layout::gemm::RowMajor, ALayout>::value) else if constexpr(is_same<tensor_layout::gemm::RowMajor, ALayout>::value)
{ {
constexpr index_t LdsSize = 32 * 4 / KPerBlock / sizeof(ADataType); constexpr index_t LdsSize = 32 * 4 / KPerBlock / sizeof(ADataType) / APackedSize;
constexpr auto MLdsLayer = LdsSize < 1 ? 1 : LdsSize; constexpr auto MLdsLayer = LdsSize < 1 ? 1 : LdsSize;
constexpr auto a_lds_block_desc = make_naive_tensor_descriptor( constexpr auto a_lds_block_desc = make_naive_tensor_descriptor(
make_tuple( make_tuple(
...@@ -788,7 +788,7 @@ struct GridwiseGemm_xdl_cshuffle_v3 ...@@ -788,7 +788,7 @@ struct GridwiseGemm_xdl_cshuffle_v3
else if constexpr(is_same<tensor_layout::gemm::ColumnMajor, BLayout>::value) else if constexpr(is_same<tensor_layout::gemm::ColumnMajor, BLayout>::value)
{ {
// NLdsLayer * K0 as logical Bank // NLdsLayer * K0 as logical Bank
constexpr index_t LdsSize = 32 * 4 / KPerBlock / sizeof(BDataType); constexpr index_t LdsSize = 32 * 4 / KPerBlock / sizeof(BDataType) / BPackedSize;
constexpr index_t NLdsLayer = LdsSize < 1 ? 1 : LdsSize; constexpr index_t NLdsLayer = LdsSize < 1 ? 1 : LdsSize;
constexpr auto b_lds_block_desc = make_naive_tensor_descriptor( constexpr auto b_lds_block_desc = make_naive_tensor_descriptor(
make_tuple( make_tuple(
......
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