Skip to content

GitLab

Unverified Commit e0eabf02 authored by Haocong WANG's avatar Haocong WANG Committed by GitHub
Browse files

Apply suggestions from code review 


Co-authored-by: default avatarJohn Afaganis <john.afaganis@amd.com>
parent 8ce41034
Hide whitespace changes
Inline Side-by-side
Showing with 0 additions and 4 deletions
example/65_gemm_multiply_multiply/CMakeLists.txt
View file @ e0eabf02
add_example_executable(example_gemm_multiply_multiply_xdl_fp8 gemm_multiply_multiply_xdl_fp8.cpp) add_example_executable(example_gemm_multiply_multiply_xdl_fp8 gemm_multiply_multiply_xdl_fp8.cpp)
# target_compile_options(example_gemm_multiply_multiply_xdl_fp8 PRIVATE -mllvm -greedy-reverse-local-assignment=1 -save-temps=$PWD -Wno-gnu-line-marker)
add_example_executable(example_gemm_multiply_multiply_xdl_fp8_ab_scale gemm_multiply_multiply_xdl_fp8_ab_scale.cpp) add_example_executable(example_gemm_multiply_multiply_xdl_fp8_ab_scale gemm_multiply_multiply_xdl_fp8_ab_scale.cpp)
add_example_executable(example_gemm_multiply_multiply_xdl_fp8_bpreshuffle gemm_multiply_multiply_xdl_fp8_bpreshuffle.cpp) add_example_executable(example_gemm_multiply_multiply_xdl_fp8_bpreshuffle gemm_multiply_multiply_xdl_fp8_bpreshuffle.cpp)
# target_compile_options(example_gemm_multiply_multiply_xdl_fp8_bpreshuffle PRIVATE -mllvm -greedy-reverse-local-assignment=1 -save-temps=$PWD -Wno-gnu-line-marker)
add_example_executable(example_gemm_add_add_xdl_fp16 gemm_add_add_xdl_fp16.cpp) add_example_executable(example_gemm_add_add_xdl_fp16 gemm_add_add_xdl_fp16.cpp)
add_example_executable(example_gemm_multiply_multiply_xdl_int8 gemm_multiply_multiply_xdl_int8.cpp) add_example_executable(example_gemm_multiply_multiply_xdl_int8 gemm_multiply_multiply_xdl_int8.cpp)
example/65_gemm_multiply_multiply/gemm_multiply_multiply_xdl_fp8_bpreshuffle.cpp
View file @ e0eabf02
...@@ -136,7 +136,6 @@ using CDEElementOp = MultiplyMultiply; ...@@ -136,7 +136,6 @@ using CDEElementOp = MultiplyMultiply;
static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::Default; static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::Default;
// using DeviceOpInstance = ck::tensor_operation::device::DeviceGemmMultiD_Xdl_CShuffle_V3
using DeviceOpInstance = ck::tensor_operation::device::DeviceGemmMultiD_Xdl_CShuffle_V3_BPreshuffle using DeviceOpInstance = ck::tensor_operation::device::DeviceGemmMultiD_Xdl_CShuffle_V3_BPreshuffle
// clang-format off // clang-format off
< Row, Col, DsLayout, ELayout, A0DataType, B0DataType, DsDataType, EDataType, AccDataType, CShuffleDataType, < Row, Col, DsLayout, ELayout, A0DataType, B0DataType, DsDataType, EDataType, AccDataType, CShuffleDataType,
......
profiler/include/profiler/profile_gemm_multiply_multiply_impl.hpp
View file @ e0eabf02
...@@ -189,7 +189,6 @@ bool profile_gemm_multiply_multiply_impl(int do_verification, ...@@ -189,7 +189,6 @@ bool profile_gemm_multiply_multiply_impl(int do_verification,
for(auto& op_ptr : op_ptrs) for(auto& op_ptr : op_ptrs)
{ {
// Seems like when performance measurement has bug when spiltK is large // Seems like when performance measurement has bug when spiltK is large
// std::vector<int> kbatch_list = {1, 2, 4, 8, 16, 19, 32, 38};
std::vector<int> kbatch_list = {1, 2, 4, 8, 16}; std::vector<int> kbatch_list = {1, 2, 4, 8, 16};
if(KBatch > 0) if(KBatch > 0)
......
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