Skip to content

GitLab

You need to sign in or sign up before continuing.
Commit 4dba005f authored by Adam Osewski's avatar Adam Osewski
Browse files

Add wavelet to Gemm standalone test.

parent 0370de0b
Hide whitespace changes
Inline Side-by-side
Showing with 127 additions and 0 deletions
test/gemm/CMakeLists.txt
View file @ 4dba005f
...@@ -18,6 +18,7 @@ add_library(gemm_standalone_xdl_fp16_instances STATIC ...@@ -18,6 +18,7 @@ add_library(gemm_standalone_xdl_fp16_instances STATIC
instance/gemm_f16_nn_instance.cpp instance/gemm_f16_nn_instance.cpp
instance/gemm_f16_nt_instance.cpp instance/gemm_f16_nt_instance.cpp
instance/gemm_f16_tn_instance.cpp instance/gemm_f16_tn_instance.cpp
instance/gemm_wavelet_f16_tn_instance.cpp
instance/gemm_f16_tt_instance.cpp instance/gemm_f16_tt_instance.cpp
) )
add_test_executable(test_gemm_standalone_xdl_fp16 gemm_standalone_xdl_fp16.cpp) add_test_executable(test_gemm_standalone_xdl_fp16 gemm_standalone_xdl_fp16.cpp)
......
test/gemm/gemm_standalone_xdl_fp16.cpp
View file @ 4dba005f
...@@ -10,6 +10,7 @@ ...@@ -10,6 +10,7 @@
#include "gemm_f16_nt_instance.hpp" #include "gemm_f16_nt_instance.hpp"
#include "gemm_f16_tn_instance.hpp" #include "gemm_f16_tn_instance.hpp"
#include "gemm_f16_tt_instance.hpp" #include "gemm_f16_tt_instance.hpp"
#include "gemm_wavelet_f16_tn_instance.hpp"
using Row = ck::tensor_layout::gemm::RowMajor; using Row = ck::tensor_layout::gemm::RowMajor;
using Col = ck::tensor_layout::gemm::ColumnMajor; using Col = ck::tensor_layout::gemm::ColumnMajor;
...@@ -74,6 +75,10 @@ int main(int argc, char* argv[]) ...@@ -74,6 +75,10 @@ int main(int argc, char* argv[])
{GemmParams{2048, 1664, 4096}, LayoutConfig{true, false, true}, add_gemm_f16_tn_256x128}, {GemmParams{2048, 1664, 4096}, LayoutConfig{true, false, true}, add_gemm_f16_tn_256x128},
{GemmParams{1024, 1664, 4096}, LayoutConfig{true, false, true}, add_gemm_f16_tn_128x128}, {GemmParams{1024, 1664, 4096}, LayoutConfig{true, false, true}, add_gemm_f16_tn_128x128},
{GemmParams{1024, 832, 4096}, LayoutConfig{true, false, true}, add_gemm_f16_tn_128x64}, {GemmParams{1024, 832, 4096}, LayoutConfig{true, false, true}, add_gemm_f16_tn_128x64},
{GemmParams{2048, 3328, 4096}, LayoutConfig{true, false, true}, add_gemm_wavelet_f16_tn_256x256},
{GemmParams{2048, 1664, 4096}, LayoutConfig{true, false, true}, add_gemm_wavelet_f16_tn_256x128},
{GemmParams{1024, 1664, 4096}, LayoutConfig{true, false, true}, add_gemm_wavelet_f16_tn_128x128},
{GemmParams{1024, 832, 4096}, LayoutConfig{true, false, true}, add_gemm_wavelet_f16_tn_128x64},
{GemmParams{2048, 3328, 4096}, LayoutConfig{true, true, true}, add_gemm_f16_tt_256x256}, {GemmParams{2048, 3328, 4096}, LayoutConfig{true, true, true}, add_gemm_f16_tt_256x256},
{GemmParams{2048, 1664, 4096}, LayoutConfig{true, true, true}, add_gemm_f16_tt_256x128}, {GemmParams{2048, 1664, 4096}, LayoutConfig{true, true, true}, add_gemm_f16_tt_256x128},
{GemmParams{1024, 1664, 4096}, LayoutConfig{true, true, true}, add_gemm_f16_tt_128x128}, {GemmParams{1024, 1664, 4096}, LayoutConfig{true, true, true}, add_gemm_f16_tt_128x128},
......
test/gemm/instance/gemm_wavelet_f16_tn_instance.cpp 0 → 100644
View file @ 4dba005f
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#include <cstdlib>
#include "ck/ck.hpp"
#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
#include "ck/tensor_operation/gpu/device/device_gemm_xdl_waveletmodel_cshuffle.hpp"
#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
#include "gemm_wavelet_f16_tn_instance.hpp"
namespace ck {
namespace tensor_operation {
namespace device {
namespace instance {
using F16 = ck::half_t;
using F32 = float;
using Row = ck::tensor_layout::gemm::RowMajor;
using Col = ck::tensor_layout::gemm::ColumnMajor;
template <ck::index_t... Is>
using S = ck::Sequence<Is...>;
using PassThrough = ck::tensor_operation::element_wise::PassThrough;
static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
using gemm_f16_tn_256x256 = std::tuple<
// clang-format off
//##################### | ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle| CData| A| B| C| GEMM| NumGemmK| ABBlockTransfer| BlockGemm| MPer| NPer| KPer| AK1| BK1| MPer| NPer| MXdl| NXdl| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds| BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CShuffle| CShuffle| CBlockTransferClusterLengths| CBlockTransfer|
//##################### | | | | Type| Type| Type| DataType| Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch| ThreadGroupSize| ThreadGroupSize| Block| Block| Block| | | XDL| XDL| Per| Per| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MWaveMPerXdl| ScalarPerVector|
//##################### | | | | | | | | | Operation| Operation| Operation| | Stage| | | | | | | | | | Wave| Wave| Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _NBlock_NWaveNPerXdl| _NWaveNPerXdl|
//##################### | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
DeviceGemm_Xdl_WaveletModel_CShuffle< Row, Col, Row, F16, F16, F32, F16, F16, PassThrough, PassThrough, PassThrough, GemmDefault, 1, 256, 256, 256, 256, 32, 8, 8, 32, 32, 4, 4, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, 1, 1, S<1, 32, 1, 8>, 8>
// clang-format on
>;
using gemm_f16_tn_256x128 = std::tuple<
// clang-format off
//##################### | ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle| CData| A| B| C| GEMM| NumGemmK| ABBlockTransfer| BlockGemm| MPer| NPer| KPer| AK1| BK1| MPer| NPer| MXdl| NXdl| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds| BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CShuffle| CShuffle| CBlockTransferClusterLengths| CBlockTransfer|
//##################### | | | | Type| Type| Type| DataType| Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch| ThreadGroupSize| ThreadGroupSize| Block| Block| Block| | | XDL| XDL| Per| Per| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MWaveMPerXdl| ScalarPerVector|
//##################### | | | | | | | | | Operation| Operation| Operation| | Stage| | | | | | | | | | Wave| Wave| Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _NBlock_NWaveNPerXdl| _NWaveNPerXdl|
//##################### | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
DeviceGemm_Xdl_WaveletModel_CShuffle< Row, Col, Row, F16, F16, F32, F16, F16, PassThrough, PassThrough, PassThrough, GemmDefault, 1, 256, 256, 256, 128, 32, 8, 8, 32, 32, 4, 2, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, 1, 1, S<1, 32, 1, 8>, 8>
// clang-format on
>;
using gemm_f16_tn_128x128 = std::tuple<
// clang-format off
//##################### | ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle| CData| A| B| C| GEMM| NumGemmK| ABBlockTransfer| BlockGemm| MPer| NPer| KPer| AK1| BK1| MPer| NPer| MXdl| NXdl| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds| BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CShuffle| CShuffle| CBlockTransferClusterLengths| CBlockTransfer|
//##################### | | | | Type| Type| Type| DataType| Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch| ThreadGroupSize| ThreadGroupSize| Block| Block| Block| | | XDL| XDL| Per| Per| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MWaveMPerXdl| ScalarPerVector|
//##################### | | | | | | | | | Operation| Operation| Operation| | Stage| | | | | | | | | | Wave| Wave| Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _NBlock_NWaveNPerXdl| _NWaveNPerXdl|
//##################### | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
DeviceGemm_Xdl_WaveletModel_CShuffle< Row, Col, Row, F16, F16, F32, F16, F16, PassThrough, PassThrough, PassThrough, GemmDefault, 1, 256, 256, 128, 128, 32, 8, 8, 32, 32, 2, 2, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, 1, 1, S<1, 32, 1, 8>, 8>
// clang-format on
>;
using gemm_f16_tn_128x64 = std::tuple<
// clang-format off
//##################### | ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle| CData| A| B| C| GEMM| NumGemmK| ABBlockTransfer| BlockGemm| MPer| NPer| KPer| AK1| BK1| MPer| NPer| MXdl| NXdl| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds| BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CShuffle| CShuffle| CBlockTransferClusterLengths| CBlockTransfer|
//##################### | | | | Type| Type| Type| DataType| Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch| ThreadGroupSize| ThreadGroupSize| Block| Block| Block| | | XDL| XDL| Per| Per| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MWaveMPerXdl| ScalarPerVector|
//##################### | | | | | | | | | Operation| Operation| Operation| | Stage| | | | | | | | | | Wave| Wave| Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _NBlock_NWaveNPerXdl| _NWaveNPerXdl|
//##################### | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
DeviceGemm_Xdl_WaveletModel_CShuffle< Row, Col, Row, F16, F16, F32, F16, F16, PassThrough, PassThrough, PassThrough, GemmDefault, 1, 256, 256, 128, 64, 32, 8, 8, 32, 32, 2, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, 1, 1, S<1, 32, 1, 8>, 8>
// clang-format on
>;
void add_gemm_wavelet_f16_tn_256x256(std::vector<std::unique_ptr<BaseOperator>>& instances)
{
add_device_operation_instances(instances, gemm_f16_tn_256x256{});
}
void add_gemm_wavelet_f16_tn_256x128(std::vector<std::unique_ptr<BaseOperator>>& instances)
{
add_device_operation_instances(instances, gemm_f16_tn_256x128{});
}
void add_gemm_wavelet_f16_tn_128x128(std::vector<std::unique_ptr<BaseOperator>>& instances)
{
add_device_operation_instances(instances, gemm_f16_tn_128x128{});
}
void add_gemm_wavelet_f16_tn_128x64(std::vector<std::unique_ptr<BaseOperator>>& instances)
{
add_device_operation_instances(instances, gemm_f16_tn_128x64{});
}
} // namespace instance
} // namespace device
} // namespace tensor_operation
} // namespace ck
test/gemm/instance/gemm_wavelet_f16_tn_instance.hpp 0 → 100644
View file @ 4dba005f
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#include <memory>
#include <vector>
#include "include/ck/tensor_operation/gpu/device/device_base.hpp"
namespace ck {
namespace tensor_operation {
namespace device {
namespace instance {
void add_gemm_wavelet_f16_tn_256x256(std::vector<std::unique_ptr<BaseOperator>>& instances);
void add_gemm_wavelet_f16_tn_256x128(std::vector<std::unique_ptr<BaseOperator>>& instances);
void add_gemm_wavelet_f16_tn_128x128(std::vector<std::unique_ptr<BaseOperator>>& instances);
void add_gemm_wavelet_f16_tn_128x64(std::vector<std::unique_ptr<BaseOperator>>& instances);
} // namespace instance
} // namespace device
} // namespace tensor_operation
} // namespace ck
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