
CUTLASS 1.3 Release - Efficient GEMM kernel targeting Volta Tensor Cores via mma.sync instruction added in CUDA 10.1.
180 lines
6.3 KiB
Plaintext
180 lines
6.3 KiB
Plaintext
/***************************************************************************************************
|
|
* Copyright (c) 2017-2019, NVIDIA CORPORATION. All rights reserved.
|
|
*
|
|
* Redistribution and use in source and binary forms, with or without modification, are permitted
|
|
* provided that the following conditions are met:
|
|
* * Redistributions of source code must retain the above copyright notice, this list of
|
|
* conditions and the following disclaimer.
|
|
* * Redistributions in binary form must reproduce the above copyright notice, this list of
|
|
* conditions and the following disclaimer in the documentation and/or other materials
|
|
* provided with the distribution.
|
|
* * Neither the name of the NVIDIA CORPORATION nor the names of its contributors may be used
|
|
* to endorse or promote products derived from this software without specific prior written
|
|
* permission.
|
|
*
|
|
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR
|
|
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
|
|
* FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE
|
|
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
|
|
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
|
|
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
|
|
* STRICT LIABILITY, OR TOR (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
|
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
*
|
|
**************************************************************************************************/
|
|
|
|
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 610))
|
|
|
|
#include "cutlass/gemm/gemm.h"
|
|
#include "cutlass/gemm/igemm_traits.h"
|
|
#include "tools/test/perf/cutlass_perf_test.h"
|
|
#include "tools/test/perf/gemm/gemm_perf_testbed.h"
|
|
#include "tools/test/perf/gemm/gemm_profiler.h"
|
|
#include "tools/test/perf/gemm/cutlass_dispatch.h"
|
|
|
|
#pragma warning( disable : 4503)
|
|
|
|
namespace perf {
|
|
|
|
////////////////////////////////////////////////////////////////////////////////////////////////////
|
|
|
|
template <typename DummyT>
|
|
int profile_igemm(TestbenchOutput<GemmProblem> &output, TestbenchOptions const &options, Config const &config) {
|
|
|
|
typedef perf::GemmProfiler<int8_t, int8_t, int, int, int> GemmProfiler;
|
|
|
|
// compute capability check
|
|
if (!options.compute_capability(6, 1)) {
|
|
return 0;
|
|
}
|
|
|
|
int results = 0;
|
|
|
|
{
|
|
typedef cutlass::gemm::IgemmTraits<
|
|
cutlass::MatrixLayout::kColumnMajor,
|
|
cutlass::MatrixLayout::kRowMajor
|
|
> GemmTraits;
|
|
|
|
typedef typename CutlassDispatchBasic<GemmTraits>::Dispatch Dispatch;
|
|
|
|
results |= profile_gemm<Dispatch, GemmProfiler>(output, "igemm_nt", options, config);
|
|
}
|
|
|
|
{
|
|
typedef cutlass::gemm::IgemmTraits<
|
|
cutlass::MatrixLayout::kColumnMajor,
|
|
cutlass::MatrixLayout::kColumnMajor
|
|
> GemmTraits;
|
|
|
|
typedef typename CutlassDispatchBasic<GemmTraits>::Dispatch Dispatch;
|
|
|
|
results |= profile_gemm<Dispatch, GemmProfiler>(output, "igemm_nn", options, config);
|
|
}
|
|
|
|
{
|
|
typedef cutlass::gemm::IgemmTraits<
|
|
cutlass::MatrixLayout::kRowMajor,
|
|
cutlass::MatrixLayout::kColumnMajor
|
|
> GemmTraits;
|
|
|
|
typedef typename CutlassDispatchBasic<GemmTraits>::Dispatch Dispatch;
|
|
|
|
results |= profile_gemm<Dispatch, GemmProfiler>(output, "igemm_tn", options, config);
|
|
}
|
|
|
|
{
|
|
typedef cutlass::gemm::IgemmTraits<
|
|
cutlass::MatrixLayout::kRowMajor,
|
|
cutlass::MatrixLayout::kRowMajor
|
|
> GemmTraits;
|
|
|
|
typedef typename CutlassDispatchBasic<GemmTraits>::Dispatch Dispatch;
|
|
|
|
results |= profile_gemm<Dispatch, GemmProfiler>(output, "igemm_tt", options, config);
|
|
}
|
|
|
|
return results;
|
|
}
|
|
|
|
template <typename DummyT>
|
|
int profile_igemm_32x32x128(TestbenchOutput<GemmProblem> &output, TestbenchOptions const &options, Config const &config) {
|
|
|
|
typedef perf::GemmProfiler<int8_t, int8_t, int, int, int> GemmProfiler;
|
|
|
|
// compute capability check
|
|
if (!options.compute_capability(6, 1)) {
|
|
return 0;
|
|
}
|
|
|
|
int results = 0;
|
|
|
|
{
|
|
typedef cutlass::gemm::IgemmTraits<cutlass::MatrixLayout::kColumnMajor,
|
|
cutlass::MatrixLayout::kColumnMajor, cutlass::Shape<128, 32, 32>, int,
|
|
cutlass::gemm::LinearScaling<int>, cutlass::Shape<32, 8, 4> > GemmTraits;
|
|
|
|
typedef typename CutlassDispatchBasic<GemmTraits>::Dispatch Dispatch;
|
|
|
|
results |= profile_gemm<Dispatch, GemmProfiler>(output, "igemm_32x32x128_nn",
|
|
options, config);
|
|
}
|
|
|
|
{
|
|
typedef cutlass::gemm::IgemmTraits<cutlass::MatrixLayout::kColumnMajor,
|
|
cutlass::MatrixLayout::kRowMajor, cutlass::Shape<128, 32, 32>, int,
|
|
cutlass::gemm::LinearScaling<int>, cutlass::Shape<32, 8, 4> > GemmTraits;
|
|
|
|
typedef typename CutlassDispatchBasic<GemmTraits>::Dispatch Dispatch;
|
|
|
|
results |= profile_gemm<Dispatch, GemmProfiler>(output, "igemm_32x32x128_nt",
|
|
options, config);
|
|
}
|
|
|
|
{
|
|
typedef cutlass::gemm::IgemmTraits<cutlass::MatrixLayout::kRowMajor,
|
|
cutlass::MatrixLayout::kColumnMajor, cutlass::Shape<128, 32, 32>, int,
|
|
cutlass::gemm::LinearScaling<int>, cutlass::Shape<32, 8, 4> > GemmTraits;
|
|
|
|
typedef typename CutlassDispatchBasic<GemmTraits>::Dispatch Dispatch;
|
|
|
|
results |= profile_gemm<Dispatch, GemmProfiler>(output, "igemm_32x32x128_tn",
|
|
options, config);
|
|
}
|
|
|
|
{
|
|
typedef cutlass::gemm::IgemmTraits<cutlass::MatrixLayout::kRowMajor,
|
|
cutlass::MatrixLayout::kRowMajor, cutlass::Shape<128, 32, 32>, int,
|
|
cutlass::gemm::LinearScaling<int>, cutlass::Shape<32, 8, 4> > GemmTraits;
|
|
|
|
typedef typename CutlassDispatchBasic<GemmTraits>::Dispatch Dispatch;
|
|
|
|
results = profile_gemm<Dispatch, GemmProfiler>(output, "igemm_32x32x128_tt",
|
|
options, config);
|
|
}
|
|
|
|
return results;
|
|
}
|
|
|
|
|
|
|
|
struct IgemmRegistrar {
|
|
IgemmRegistrar()
|
|
{
|
|
RegisterGemmProfileFunc(profile_igemm<void>);
|
|
|
|
#ifdef EXHAUSTIVE_PROF
|
|
RegisterGemmProfileFunc(profile_igemm_32x32x128<void>);
|
|
#endif // defined EXHAUSTIVE_PROF
|
|
|
|
}
|
|
};
|
|
|
|
volatile IgemmRegistrar _IgemmRegistrar;
|
|
|
|
////////////////////////////////////////////////////////////////////////////////////////////////////
|
|
|
|
} // namespace perf
|
|
|
|
#endif // if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 610))
|