Cutlass
CUDA Templates for Linear Algebra Subroutines and Solvers
device_gemm.h
Go to the documentation of this file.
1 /***************************************************************************************************
2 * Copyright (c) 2017-2018, NVIDIA CORPORATION. All rights reserved.
3 *
4 * Redistribution and use in source and binary forms, with or without modification, are permitted
5 * provided that the following conditions are met:
6 * * Redistributions of source code must retain the above copyright notice, this list of
7 * conditions and the following disclaimer.
8 * * Redistributions in binary form must reproduce the above copyright notice, this list of
9 * conditions and the following disclaimer in the documentation and/or other materials
10 * provided with the distribution.
11 * * Neither the name of the NVIDIA CORPORATION nor the names of its contributors may be used
12 * to endorse or promote products derived from this software without specific prior written
13 * permission.
14 *
15 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR
16 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
17 * FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
19 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
20 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
21 * STRICT LIABILITY, OR TOR (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
22 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
23 *
24 **************************************************************************************************/
28 #pragma once
29 
30 #if !defined(__CUDACC_RTC__)
31 #include <cuda.h>
32 #endif
33 
34 #include "cutlass/coord.h"
35 #include "cutlass/util/platform.h"
36 namespace cutlass {
37 namespace gemm {
38 
39 template<typename DeviceGemmTraits_ >
40 struct DeviceGemm {
42  typedef DeviceGemmTraits_ Traits;
44  typedef typename Traits::Params Params;
45 
47 #if !defined(__CUDACC_RTC__)
48  static __host__ cudaError_t launch(Params const& params) {
50  Traits::GemmTraits::KernelClass::launch(params.GemmParams);
51  cudaError_t err = cudaGetLastError();
52  if (err != cudaSuccess)
53  return err;
54  Traits::ReductionTraits::KernelClass::launch(params.ReductionParams);
55  return cudaGetLastError();
56  }
57 #endif
58 
62 
64  CUTLASS_DEVICE DeviceGemm() {}
65 };
66 } // namespace device_gemm
67 } // namespace cutalss
DeviceGemmTraits_ Traits
The Traits.
Definition: device_gemm.h:42
Definition: convert.h:33
A Coord is a coordinate of arbitrary rank into a tensor or matrix.
Definition: device_gemm.h:40
Traits::Params Params
Use the params object defined in traits.
Definition: device_gemm.h:44
C++ features that may be otherwise unimplemented for CUDA device functions.
CUTLASS_DEVICE DeviceGemm()
Ctor.
Definition: device_gemm.h:64
static __host__ cudaError_t launch(Params const &params)
Support for NVRTC.
Definition: device_gemm.h:49