google-research

1
// Copyright 2024 The Google Research Authors.
2
//
3
// Licensed under the Apache License, Version 2.0 (the "License");
4
// you may not use this file except in compliance with the License.
5
// You may obtain a copy of the License at
6
//
7
//     http://www.apache.org/licenses/LICENSE-2.0
8
//
9
// Unless required by applicable law or agreed to in writing, software
10
// distributed under the License is distributed on an "AS IS" BASIS,
11
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
// See the License for the specific language governing permissions and
13
// limitations under the License.
14

15
#if GOOGLE_CUDA
16
#define EIGEN_USE_GPU  // For definition of Eigen::GpuDevice.
17
#include "sparse/ops/cc/common.h"
18
#include "sparse/ops/cc/spmm_launcher.h"
19
#include "sputnik/sputnik.h"
20

21
namespace sgk {
22

23
// CUDA kernel launcher.
24
void LaunchSpmm(const Eigen::GpuDevice &d, int m, int k, int n, int nonzeros,
25
                const float *values, const int *row_indices,
26
                const int *row_offsets, const int *column_indices,
27
                const float *dense_matrix, const float *bias,
28
                float *output_matrix) {
29
  // TODO(tgale): There should be a TensorFlow approach to checking
30
  // cudaError_t objects correctly. Switch to this, whatever it is.
31
  //
32
  // NOTE: Passing nullptr as bias will execute the standard spmm w/
33
  // no bias or relu.
34
  CUDA_CALL(sputnik::CudaSpmmBiasRelu(m, k, n, nonzeros, row_indices, values,
35
                                      row_offsets, column_indices, dense_matrix,
36
                                      bias, output_matrix, d.stream()));
37
}
38

39
}  // namespace sgk
40

41
#endif  // GOOGLE_CUDA
42