forked from NVIDIA/CUDALibrarySamples
-
Notifications
You must be signed in to change notification settings - Fork 0
/
spmm_coo_batched_example.c
232 lines (223 loc) · 11.5 KB
/
spmm_coo_batched_example.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
/*
* Copyright 1993-2022 NVIDIA Corporation. All rights reserved.
*
* NOTICE TO LICENSEE:
*
* This source code and/or documentation ("Licensed Deliverables") are
* subject to NVIDIA intellectual property rights under U.S. and
* international Copyright laws.
*
* These Licensed Deliverables contained herein is PROPRIETARY and
* CONFIDENTIAL to NVIDIA and is being provided under the terms and
* conditions of a form of NVIDIA software license agreement by and
* between NVIDIA and Licensee ("License Agreement") or electronically
* accepted by Licensee. Notwithstanding any terms or conditions to
* the contrary in the License Agreement, reproduction or disclosure
* of the Licensed Deliverables to any third party without the express
* written consent of NVIDIA is prohibited.
*
* NOTWITHSTANDING ANY TERMS OR CONDITIONS TO THE CONTRARY IN THE
* LICENSE AGREEMENT, NVIDIA MAKES NO REPRESENTATION ABOUT THE
* SUITABILITY OF THESE LICENSED DELIVERABLES FOR ANY PURPOSE. IT IS
* PROVIDED "AS IS" WITHOUT EXPRESS OR IMPLIED WARRANTY OF ANY KIND.
* NVIDIA DISCLAIMS ALL WARRANTIES WITH REGARD TO THESE LICENSED
* DELIVERABLES, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY,
* NONINFRINGEMENT, AND FITNESS FOR A PARTICULAR PURPOSE.
* NOTWITHSTANDING ANY TERMS OR CONDITIONS TO THE CONTRARY IN THE
* LICENSE AGREEMENT, IN NO EVENT SHALL NVIDIA BE LIABLE FOR ANY
* SPECIAL, INDIRECT, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, OR ANY
* DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS,
* WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
* ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE
* OF THESE LICENSED DELIVERABLES.
*
* U.S. Government End Users. These Licensed Deliverables are a
* "commercial item" as that term is defined at 48 C.F.R. 2.101 (OCT
* 1995), consisting of "commercial computer software" and "commercial
* computer software documentation" as such terms are used in 48
* C.F.R. 12.212 (SEPT 1995) and is provided to the U.S. Government
* only as a commercial end item. Consistent with 48 C.F.R.12.212 and
* 48 C.F.R. 227.7202-1 through 227.7202-4 (JUNE 1995), all
* U.S. Government End Users acquire the Licensed Deliverables with
* only those rights set forth herein.
*
* Any use of the Licensed Deliverables in individual and commercial
* software must include, in the user documentation and internal
* comments to the code, the above Disclaimer and U.S. Government End
* Users Notice.
*/
#include <cuda_runtime_api.h> // cudaMalloc, cudaMemcpy, etc.
#include <cusparse.h> // cusparseSpMM
#include <stdio.h> // printf
#include <stdlib.h> // EXIT_FAILURE
#define CHECK_CUDA(func) \
{ \
cudaError_t status = (func); \
if (status != cudaSuccess) { \
printf("CUDA API failed at line %d with error: %s (%d)\n", \
__LINE__, cudaGetErrorString(status), status); \
return EXIT_FAILURE; \
} \
}
#define CHECK_CUSPARSE(func) \
{ \
cusparseStatus_t status = (func); \
if (status != CUSPARSE_STATUS_SUCCESS) { \
printf("CUSPARSE API failed at line %d with error: %s (%d)\n", \
__LINE__, cusparseGetErrorString(status), status); \
return EXIT_FAILURE; \
} \
}
int main(void) {
// Host problem definition
int A_num_rows = 4;
int A_num_cols = 4;
int A_nnz = 9;
int B_num_rows = A_num_cols;
int B_num_cols = 3;
int ldb = B_num_rows;
int ldc = A_num_rows;
int B_size = ldb * B_num_cols;
int C_size = ldc * B_num_cols;
int num_batches = 2;
int hA_rows[] = { 0, 0, 0, 1, 2, 2, 2, 3, 3 };
int hA_columns1[] = { 0, 2, 3, 1, 0, 2, 3, 1, 3 };
int hA_columns2[] = { 1, 2, 3, 0, 0, 1, 3, 1, 2 };
float hA_values1[] = { /*0*/ 1.0f, 2.0f, 3.0f,
4.0f, /*0*/ /*0*/ /*0*/
5.0f, /*0*/ 6.0f, 7.0f,
/*0*/ 8.0f, /*0*/ 9.0f };
float hA_values2[] = { /*0*/ 10.0f, 11.0f, 12.0f,
13.0f, /*0*/ /*0*/ /*0*/
14.0f, 15.0f, /*0*/ 16.0f,
/*0*/ 17.0f, 18.0f /*0*/ };
float hB1[] = { 1.0f, 2.0f, 3.0f, 4.0f,
5.0f, 6.0f, 7.0f, 8.0f,
9.0f, 10.0f, 11.0f, 12.0f };
float hB2[] = { 6.0f, 4.0f, 3.0f, 2.0f,
1.0f, 6.0f, 9.0f, 8.0f,
9.0f, 3.0f, 2.0f, 5.0f };
float hC1[] = { 0.0f, 0.0f, 0.0f, 0.0f,
0.0f, 0.0f, 0.0f, 0.0f,
0.0f, 0.0f, 0.0f, 0.0f };
float hC2[] = { 0.0f, 0.0f, 0.0f, 0.0f,
0.0f, 0.0f, 0.0f, 0.0f,
0.0f, 0.0f, 0.0f, 0.0f };
float hC1_result[] = { 19.0f, 8.0f, 51.0f, 52.0f,
43.0f, 24.0f, 123.0f, 120.0f,
67.0f, 40.0f, 195.0f, 188.0f };
float hC2_result[] = { 97.0f, 78.0f, 176.0f, 122.0f,
255.0f, 13.0f, 232.0f, 264.0f,
112.0f, 117.0f, 251.0f, 87.0f };
float alpha = 1.0f;
float beta = 0.0f;
//--------------------------------------------------------------------------
// Device memory management
int *dA_rows, *dA_columns;
float *dA_values, *dB, *dC;
CHECK_CUDA( cudaMalloc((void**) &dA_rows,
A_nnz * num_batches * sizeof(int)) )
CHECK_CUDA( cudaMalloc((void**) &dA_columns,
A_nnz * num_batches * sizeof(int)) )
CHECK_CUDA( cudaMalloc((void**) &dA_values,
A_nnz * num_batches * sizeof(float)) )
CHECK_CUDA( cudaMalloc((void**) &dB,
B_size * num_batches * sizeof(float)) )
CHECK_CUDA( cudaMalloc((void**) &dC,
C_size * num_batches * sizeof(float)) )
CHECK_CUDA( cudaMemcpy(dA_rows, hA_rows, A_nnz * sizeof(int),
cudaMemcpyHostToDevice) )
CHECK_CUDA( cudaMemcpy(dA_rows + A_nnz, hA_rows, A_nnz * sizeof(int),
cudaMemcpyHostToDevice) )
CHECK_CUDA( cudaMemcpy(dA_columns, hA_columns1, A_nnz * sizeof(int),
cudaMemcpyHostToDevice) )
CHECK_CUDA( cudaMemcpy(dA_columns + A_nnz, hA_columns2, A_nnz * sizeof(int),
cudaMemcpyHostToDevice) )
CHECK_CUDA( cudaMemcpy(dA_values, hA_values1, A_nnz * sizeof(float),
cudaMemcpyHostToDevice) )
CHECK_CUDA( cudaMemcpy(dA_values + A_nnz, hA_values2, A_nnz * sizeof(float),
cudaMemcpyHostToDevice) )
CHECK_CUDA( cudaMemcpy(dB, hB1, B_size * sizeof(float),
cudaMemcpyHostToDevice) )
CHECK_CUDA( cudaMemcpy(dB + B_size, hB2, B_size * sizeof(float),
cudaMemcpyHostToDevice) )
CHECK_CUDA( cudaMemcpy(dC, hC1, C_size * sizeof(float),
cudaMemcpyHostToDevice) )
CHECK_CUDA( cudaMemcpy(dC + C_size, hC2, C_size * sizeof(float),
cudaMemcpyHostToDevice) )
//--------------------------------------------------------------------------
// CUSPARSE APIs
cusparseHandle_t handle = NULL;
cusparseSpMatDescr_t matA;
cusparseDnMatDescr_t matB, matC;
void* dBuffer = NULL;
size_t bufferSize = 0;
CHECK_CUSPARSE( cusparseCreate(&handle) )
// Create sparse matrix A in COO format
CHECK_CUSPARSE( cusparseCreateCoo(&matA, A_num_rows, A_num_cols, A_nnz,
dA_rows, dA_columns, dA_values,
CUSPARSE_INDEX_32I,
CUSPARSE_INDEX_BASE_ZERO, CUDA_R_32F) )
CHECK_CUSPARSE( cusparseCooSetStridedBatch(matA, num_batches, A_nnz) )
// Alternatively, the following code can be used for matA broadcast
// CHECK_CUSPARSE( cusparseCooSetStridedBatch(matA, num_batches, 0) )
// Create dense matrix B
CHECK_CUSPARSE( cusparseCreateDnMat(&matB, B_num_rows, B_num_cols, ldb, dB,
CUDA_R_32F, CUSPARSE_ORDER_COL) )
CHECK_CUSPARSE( cusparseDnMatSetStridedBatch(matB, num_batches, B_size) )
// Create dense matrix C
CHECK_CUSPARSE( cusparseCreateDnMat(&matC, A_num_rows, B_num_cols, ldc, dC,
CUDA_R_32F, CUSPARSE_ORDER_COL) )
CHECK_CUSPARSE( cusparseDnMatSetStridedBatch(matC, num_batches, C_size) )
// allocate an external buffer if needed
CHECK_CUSPARSE( cusparseSpMM_bufferSize(
handle,
CUSPARSE_OPERATION_NON_TRANSPOSE,
CUSPARSE_OPERATION_NON_TRANSPOSE,
&alpha, matA, matB, &beta, matC, CUDA_R_32F,
CUSPARSE_SPMM_COO_ALG4, &bufferSize) )
CHECK_CUDA( cudaMalloc(&dBuffer, bufferSize) )
// execute SpMM
CHECK_CUSPARSE( cusparseSpMM(handle,
CUSPARSE_OPERATION_NON_TRANSPOSE,
CUSPARSE_OPERATION_NON_TRANSPOSE,
&alpha, matA, matB, &beta, matC, CUDA_R_32F,
CUSPARSE_SPMM_COO_ALG4, dBuffer) )
// destroy matrix/vector descriptors
CHECK_CUSPARSE( cusparseDestroySpMat(matA) )
CHECK_CUSPARSE( cusparseDestroyDnMat(matB) )
CHECK_CUSPARSE( cusparseDestroyDnMat(matC) )
CHECK_CUSPARSE( cusparseDestroy(handle) )
//--------------------------------------------------------------------------
// device result check
CHECK_CUDA( cudaMemcpy(hC1, dC, C_size * sizeof(float),
cudaMemcpyDeviceToHost) )
CHECK_CUDA( cudaMemcpy(hC2, dC + C_size, C_size * sizeof(float),
cudaMemcpyDeviceToHost) )
int correct = 1;
for (int i = 0; i < A_num_rows; i++) {
for (int j = 0; j < B_num_cols; j++) {
if (hC1[i + j * ldc] != hC1_result[i + j * ldc]) {
correct = 0; // direct floating point comparison is not reliable
break;
}
if (hC2[i + j * ldc] != hC2_result[i + j * ldc]) {
correct = 0;
break;
}
}
}
if (correct)
printf("spmm_coo_batched_example test PASSED\n");
else
printf("spmm_coo_batched_example test FAILED: wrong result\n");
//--------------------------------------------------------------------------
// device memory deallocation
CHECK_CUDA( cudaFree(dBuffer) )
CHECK_CUDA( cudaFree(dA_rows) )
CHECK_CUDA( cudaFree(dA_columns) )
CHECK_CUDA( cudaFree(dA_values) )
CHECK_CUDA( cudaFree(dB) )
CHECK_CUDA( cudaFree(dC) )
return EXIT_SUCCESS;
}