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
|
/* { dg-do run { target openacc_nvidia_accel_selected } } */
/* { dg-additional-options "-lcuda -lcublas -lcudart" } */
#include <stdio.h>
#include <stdlib.h>
#include <cuda.h>
#include <cuda_runtime_api.h>
#include <cublas_v2.h>
#include <openacc.h>
void
saxpy (int n, float a, float *x, float *y)
{
int i;
for (i = 0; i < n; i++)
{
y[i] = a * x[i] + y[i];
}
}
void
context_check (CUcontext ctx1)
{
CUcontext ctx2, ctx3;
CUresult r;
r = cuCtxGetCurrent (&ctx2);
if (r != CUDA_SUCCESS)
{
fprintf (stderr, "cuCtxGetCurrent failed: %d\n", r);
exit (EXIT_FAILURE);
}
if (ctx1 != ctx2)
{
fprintf (stderr, "new context established\n");
exit (EXIT_FAILURE);
}
ctx3 = (CUcontext) acc_get_current_cuda_context ();
if (ctx1 != ctx3)
{
fprintf (stderr, "acc_get_current_cuda_context returned wrong value\n");
exit (EXIT_FAILURE);
}
return;
}
int
main (int argc, char **argv)
{
cublasStatus_t s;
cublasHandle_t h;
CUcontext pctx;
CUresult r;
int i;
const int N = 256;
float *h_X, *h_Y1, *h_Y2;
float *d_X,*d_Y;
float alpha = 2.0f;
float error_norm;
float ref_norm;
/* Test 4 - OpenACC creates, cuBLAS shares. */
acc_set_device_num (0, acc_device_nvidia);
r = cuCtxGetCurrent (&pctx);
if (r != CUDA_SUCCESS)
{
fprintf (stderr, "cuCtxGetCurrent failed: %d\n", r);
exit (EXIT_FAILURE);
}
h_X = (float *) malloc (N * sizeof (float));
if (h_X == 0)
{
fprintf (stderr, "malloc failed: for h_X\n");
exit (EXIT_FAILURE);
}
h_Y1 = (float *) malloc (N * sizeof (float));
if (h_Y1 == 0)
{
fprintf (stderr, "malloc failed: for h_Y1\n");
exit (EXIT_FAILURE);
}
h_Y2 = (float *) malloc (N * sizeof (float));
if (h_Y2 == 0)
{
fprintf (stderr, "malloc failed: for h_Y2\n");
exit (EXIT_FAILURE);
}
for (i = 0; i < N; i++)
{
h_X[i] = rand () / (float) RAND_MAX;
h_Y2[i] = h_Y1[i] = rand () / (float) RAND_MAX;
}
#pragma acc parallel copyin (h_X[0:N]), copy (h_Y2[0:N]) copy (alpha)
{
int i;
for (i = 0; i < N; i++)
{
h_Y2[i] = alpha * h_X[i] + h_Y2[i];
}
}
r = cuCtxGetCurrent (&pctx);
if (r != CUDA_SUCCESS)
{
fprintf (stderr, "cuCtxGetCurrent failed: %d\n", r);
exit (EXIT_FAILURE);
}
d_X = (float *) acc_copyin (&h_X[0], N * sizeof (float));
if (d_X == NULL)
{
fprintf (stderr, "copyin error h_Y1\n");
exit (EXIT_FAILURE);
}
d_Y = (float *) acc_copyin (&h_Y1[0], N * sizeof (float));
if (d_Y == NULL)
{
fprintf (stderr, "copyin error h_Y1\n");
exit (EXIT_FAILURE);
}
s = cublasCreate (&h);
if (s != CUBLAS_STATUS_SUCCESS)
{
fprintf (stderr, "cublasCreate failed: %d\n", s);
exit (EXIT_FAILURE);
}
context_check (pctx);
s = cublasSaxpy (h, N, &alpha, d_X, 1, d_Y, 1);
if (s != CUBLAS_STATUS_SUCCESS)
{
fprintf (stderr, "cublasSaxpy failed: %d\n", s);
exit (EXIT_FAILURE);
}
context_check (pctx);
acc_memcpy_from_device (&h_Y1[0], d_Y, N * sizeof (float));
context_check (pctx);
error_norm = 0;
ref_norm = 0;
for (i = 0; i < N; ++i)
{
float diff;
diff = h_Y1[i] - h_Y2[i];
error_norm += diff * diff;
ref_norm += h_Y2[i] * h_Y2[i];
}
error_norm = (float) sqrt ((double) error_norm);
ref_norm = (float) sqrt ((double) ref_norm);
if ((fabs (ref_norm) < 1e-7) || ((error_norm / ref_norm) >= 1e-6f))
{
fprintf (stderr, "math error\n");
exit (EXIT_FAILURE);
}
free (h_X);
free (h_Y1);
free (h_Y2);
acc_free (d_X);
acc_free (d_Y);
context_check (pctx);
s = cublasDestroy (h);
if (s != CUBLAS_STATUS_SUCCESS)
{
fprintf (stderr, "cublasDestroy failed: %d\n", s);
exit (EXIT_FAILURE);
}
context_check (pctx);
acc_shutdown (acc_device_nvidia);
r = cuCtxGetCurrent (&pctx);
if (r != CUDA_SUCCESS)
{
fprintf (stderr, "cuCtxGetCurrent failed: %d\n", r);
exit (EXIT_FAILURE);
}
if (pctx)
{
fprintf (stderr, "Unexpected context\n");
exit (EXIT_FAILURE);
}
return EXIT_SUCCESS;
}
|
