-
Notifications
You must be signed in to change notification settings - Fork 11
/
utils.cu
169 lines (141 loc) · 4.99 KB
/
utils.cu
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
#include "utils.h"
#include <stdio.h>
// Complement & Compare technique, see
// http://www.exploringbinary.com/ten-ways-to-check-if-an-integer-is-a-power-of-two-in-c/
int isPowerOfTwo (unsigned int x)
{
return ((x != 0) && ((x & (~x + 1)) == x));
}
// Print device properties
void printDevProp(cudaDeviceProp devProp)
{
printf("Major revision number: %d\n", devProp.major);
printf("Minor revision number: %d\n", devProp.minor);
printf("Name: %s\n", devProp.name);
printf("Total global memory: %zu\n", devProp.totalGlobalMem);
printf("Total shared memory per block: %zu\n", devProp.sharedMemPerBlock);
printf("Total registers per block: %d\n", devProp.regsPerBlock);
printf("Warp size: %d\n", devProp.warpSize);
printf("Maximum memory pitch: %zu\n", devProp.memPitch);
printf("Maximum threads per block: %d\n", devProp.maxThreadsPerBlock);
for (int i = 0; i < 3; ++i)
printf("Maximum dimension %d of block: %d\n", i, devProp.maxThreadsDim[i]);
for (int i = 0; i < 3; ++i)
printf("Maximum dimension %d of grid: %d\n", i, devProp.maxGridSize[i]);
printf("Clock rate: %d\n", devProp.clockRate);
printf("Total constant memory: %zu\n", devProp.totalConstMem);
printf("Texture alignment: %zu\n", devProp.textureAlignment);
printf("Concurrent copy and execution: %s\n", (devProp.deviceOverlap ? "Yes" : "No"));
printf("Number of multiprocessors: %d\n", devProp.multiProcessorCount);
printf("Kernel execution timeout: %s\n", (devProp.kernelExecTimeoutEnabled ? "Yes" : "No"));
return;
}
void printAllDevices(bool debug)
{
// Number of CUDA devices
int devCount;
cudaGetDeviceCount(&devCount);
if (devCount < 1) {
fprintf(stderr, "Error: no CUDA devices?\n");
}
if (debug) {
printf("CUDA Device Query...\n");
printf("There are %d CUDA devices.\n", devCount);
}
// For some reason, systems that don't have CUDA devices might
// print infinitely many of them. If we run the program accidently,
// the program might hang while printing. We don't want that.
if (devCount > 5)
printf("Printing first 5 devices.\n");
// Iterate through devices
for (int i = 0; i < min(5, devCount); ++i)
{
// Get device properties
if (debug)
printf("\nCUDA Device #%d\n", i);
cudaDeviceProp devProp;
cudaGetDeviceProperties(&devProp, i);
if (debug)
printDevProp(devProp);
}
}
__global__
void hello(char *a, int *b)
{
int idx = blockIdx.x * blockDim.x + threadIdx.x;
a[idx] += b[idx];
}
// Prints Hello, World if the GPU code is working right.
void helloWorld()
{
const int N = 16;
const int blocksize = 16;
char a[N] = "Hello \0\0\0\0\0\0";
int b[N] = {15, 10, 6, 0, -11, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
char *ad;
int *bd;
const int csize = N*sizeof(char);
const int isize = N*sizeof(int);
checkCuda(cudaMalloc( (void**)&ad, csize ));
checkCuda(cudaMalloc( (void**)&bd, isize ));
checkCuda(cudaMemcpy( ad, a, csize, cudaMemcpyHostToDevice ));
checkCuda(cudaMemcpy( bd, b, isize, cudaMemcpyHostToDevice ));
dim3 dimBlock( blocksize, 1 );
dim3 dimGrid( 1, 1 );
hello<<<dimGrid, dimBlock>>>(ad, bd);
checkCuda(cudaMemcpy( a, ad, csize, cudaMemcpyDeviceToHost ));
checkCuda(cudaFree( ad ));
checkCuda(cudaFree( bd ));
if (strcmp(a, "World!")) {
fprintf(stderr, "Error: Expected \"World!\", got \"%s\"\n", a);
exit(-1);
}
}
// Run a couple tests to see that CUDA works properly.
void cudaCheck(bool debug)
{
if (debug) {
printf("Calling cudaFree(0) no-op...\n");
}
checkCuda(cudaFree(0));
if (debug) {
printf("Calling cudaFree(0) succeeded!\n");
}
printAllDevices(debug);
helloWorld();
}
/* Slow FFT for testing if needed.
vector<complex> dft(vector<double>& in)
{
vector<complex> out(in.size());
for (int k = 0; k < out.size(); k++) {
out[k] = makeComplex(0, 0);
for (int n = 0; n < in.size(); n++) {
complex exponent = makeComplex(0, -2.0f * M_PI * k * n / in.size());
out[k] = cuCadd(out[k], cuComplexScalarMult(in[n], cuComplexExponential(exponent)));
}
}
return out;
}
vector<complex> idft_complex(vector<complex>& in)
{
vector<complex> out(in.size());
for (int k = 0; k < out.size(); k++) {
out[k] = makeComplex(0, 0);
for (int n = 0; n < in.size(); n++) {
complex exponent = makeComplex(0, 2.0f * M_PI * k * n / in.size());
out[k] = cuCadd(out[k], cuCmul(in[n], cuComplexExponential(exponent)));
}
}
return out;
}
vector<double> idft(vector<complex>& in)
{
vector<complex> ift = idft_complex(in);
vector<double> out(ift.size());
for (int i = 0; i < ift.size(); i++) {
out[i] = cuCreal(ift[i]);
}
return out;
}
*/