-
Notifications
You must be signed in to change notification settings - Fork 0
/
hysub.cu
98 lines (89 loc) · 2.64 KB
/
hysub.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
#include <stdio.h>
#include <stdlib.h>
#include <cuda.h>
void checkCUDAError(const char *msg);
__global__ void Kernel(int *dat);
extern "C" void cumain(int myid, int gx, int gy,int bx, int by, int bz) {
int *dat_local, *dat_remote;
//int gx,gy;
//int bx,by,bz;
int size;
int numthreads,j;
int mydev;
cudaSetDevice(myid);
cudaGetDevice(&mydev);
dim3 dimGrid(gx,gy);
dim3 dimBlock(bx,by,bz);
numthreads=gx*gy*bx*by*bz;
size=6*sizeof(int)*numthreads;
cudaMalloc((void**) &dat_remote, size);
checkCUDAError("cudaMalloc");
dat_local=(int*)malloc(size);
Kernel<<<dimGrid,dimBlock>>>(dat_remote);
checkCUDAError("Kernel");
cudaMemcpy(dat_local, dat_remote, size,cudaMemcpyDeviceToHost);
checkCUDAError("copy");
for(int i=0;i<numthreads;i++) {
j=i*6;
printf("%4.4d %2.2d %6d %3d %3d %3d %3d %3d\n",myid,mydev,
dat_local[j],
dat_local[j+1],dat_local[j+2],
dat_local[j+3],dat_local[j+4],dat_local[j+5]);
}
}
// To run at normal speed define SLOW as blank.
// The program should return the same results independent of
// the setting for kmax and jmax.
// Set jmax to a larger value to slow it down more.
#ifndef SLOW
#define SLOW slow
#endif
int __device__ slow(int input){
int i;
int jmax,kmax;
jmax=5;
kmax=100000;
for (int j=1; j <=jmax ; j++) {
i=j;
if(j == jmax)i=input;
for (int k=1; k< kmax; k++) {
i= int(i*(1.00001*(sin((double)i)*sin((double)i)+cos((double)i)*cos((double)i))));
}
}
return(i);
}
__global__ void Kernel(int *dat) {
/* get my block within a grid */
int myblock=blockIdx.x+blockIdx.y*gridDim.x;
/* how big is each block within a grid */
int blocksize=blockDim.x*blockDim.y*blockDim.z;
/* get thread within a block */
int subthread=threadIdx.z*(blockDim.x*blockDim.y)+threadIdx.y*blockDim.x+threadIdx.x;
/* find my thread */
int thread=myblock*blocksize+subthread;
#if __DEVICE_EMULATION__
printf("gridDim=(%3d %3d) blockIdx=(%3d %3d) blockDim=(%3d %3d %3d) threadIdx=(%3d %3d %3d) %6d\n",
gridDim.x,gridDim.y,
blockIdx.x,blockIdx.y,
blockDim.x,blockDim.y,blockDim.z,
threadIdx.x,threadIdx.y,threadIdx.z,thread);
#endif
/* starting index into array */
int index=thread*6;
dat[index]=SLOW(thread);
dat[index+1]=SLOW(blockIdx.x);
dat[index+2]=SLOW(blockIdx.y);
dat[index+3]=SLOW(threadIdx.x);
dat[index+4]=SLOW(threadIdx.y);
dat[index+5]=SLOW(threadIdx.z);
}
void checkCUDAError(const char *msg)
{
cudaError_t err = cudaGetLastError();
if( cudaSuccess != err)
{
fprintf(stderr, "Cuda error: %s: %s.\n", msg,
cudaGetErrorString( err) );
exit(EXIT_FAILURE);
}
}