int main()
{
srand (time(0));
int temp;
for (int i =0;i<SIZE;i++){
temp=rand() % 10000 + 1;
data[i]=temp;
data2[i]=temp;
}
ResetMilli();
bitonic_cpu(data, SIZE);
printf("CPU:%f\n", GetSeconds());
ResetMilli();
#ifdef GPU
bitonic_gpu(data2, SIZE);
printf("GPU:%f\n", GetSeconds());
for (int i=0;i<SIZE;i++)
if (data[i] != data2[i])
{
printf("Error at %d ", i);
return(1);
}
#endif
// Print result
if (SIZE <= MAXPRINTSIZE)
for (int i=0;i<SIZE;i++)
printf("%d ", data[i]);
printf("\nYour sorting looks correct!\n");
}
int main()
{
const int N = 1024;
float *a = new float[N*N];
float *b = new float[N*N];
float *c = new float[N*N];
for (int i = 0; i < N; i++)
for (int j = 0; j < N; j++)
{
a[i+j*N] = 10 + i;
b[i+j*N] = (float)j / N;
}
ResetMilli();
add_matrix(a, b, c, N);
printf("Total time in ms: %d\n", GetMicroseconds());
/* for (int i = 0; i < N; i++) */
/* { */
/* for (int j = 0; j < N; j++) */
/* { */
/* printf("%0.2f ", c[i+j*N]); */
/* } */
/* printf("\n"); */
/* } */
}
////////////////////////////////////////////////////////////////////////////////
// main computation function
////////////////////////////////////////////////////////////////////////////////
void computeImages()
{
//read in full size of memory
image = readppm("maskros512.ppm", &n, &m);
out = (unsigned char*) malloc(n*m*3);
cl_mem in_data, out_data;
cl_int ciErrNum = CL_SUCCESS;
// Create space for data and copy image to device (note that we could also use clEnqueueWriteBuffer to upload)
in_data = clCreateBuffer(cxGPUContext, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR,
3*n*m * sizeof(unsigned char), image, &ciErrNum);
printCLError(ciErrNum,6);
out_data = clCreateBuffer(cxGPUContext, CL_MEM_WRITE_ONLY,
3*n*m * sizeof(unsigned char), NULL, &ciErrNum);
printCLError(ciErrNum,7);
// set the args values
ciErrNum = clSetKernelArg(theKernel, 0, sizeof(cl_mem), (void *) &in_data);
ciErrNum |= clSetKernelArg(theKernel, 1, sizeof(cl_mem), (void *) &out_data);
ciErrNum |= clSetKernelArg(theKernel, 2, sizeof(cl_uint), (void *) &n);
ciErrNum |= clSetKernelArg(theKernel, 3, sizeof(cl_uint), (void *) &m);
printCLError(ciErrNum,8);
// Computing arrangement
//size_t localWorkSize, globalWorkSize;
size_t globalWorkSize[3] = {512, 512, 1};
size_t localWorkSize[3] = {16, 16, 1}; //256 threads in each block
// 32*32 (1024) blocks in total to have 512*512 threads in total
printf("Startup time %lf\n", GetSeconds());
// Compute!
cl_event event;
ResetMilli();
ciErrNum = clEnqueueNDRangeKernel(commandQueue, theKernel, 2, NULL, &globalWorkSize, &localWorkSize, 0, NULL, &event);
printCLError(ciErrNum,9);
ciErrNum = clWaitForEvents(1, &event); // Synch
printCLError(ciErrNum,10);
printf("time %lf\n", GetSeconds());
ciErrNum = clEnqueueReadBuffer(commandQueue, out_data, CL_TRUE, 0, 3*n*m * sizeof(unsigned char), out, 0, NULL, &event);
printCLError(ciErrNum,11);
clWaitForEvents(1, &event); // Synch
printCLError(ciErrNum,10);
clReleaseMemObject(in_data);
clReleaseMemObject(out_data);
return;
}
int main()
{
ResetMilli();
bitonic_cpu(data, SIZE);
printf("%f\n", GetSeconds());
ResetMilli();
bitonic_gpu(data2, SIZE);
printf("%f\n", GetSeconds());
for (int i=0;i<SIZE;i++)
if (data[i] != data2[i])
{
printf("Error at %d ", i);
return(1);
}
// Print result
if (SIZE <= MAXPRINTSIZE)
for (int i=0;i<SIZE;i++)
printf("%d ", data[i]);
printf("\nYour sorting looks correct!\n");
}
// Main program, inits
int main( int argc, char** argv)
{
glutInit(&argc, argv);
glutInitDisplayMode( GLUT_SINGLE | GLUT_RGBA );
glutInitWindowSize( 1024, 512 );
glutCreateWindow("CUDA on live GL");
glutDisplayFunc(Draw);
ResetMilli();
if (init_OpenCL()<0)
{
printf("OpenCL could not be initialized!\n");
close_OpenCL();
return;
}
readAndBuildKernel("filter.cl");
computeImages();
close_OpenCL();
glutMainLoop();
}
