int main(int argc, char ** argv)
{
float * h_raw, * h_out, * outCPU;
cl_mem d_raw, d_out;
try {
if(argc!=2) {
printf("need one parameter here!!!");
exit(-1);
}
_clInit(1, "gpu", 0);
#if defined TIME
double start_time = 0;
double end_time = 0;
string dat_name="data.dat";
FILE * fp = fopen(dat_name.c_str(), "a+");
if(fp==NULL)
{
printf("failed to open file!!!\n");
exit(-1);
}
#endif
int cdim = atoi(argv[1]); //{384};
int rdim = atoi(argv[1]); //{288};
printf("cdim=%d, rdim=%d\n", cdim, rdim);
h_raw = (float *)malloc(cdim * rdim * sizeof(float));
h_out = (float *)malloc(cdim * rdim * sizeof(float));
outCPU = (float *)malloc(cdim * rdim * sizeof(float));
fill<float>(h_raw, cdim * rdim, 5);
d_raw = _clMalloc(cdim * rdim * sizeof(float));
d_out = _clMalloc(cdim * rdim * sizeof(float));
_clMemcpyH2D(d_raw, h_raw, cdim * rdim * sizeof(float));
printf("-0\n");
#ifdef VARIFY
CPURun(h_raw, outCPU, cdim, rdim);
#endif //VARIFY
/**************************1****************************/
#ifdef TIME
start_time = gettime();
#endif
printf("-1.1\n");
broadcast(d_raw, d_out, cdim, rdim);
printf("-1.2\n");
#ifdef TIME
end_time = gettime();
fprintf(fp, "%lf\t", (end_time - start_time));
#endif
#ifdef VARIFY
_clMemcpyD2H(h_out, d_out, cdim * rdim * sizeof(float));
verify_array<float>(outCPU, h_out, cdim * rdim);
#endif //VARIFY
/**************************2****************************/
#ifdef TIME
start_time = gettime();
#endif
broadcast_lm(d_raw, d_out, cdim, rdim);
printf("-2\n");
#ifdef TIME
end_time = gettime();
fprintf(fp, "%lf\t", (end_time - start_time));
#endif
#ifdef VARIFY
_clMemcpyD2H(h_out, d_out, cdim * rdim * sizeof(float));
verify_array<float>(outCPU, h_out, cdim * rdim);
#endif //VARIFY
#ifdef TIME
fprintf(fp, "\n");
fclose(fp);
#endif
}
catch(string msg) {
printf("ERR:%s\n", msg.c_str());
printf("Error catched\n");
}
_clFree(d_raw);
_clFree(d_out);
_clRelease();
if(h_raw!=NULL) free(h_raw);
if(h_out!=NULL) free(h_out);
if(outCPU!=NULL) free(outCPU);
return 1;
}
int main(int argc, char ** argv)
{
uint * in = NULL, * out_cpu = NULL, * out_gpu = NULL;
cl_mem d_in = NULL, d_out = NULL;
try{
if(argc!=2){
printf("need 1 parameter here!!!");
exit(-1);
}
_clInit(1, "gpu", 0);
uint iter = 100;
#if defined TIME
double start_time = 0.0;
double end_time = 0.0;
double deltaT = 0.0;
string dat_name="data.dat";
FILE * fp = fopen(dat_name.c_str(), "a+");
if(fp==NULL)
{
printf("failed to open file!!!\n");
exit(-1);
}
#endif
// parameters
uint side = atoi(argv[1]);
uint wData = side;
uint hData = side;
uint size = wData * hData;
printf("wData=%d, hData=%d\n", wData, hData);
// allocate memory space on the host and device side
in = (uint * )malloc(size * sizeof(uint));
out_cpu = (uint * )malloc(size * sizeof(uint));
out_gpu = (uint * )malloc(size * sizeof(uint));
d_in = _clMalloc(size * sizeof(uint));
d_out = _clMalloc(size * sizeof(uint));
// initialization
fill<uint>(in, size, 16);
// copy data from host to device
_clMemcpyH2D(d_in, in, size * sizeof(uint));
// warm-up
mt_1(d_in, d_out, wData, hData);
mt_2(d_in, d_out, wData, hData);
mt_3(d_in, d_out, wData, hData);
#ifdef VARIFY
CPURun(in, out_cpu, wData, hData);
#endif //VARIFY
/**************************1****************************/
#ifdef TIME
deltaT = 0.0;
#endif
for(int i=0; i<iter; i++)
{
#ifdef TIME
start_time = gettime();
#endif
mt_1(d_in, d_out, wData, hData);
#ifdef TIME
end_time = gettime();
deltaT += end_time - start_time;
#endif
}
#ifdef TIME
fprintf(fp, "%lf\t", deltaT/(double)iter);
#endif
#ifdef VARIFY
_clMemcpyD2H(out_gpu, d_out, size * sizeof(uint));
verify_array_int<uint>(out_cpu, out_gpu, size);
#endif //VARIFY
/**************************2****************************/
#ifdef TIME
deltaT = 0.0;
#endif
for(int i=0; i<iter; i++)
{
#ifdef TIME
start_time = gettime();
#endif
mt_2(d_in, d_out, wData, hData);
#ifdef TIME
end_time = gettime();
deltaT += end_time - start_time;
#endif
}
#ifdef TIME
fprintf(fp, "%lf\t", deltaT/(double)iter);
#endif
#ifdef VARIFY
_clMemcpyD2H(out_gpu, d_out, size * sizeof(uint));
verify_array_int<uint>(out_cpu, out_gpu, size);
#endif //VARIFY
/**************************3****************************/
#ifdef TIME
deltaT = 0.0;
#endif
for(int i=0; i<iter; i++)
{
#ifdef TIME
start_time = gettime();
#endif
mt_3(d_in, d_out, wData, hData);
#ifdef TIME
end_time = gettime();
deltaT += end_time - start_time;
#endif
}
#ifdef TIME
fprintf(fp, "%lf\t", deltaT/(double)iter);
#endif
#ifdef VARIFY
_clMemcpyD2H(out_gpu, d_out, size * sizeof(uint));
verify_array_int<uint>(out_cpu, out_gpu, size);
#endif //VARIFY
#ifdef TIME
fprintf(fp, "\n");
fclose(fp);
#endif
}
catch(string msg){
printf("ERR:%s\n", msg.c_str());
printf("Error catched\n");
}
_clFree(d_in);
_clFree(d_out);
_clRelease();
if(in!=NULL) free(in);
if(out_cpu!=NULL) free(out_cpu);
if(out_gpu!=NULL) free(out_gpu);
return 1;
}
int main(int argc, char *argv[]) {
//printf("working");
if (argc == 9) {
process_count = atoi(argv[2]);
kproc_count = atoi(argv[4]);
ioproc_count = atoi(argv[6]);
pcproc_count = atoi(argv[8]);
calcproc_count = process_count - (kproc_count + ioproc_count + (pcproc_count * 2));
} else {
process_count = 15;
kproc_count = 3;
ioproc_count = 3;
pcproc_count = 3;
calcproc_count = 3;
}
ReadyQPtr = QueueConstructor();
int i;
for (i = 0; i < kproc_count; i++) {
PCBPtr a = PCBConstructor(processidcount, 2, 2, -1);
enqueue(ReadyQPtr, a);
processidcount++;
}
for (i = 0; i < pcproc_count; i++) {
PCBPtr c = PCBConstructor(processidcount, 3, 2, i);
enqueue(ReadyQPtr, c);
processidcount++;
PCBPtr d = PCBConstructor(processidcount, 4, 2, i);
enqueue(ReadyQPtr, d);
processidcount++;
}
for (i = 0; i < calcproc_count; i++) {
PCBPtr e = PCBConstructor(processidcount, 0, 2, -1);
enqueue(ReadyQPtr, e);
processidcount++;
}
for (i = 0; i < ioproc_count; i++) {
PCBPtr b = PCBConstructor(processidcount, 1, 2, -1);
enqueue(ReadyQPtr, b);
processidcount++;
}
for (i = 0; i < pcproc_count; i++) {
sharedMemory[i] = 0;
}
//initialize interrupt flags
IO1INT = 0;
IO2INT = 0;
KBINT = 0;
TIMERINT = 0;
KBHASPCB = 0;
//printf("\nProcess Count %d", process_count);
//printQueue(ReadyQPtr);
//printf("\nProcess Count2");
cpuPtr cpu = cpuConstructor();
//printf("\nCPU Program Complete");
CPURun(cpu);
return 0;
}