int main() {
SIM_SLEEP_OFF();
// some variables
int corenum = RigelGetCoreNum();
int threadnum = RigelGetThreadNum();
int numcores = RigelGetNumCores();
int numthreads = RigelGetNumThreads();
int th_per_core = RigelGetNumThreadsPerCore();
int th_per_cluster = RigelGetNumThreadsPerCluster();
int rows_per_thread = SIZE / numthreads; // statically divide work up
int start = rows_per_thread * threadnum;
int end = start + rows_per_thread;
int i=0, j=0;
// setup on thread 0
if(threadnum == 0) {
BARRIER_INIT(&bi);
// init values
for(i=0;i<SIZE;i++){
for(j=0;j<SIZE;j++){
MATRIX[j][i] = j;
}
VECTOR[i] = i;
RESULT[i] = 0xbeef;
}
//printf("rpf:%d numthreads:%d\n", rows_per_thread, numthreads);
ClearTimer(0);
// get started
StartTimer(0);
atomic_flag_set(&Init_Flag);
}
// wait for core 0 to finish setup
atomic_flag_spin_until_set(&Init_Flag);
// do the work
for( i=start; i<end; i++ ) {
RESULT[i] = MVM(i);
}
BARRIER_ENTER(&bi);
// cleanup on thread 0
if(threadnum == 0) {
StopTimer(0);
// print results
for(i=0; i<SIZE; i++) {
RigelPrint(RESULT[i]);
}
}
return 1;
}
void __rigel_premain(void)
{
int tid = RigelGetThreadNum();
if(tid == 0) {
SIM_SLEEP_ON();
SIM_SLEEP_OFF();
_pt[tid].funcptr = (void *)0xFFFFFFFF; //Make sure we don't get a thread spawned on us by accident
}
else {
RigelPrint(0xFEDC);
while(1) {
while(_pt[tid].funcptr == NULL);
void *retval = _pt[tid].funcptr(_pt[tid].funcarg);
RigelBroadcastUpdate(retval, _pt[tid].retval);
RigelBroadcastUpdate(NULL, _pt[tid].funcptr);
}
}
}
int main(int argc, char *argv[])
{
if(RigelGetThreadNum() == 0) {
SIM_SLEEP_OFF();
}
void *mallocs[100];
for(int i = 0; i < 100; i++) {
mallocs[i] = malloc(RigelRandUInt(4, 2000)*4);
RigelPrint(i);
}
for(int i = 0; i < 100; i++) {
free(mallocs[i]);
RigelPrint(0xFFFF0000 | i);
}
//SIM_SLEEP_OFF();
//}
return 0;
}
