bda/gpu-mpi-tests
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

distribute total across ranks

Browse Source 
useful for test < 6 ranks per node
This commit is contained in:
Bryce Allen
2020-08-07 17:53:16 -04:00
parent 538c22a22f
commit c32b86422f
1 changed files with 36 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

48
mpi_daxpy_nvtx.cc
View File

@@ -68,7 +68,8 @@ void set_rank_device(int n_ranks, int rank) {
int main(int argc, char **argv) {
int n = 4*MB;
const int nall = 48*MB;
int n = 0;
int world_size, world_rank;
size_t free_mem, total_mem;
@@ -84,7 +85,8 @@ int main(int argc, char **argv) {
double g_end_time = 0.0;
#ifndef MANAGED
double *h_x, *h_y, *h_allx, *h_ally;
double *h_x, *h_y;
double *h_allx, *h_ally;
#endif
double *d_x, *d_y;
@@ -97,6 +99,12 @@ int main(int argc, char **argv) {
MPI_Comm_size(MPI_COMM_WORLD, &world_size);
MPI_Comm_rank(MPI_COMM_WORLD, &world_rank);
n = nall / world_size;
if (world_rank == 0) {
printf("%d ranks, %d elements each, total %d\n", world_size, n, nall);
}
/*
x = (double *)malloc(n*sizeof(*x));
if (x == NULL) {
@@ -159,9 +167,11 @@ int main(int argc, char **argv) {
#endif
nvtxRangePop();
if (world_rank == 0) {
CHECK( "memInfo", cudaMemGetInfo(&free_mem, &total_mem) );
printf("GPU memory %0.3f / %0.3f (%0.3f)\n", free_mem/(double)MB,
printf("GPU memory %0.3f / %0.3f (%0.3f) MB\n", free_mem/(double)MB,
(double)total_mem/MB, (double)(total_mem-free_mem)/MB);
}
nvtxRangePushA("initializeArrays");
#ifdef MANAGED
@@ -175,9 +185,9 @@ int main(int argc, char **argv) {
h_y[i] = -h_x[i];
}
nvtxRangePushA("copyInput");
CHECK("d_x = x",
CHECK("d_x = h_x",
cudaMemcpy(d_x, h_x, n*sizeof(*h_x), cudaMemcpyHostToDevice) );
CHECK("d_y = y",
CHECK("d_y = h_y",
cudaMemcpy(d_y, h_y, n*sizeof(*h_y), cudaMemcpyHostToDevice) );
nvtxRangePop();
#endif
@@ -191,6 +201,13 @@ int main(int argc, char **argv) {
MEMINFO("d_x", d_x, sizeof(d_x));
MEMINFO("d_y", d_y, sizeof(d_y));
#ifndef MANAGED
MEMINFO("h_x", h_x, sizeof(h_x));
MEMINFO("h_y", h_y, sizeof(h_y));
MEMINFO("h_allx", h_allx, sizeof(h_allx));
MEMINFO("h_ally", h_ally, sizeof(h_ally));
#endif
k_start_time = MPI_Wtime();
nvtxRangePushA("cublasDaxpy");
CHECK("daxpy",
@@ -237,9 +254,9 @@ int main(int argc, char **argv) {
sum += d_ally[i];
}
#else
nvtxRangePushA("copyOutput");
nvtxRangePushA("copyAlly");
CHECK("h_ally = d_ally",
cudaMemcpy(h_ally, d_ally, n*sizeof(*h_ally),
cudaMemcpy(h_ally, d_ally, n*sizeof(*h_ally)*world_size,
cudaMemcpyDeviceToHost) );
nvtxRangePop();
for (int i=0; i<n*world_size; i++) {
@@ -252,10 +269,10 @@ int main(int argc, char **argv) {
// cleanup
nvtxRangePushA("free");
#ifndef MANAGED
cudaFree(h_x);
cudaFree(h_y);
cudaFree(h_allx);
cudaFree(h_ally);
cudaFreeHost(h_x);
cudaFreeHost(h_y);
cudaFreeHost(h_allx);
cudaFreeHost(h_ally);
#endif
cudaFree(d_x);
cudaFree(d_y);
@@ -270,9 +287,12 @@ int main(int argc, char **argv) {
cublasDestroy(handle);
MPI_Finalize();
printf("total time: %0.3f\n", end_time-start_time);
printf("kernel time: %0.3f\n", k_end_time-k_start_time);
printf("gather time: %0.3f\n", g_end_time-g_start_time);
printf("%d/%d TIME total : %0.3f\n", world_rank, world_size,
end_time-start_time);
printf("%d/%d TIME kernel: %0.3f\n", world_rank, world_size,
k_end_time-k_start_time);
printf("%d/%d TIME gather: %0.3f\n", world_rank, world_size,
g_end_time-g_start_time);
return EXIT_SUCCESS;
}