670c453aeb
Summary: This was a lot of code that was only used for upstream LLVM builds of AMDGPU offloading. We have a generic and fast `malloc` in `libc` now so just use that. Simplifies code, can be added back if we start providing alternate forms but I don't think there's a single use-case that would justify it yet.
43 lines
1.1 KiB
C
43 lines
1.1 KiB
C
// RUN: %libomptarget-compile-run-and-check-generic
|
|
// RUN: %libomptarget-compileopt-run-and-check-generic
|
|
|
|
#include <omp.h>
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
|
|
int main() {
|
|
long unsigned **DP = 0;
|
|
int N = 32;
|
|
int Threads = 64;
|
|
int Teams = 10;
|
|
|
|
#pragma omp target map(from : DP)
|
|
DP = (long unsigned **)malloc(sizeof(long unsigned *) * Threads * Teams);
|
|
|
|
#pragma omp target teams distribute parallel for num_teams(Teams) \
|
|
thread_limit(Threads)
|
|
for (int i = 0; i < Threads * Teams; ++i)
|
|
DP[i] = (long unsigned *)malloc(sizeof(long unsigned) * N);
|
|
|
|
#pragma omp target teams distribute parallel for num_teams(Teams) \
|
|
thread_limit(Threads)
|
|
for (int i = 0; i < Threads * Teams; ++i) {
|
|
for (int j = 0; j < N; ++j) {
|
|
DP[i][j] = i + j;
|
|
}
|
|
}
|
|
|
|
long unsigned s = 0;
|
|
#pragma omp target teams distribute parallel for num_teams(Teams) \
|
|
thread_limit(Threads) reduction(+ : s)
|
|
for (int i = 0; i < Threads * Teams; ++i) {
|
|
for (int j = 0; j < N; ++j) {
|
|
s += DP[i][j];
|
|
}
|
|
}
|
|
|
|
// CHECK: Sum: 6860800
|
|
printf("Sum: %li\n", s);
|
|
return 0;
|
|
}
|