File: malloc.c

package info (click to toggle)
llvm-toolchain-21 1%3A21.1.8-2
  • links: PTS, VCS
  • area: main
  • in suites: sid
  • size: 2,236,724 kB
  • sloc: cpp: 7,619,776; ansic: 1,433,958; asm: 1,058,748; python: 252,197; f90: 94,671; objc: 70,753; lisp: 42,813; pascal: 18,401; sh: 8,601; ml: 5,111; perl: 4,720; makefile: 3,719; awk: 3,523; javascript: 2,272; xml: 892; fortran: 770
file content (37 lines) | stat: -rw-r--r-- 991 bytes parent folder | download | duplicates (11)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
// RUN: %libomptarget-compile-generic && %libomptarget-run-generic
// RUN: %libomptarget-compileopt-generic && %libomptarget-run-generic

#include <stdio.h>
#include <stdlib.h>

int main() {
  long unsigned *DP = 0;
  int N = 32;
  int Threads = 64;
  int Teams = 10;

  // Allocate ~55MB on the device.
#pragma omp target map(from : DP)
  DP = (long unsigned *)malloc(sizeof(long unsigned) * N * Threads * Teams);

#pragma omp target teams distribute parallel for num_teams(Teams)              \
    thread_limit(Threads) is_device_ptr(DP)
  for (int i = 0; i < Threads * Teams; ++i) {
    for (int j = 0; j < N; ++j) {
      DP[i * N + 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 * N + j];
    }
  }

  // CHECK: Sum: 6860800
  printf("Sum: %li\n", s);
  return 0;
}