[llvm-bugs] [Bug 37580] New: Handling doacross loops: no collapse

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=37580

            Bug ID: 37580
           Summary: Handling doacross loops: no collapse
           Product: OpenMP
           Version: unspecified
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Clang Compiler Support
          Assignee: unassignedclangbugs at nondot.org
          Reporter: protze at itc.rwth-aachen.de
                CC: llvm-bugs at lists.llvm.org

#include <stdio.h>
#include <omp.h>
int main()
{
  int i, j;
#pragma omp parallel for ordered(2) num_threads(2) schedule(static,1) \
                                                   private(i,j)
  for (i = 0; i < 10; i++)
    for (j = 0; j < 10; j++)
    {
#pragma omp ordered depend(sink:i-1,j) depend (sink:i,j-1)
      printf ("test %i i=%d j=%d\n",omp_get_thread_num(),i,j);
#pragma omp ordered depend(source)
    }
  return 0;
}


According to the OpenMP spec, this code should execute i=0 and i=2 on thread 0
and i=1, i=3 on thread 1. The ordered(2) clause does not change the scheduling.
The behavior of clang looks like collapse(2) is implied.

-> expected behavior: i is distributed between the threads, 
-> observed behavior: j is distributed between the threads.


Since icc in combination with the same (LLVM) OpenMP runtime provides correct
behavior (besides a bug in privatization of j), this should be a compiler bug
in clang.

Also, explicitly adding collapse(1) does not help.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20180524/4f19c42d/attachment-0001.html>