[mpich-discuss] MPI_Allreduce fail. Please help. (UPDATE) (WINDOWS XP 32bit+gfortran)

[Yonghui]

Hello MPICH2 users and developers,

 

I did some more tests on a Windows XP 32bit machine with MPICH2 32bit and
MinGW 32bit compilers. The code in my previous email failed when calling
MPI_Allreduce in fortran. So here are the results:

 

1.       Works on Linux (Ubuntu 11.10) and OS X (thanks to Rajeev Thakur's
email). But fail on Windows xp 32bit and  Windows 7 64bit (running the 32bit
code).

2.       No problem if the matrix is much smaller (e.g. 10x10x10 matrix).

3.       Works if not using MPI_IN_PLACE.

 

So I  guess it could be a MPI_IN_PLACE problem happened in the windows
version fortran functions.

Or it could be the problem of MinGW?

 

The c code works fine. And in c version MPI_IN_PLACE = 0xffffffff but
fortran version MPI_IN_PLACE is 0. I can't tell whether they should be
different since I don't know how does it implemented.

 

Please help me out! Thanks.

Y L

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20120905/7c7f8866/attachment.html>