[mpich-discuss] MPI_Send/MPI_Recv - getting delayed indeffinitely

Pavan Balaji balaji at mcs.anl.gov
Thu Dec 26 01:44:27 CST 2013


On Dec 26, 2013, at 1:38 AM, Madhawa Bandara <madawa911 at gmail.com> wrote:
> Are you suggesting using non-blocking send/recv?

On the master process, yes.  On the slave, it won't matter.

An alternative is to use MPI_ANY_SOURCE receive, but nonblocking is better.

--
Pavan Balaji
http://www.mcs.anl.gov/~balaji




More information about the discuss mailing list