On Aug 11, 2004, at 8:08 AM, Jaroslaw Zola wrote:
> Finally, I have written very short wrapper which performs threads
> synchronization on the application level. So I have "multiple threads
> support" for basic routines (MPI_Send, MPI_Recv) from LAM
> implementation.
Note that you should be able to use MPI_THREAD_SERIALIZED to the same
effect -- it puts a global lock around all MPI functions so that only
one thread will be in MPI at a time.
--
{+} Jeff Squyres
{+} jsquyres_at_[hidden]
{+} http://www.lam-mpi.org/
|
