Hi,
You might want to use 'non-blocking' Sends and Receives in this case.
(MPI_Isend and MPI_Irecv)
When a process uses MPI_Send, it essentially blocks itself until a
matching MPI_Recv is posted on the receiving side. In your case, when
Master does an MPI_Send to itself, it is blocking on the MPI_Send and
hence cannot post a MPI_Recv, which could explain your deadlock.
Hope this helps.
--
Prashanth
pcharapa_at_[hidden]
LAM-MPI Team
Thus spoke Irshad Ahmed in the message sent on Thu, 2 Oct 2003
->Hi, Jeremy
->yes, Master is also doing same job as the slaves, and sends the results of jobs to itself.
->I think that there is problem in
->* While statement.
->Thanks for your interest.
->Waiting anxiously for the reply.
->Regards,
->
->
->Ahmed irshad
->
->---------------------------------
->Do you Yahoo!?
->The New Yahoo! Shopping - with improved product search
|
