David and Brian,
Thanks again, I see what you both mean about MPI_GATHERV, I didn't realize
the sendcount could be zero on the root processor, looks like it will solve
my problem and a cleaner, clearer solution.
Thanks,
Steve
-----Original Message-----
From: lam-bounces_at_[hidden] [mailto:lam-bounces_at_[hidden]] On Behalf Of
David Cronk
Sent: Tuesday, May 10, 2005 7:31 AM
To: General LAM/MPI mailing list
Subject: Re: LAM: Sending 3D arrays and opaque objects
Steve Lowder wrote:
> David and Brian,
> Thank you for you replies,
You are certainly welcome.
they filled in the gaps for me. I'm going to
> cache my last datatype and free the previous two. I would have liked to
use
> MPI_Gather or MPI_Gatherv but my problem dictated that the root processor
> not be a data contributor to the gathered data
This is why I suggested Gatherv. This allows you to specify how much
data is contributed by each process, and this count can be 0 for one or
more processes. For your purposes, the root would use sendcount = 0 and
recvcounts[0] = 0. You just need to make sure you get the displacements
setup correctly.
Dave.
|
