Newsgroups: comp.parallel.mpi
From: sundar@aleutian.cis.ohio-state.edu (N Sundar)
Subject: Re: MPI_Alltoall() on Cray T3D
Summary: What's the algo for doing MPI_Alltoall on T3D?
Organization: The Ohio State University, Department of Computer and Information Science
Date: 12 Nov 1996 20:08:31 -0500
Message-ID: <56b72fINNpv6@aleutian.cis.ohio-state.edu>

Thanks to the people from Edinburgh Parallel Computing Center who
answered my previous question about the implementation of
MPI_Alltoall() on Cray T3D. This has encouraged me to come back with
more questions.

I understand MPI_Alltoall() is implemented with each processor
directly sending to every other such that no processor receives from
two others in the same step. But is this an exchange algorithm i.e.
if processor x sends to y in a step, it also receives from y? Is any
attempt made to avoid contention for the network links by messages?  I
guess I'd really appreciate if somebody can mention the exact way the
schedule is generated.

Also, is MPI_Alltoallv() implemented with the same algorithm?

Thanks again.
Sundar.

-- 
------------+----------------Disclaimer: These are my opinions only.
N. S. Sundar!    I am ardently against all awkward, affected, 
GTA,CIS,OSU.!    arbitrary and aimless alliteration.
INTJ        !    WWW: http://www.cis.ohio-state.edu/~sundar/

