Newsgroups: comp.parallel.mpi
From: gorlatch@fmi.uni-passau.de (Sergei Gorlatch)
Subject: Re: shift in MPI?
Organization: University of Passau, Germany
Date: 29 Jan 97 11:53:55 GMT
Message-ID: <32ef3a53.0@news.rz.uni-passau.de>

In article <5cn286$f1p@murrow.corp.sgi.com>, salo@mrjones.engr.sgi.com (Eric Salo) writes:
|> > what about MPI_Sendrecv ???
|> 
|> I hate this function. It presents the illusion of optimization when in
|> reality it virtually guarantees that your code will be as slow as it
|> can possibly be. You would almost certainly be better off just rolling

Thanks for the warning, Eric.
Frankly I don't see why this seemingly quite simple pattern
of communication cannot be implemented efficiently.
Any inherent reasons?

Thanks,

Sergei