Newsgroups: comp.parallel.mpi
From: Reuben Pasquini <pasquini@cs.purdue.edu>
Subject: probe vs. nonblocking recv test
Organization: Purdue University
Date: Thu, 07 Aug 1997 15:51:54 -0500
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Message-ID: <33EA356A.2184@cs.purdue.edu>

Hi,

My question is this:
Would it be more efficient to simply execute a non-blocking
receive and periodically test the receive handle for completion,
or is network probing more efficient, or are they the same ?

Does the answer depend on my platform (the application currently
runs on a shared memory multiprocessor,
workstations on an ethernet, and the IBM SP2) ?

I just want to check if anybody knows the answers before I spend
the time rewriting my code!

Thanks for the help.

Reuben