Cornell Theory Center

Presentation
MPI Persistent Communication

10/98

This is the presentation layer of a two-part module. For an explanation of the layers and how to navigate within and between them, return to the top page of this module.

1. Overview
2. Internals
3. Usage
4. Performance

References Lab Exercises Evaluation

1. Overview

Purpose

reduce communication overhead

Use when

point-to-point calls are repeated many times

example: boundary values

Notes

non-blocking

variant for each communication mode

can be used for send and/or receive

2. Internals

Request object

internal representation of a communication event

includes mode, arguments, status

Conventional communication

Called	Equivalent	Effect
MPI_Isend	MPI_Send_init	create request object
	MPI_Start	start communication
MPI_Wait	MPI_Wait	end communication
	MPI_Request_free	release request

Persistent communication

only call MPI_Send_init and MPI_Request_free once
if request can be reused

3. Usage

Alternate foils for hardcopy presentations

4. Performance

Improvement in wallclock time,
Persistent vs. conventional communication

One task sends, one receives
Timings done with PE version 2.1.0.6
Improvement was independent of the number of iterations

msize (bytes)	mode	improvement	mode	improvement
8	async	19 %	sync	15 %
4096	async	11 %	sync	4.7 %
8192	async	5.9 %	sync	2.9 %
800,000	-	-	sync	0 %
8,000,000	-	-	sync	0 %

References

Message Passing Interface Forum (June 1995) MPI: A Message Passing Interface Standard.

CTC's MPI Documentation

Lab exercise for MPI Persistent Communication

Please complete this short evaluation form. Thank you!

URL http://www.tc.cornell.edu/Edu/Talks/MPI/Persistent/less.html