Newsgroups: comp.parallel.pvm
Path: ukc!uknet!pipex!howland.reston.ans.net!europa.eng.gtefsd.com!MathWorks.Com!yeshua.marcam.com!zip.eecs.umich.edu!caen!batcomputer!ghost.dsi.unimi.it!univ-lyon1.fr!swidir.switch.ch!scsing.switch.ch!news.dfn.de!urmel.informatik.rwth-aachen.de!newsserver.rrzn.uni-hannover.de!rz.uni-hildesheim.de!baghira.han.de!erni.escape.de!dagoba.escape.de!strauss
From: strauss@dagoba.escape.de (Frank Strauss)
Subject: Re: Pvm_Data_In_Place?
References: <DAHLE.94Feb19145957@dalton.uio.no>
Organization: Private Site
Date: Wed, 23 Feb 1994 18:26:52 GMT
Message-ID: <1994Feb23.182652.3146@dagoba.escape.de>
Lines: 19

dahle@dalton.uio.no (Paal Dahle) writes:

>  Also, I have tried the Pvm_Data_Raw option several times, but so far I
>have not had any kind of improvement in the speed of data transfer.
>  Will anyone please tell me what kind of improvement it is reasonable to
>expect from this encoding option!

Remember that XDR is only used on real heterogenous networks if you
use Pvm_Data_Default. E.g. PVM3 will detect that the virtual machine
consists of just SUN4 nodes and will not use XDR. So it behaves
like using Pvm_Data_Raw. If you would like to see speed improvements
use at least one other architecture and a very fine granularity for
having much traffic. If you have nothing else to do :-), let us know
your results, please. BTW, what happens, if two architectures with
obviously same byte orders and FP representation are used?

-- 
Frank Strauss, TU Braunschweig                  strauss@{ibr|ips}.cs.tu-bs.de
Buergerstr.13, 38118 BS, (0531)-503873    strauss@dagoba.{escape|priconet}.de

