From: hjstein@bfr.co.il (Harvey J. Stein)
Newsgroups: comp.parallel.mpi,comp.parallel.pvm
Subject: Re: Poor MPI performance with collection of Linux workstations
Date: 18 Mar 1999 12:46:21 +0300
Organization: Unspecified Organization
Message-Id: <m2d8275dw2.fsf@blinky.bfr.co.il>
References: <36ED325A.69D8C9B8@rrsg.ee.uct.ac.za>
    <7cn7up$4tq$1@grandprime.binc.net> <m2r9qor05g.fsf@blinky.bfr.co.il>
    <36F00B68.99E9187@ece.neu.edu>
Cc: hjstein@bfr.co.il
Xref: ukc comp.parallel.mpi:4776 comp.parallel.pvm:8144


Srinidhi <srinidhi@ece.neu.edu> writes:

 > Harvey J. Stein wrote:
 > 
 > >  > I don't have an answer to your problem, but I have a similar problem.
 > >  > I'm running Linux and have just recently upgraded all machines to the
 > >  > 2.2.3 kernel, hoping my problems would be solved. My problems persist. I
 > >  > have nine machines, one of which is a 4-processor box.
 > >
 > > Are they small packets?  If so, here's a patch from the linux-kernel
 > > mailing list that might help:
 > 
 > I have a similar problem. A 1 Kbyte or larger packet takes less
 > time (~ 1ms) than a 4 byte mesage (~15ms). This is for PVM on Linux
 > (Redhat 5.1). Does this patch also fix the problem in PVM? PVM does
 > not use TCP/IP, AFAIK.

AFAIK, the patch only has to do with TCP, not with UDP.  I suggest
emailing a bug report to the linux-kernel mailing list.  Be sure to
include the output of a tcpdump showing the problem, or people won't
really be able to do anything about it.

-- 
Harvey J. Stein
BFM Financial Research
hjstein@bfr.co.il

