Re: MPI vs PVM
>>>>> "vince" == Vince Mulhollon <vlm@norlight.com> writes:
vince> Kind of like asking what is the best beer, they all work, but individuals
vince> have strong preferences about them...
vince> The "correct" answer is find the open source project that is the most
vince> similar to your task, and use whatever it uses.
This is honestly the truth...
(in our application, which is a higher-level application for
an easily useable/buildable system interactively building parallel
computations for statistical analysis, we've managed to find counter
examples to most of the conventional wisdoms regarding MPI (LAM-MPI
implementation, we hope to add MPICH as well) and PVM. There's a
socket interface as well that we've got as a back end).
(We use R (www.r-project.org) as the workhorse in front of ones choice
of message passing library...).
best,
-tony
--
A.J. Rossini Rsrch. Asst. Prof. of Biostatistics
U. of Washington Biostatistics rossini@u.washington.edu
FHCRC/SCHARP/HIV Vaccine Trials Net rossini@scharp.org
-------------- http://software.biostat.washington.edu/ ----------------
FHCRC: M: 206-667-7025 (fax=4812)|Voicemail is pretty sketchy/use Email
UW: Th: 206-543-1044 (fax=3286)|Change last 4 digits of phone to FAX
(my tuesday/wednesday/friday locations are completely unpredictable.)
Reply to: