Newsgroups: comp.parallel.pvm
From: sucha@immunex.com (Sucha Sudarsanam)
Subject: PVM om desktop machines
Organization: Immunex Corporation
Date: Sun, 03 Dec 1995 11:21:24 -0800
Message-ID: <sucha-0312951121240001@gumbo.immunex.com>

Hi,

The time has come to take the lowly desktop machines seriously for PVM
implementation. We have benchmarked a (real life problem!) protein loop
modeling algorithm developed by our group (Sudarsanam et al., Prot. Sci,
4, 1412-1420, 1995) that we have parallelized using PVM (Sudarsanam et
al., Comput. Appl. Bio. Sci. In press). Without going into all the
details, it appears that the Mac PowerPC 9500/132 performs about as well
as a very expensive DEC Alpha 4/2100 (275 MHz).  PPC 9500 outperforms just
about every expensive workstation we bought in the past two years or so.
Thus, in our case, adding a single PPC to the virtual machine would mean a
lot more than adding a few traditional Unix workstations.

I realize it may not be  very straightforward to port PVM to desktop
machines (although, it appears that Windows machines can now run PVM). Is
any one working on a Mac version of PVM?

I would also like to hear from the PVM development team on their plans to
port to desktop machines. Given that the basic idea behind PVM is to make
use of existing computing resources efficiently, it is imperative that it
be ported to the more ubiquitous desktop machines.

Any comments?

Thanks,
Sucha

-- 
Sucha Sudarsanam
Dept. of Protein Chemistry
Immunex Corporation
51 University Street
Seattle WA 98101

