Hi,
I found several Python bindings for the MPI protocol,
can anybody report some opinions and experiences ?
Peter Christen and I have used Ole Nielsen's PyPar package (see
http://datamining.anu.edu.au/~ole/pypar/) in our Febrl probabilistic
record linkage engine (see
http://datamining.anu.edu.au/projects/linkage.html) with excellent
results. I have also been experimenting with its use in a parallel data
summarisation engine. It is easy to install and is very easy to use,
mainly because it only wraps a small portion of MPI - but just those
parts you really need (at least we haven't missed anything yet). Its
ability to pass any picklable Python object around the cluster is a real
bonus, and the speed at which it passes Numeric Python arrays using
"raw" mode is very impressive. Ole is also very helpful, and is actively
involved in parallel computing work on a daily basis (using PyPar, no
doubt).
--
Tim C
PGP/GnuPG Key 1024D/EAF993D0 available from keyservers everywhere
or at
http://members.optushome.com.au/tchur/pubkey.asc
Key fingerprint = 8C22 BF76 33BA B3B5 1D5B EB37 7891 46A9 EAF9 93D0
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)
iD8DBQA/DHcneJFGqer5k9ARAlVOAKDzm4Xld02G+C6qRHh9ZaMX2lgTfQCgk1nw
lTw4djQgOQ4HsbzKXVRlV/0=
=hyxe
-----END PGP SIGNATURE-----
