Newsgroups: comp.parallel.mpi
From: lusk@donner.mcs.anl.gov (Rusty Lusk)
Subject: Re: MPI for shared memory machines?
Organization: Argonne National Laboratory, Chicago, Illinois
Date: Fri, 5 Aug 1994 14:48:16 GMT
Message-ID: <Cu2H4H.HLs@mcs.anl.gov>

In article <LUIS.94Aug4233243@gaia.electrum.kth.se>, luis@gaia.electrum.kth.se (Luis Barriga) writes:
>
>Well, it seems like overkill. MPI on top of P4 which is on top of
>SYSV which itself is a mess. I am also intererested in high
>performance and I look forward to is MPI/P4 on on top of user-level
>threads that Solaris is offering.
>

The ANL/MSU implmentation of MPI is built on a device interface that
makes porting convenient.  We are adding various implementations of
the device as fast as we can.  There are special implementations for
the IBM-SP1, Intel Delta, IPSC860, and Paragon, and the CM5.  Another
"device" is Chameleon-p4, which has given us quick, wide portability
to just about everything else.  You can also use PVM as the underlying
transport mechanism.  But direct shared-memory implementations are in
the works (probably SGI first), as well as a cut-down TCP/IP
implementation that bypasses p4.  Expect these in the fall.

Rusty Lusk

