Newsgroups: comp.parallel.mpi
From: neely@crevasse.llnl.gov (Rob Neely)
Subject: Re: MPI on T3D
Keywords: t3d,mpi,psc
Organization: Lawrence Livermore National Laboratory, NCD
Date: 10 Oct 1994 17:51:28 GMT
Message-ID: <37buv0$5si@lll-winken.llnl.gov>


> We at Edinburgh Parallel Computing Centre are currently working on a
> native implementation of the full MPI interface for the Cray T3D in 

Any early indications on performance?  Do you expect to be able to
obtain better performance than the PVM layer CRI provided?  Are you able
to take advantage of the block transfer (SHMEM) stuff, either in point-
to-point or group comm?

Some measured numbers that I saw for the machine were:

     latency    bandwidth
PVM     80 us      30 M/s
SHMEM   1 us       120 M/s

Although I realize that the SHMEM layer is inherently faster, this 
discrepancy (especially in latencies) seemed a little extreme.  Others
that I talked to seemed to think that the MPI port *should* be able to
do much better than the PVM layer.

--Rob Neely, LLNL

