Newsgroups: comp.parallel.mpi
From: edemaine@ug.cs.dal.ca (Erik Demaine)
Subject: Re: MPI on an SP2
Organization: Math, Stats & CS, Dalhousie University, Halifax, NS, Canada
Date: Tue, 17 Jan 1995 23:07:28 GMT
Message-ID: <D2Ko8H.C5B@cs.dal.ca>

NOTE: THIS INCLUDES A QUESTION, NOT JUST A FOLLOWUP

Erik Demaine (I myself, edemaine@ug.cs.dal.ca) wrote:
: Does the implementation of MPI on an SP2 come with the regular software?
: If so, how do I access it?  Any special "gotchas"?  How do I launch an
: MPI program on an SP2?
: If not, what portable imp. should I use? (for example, MPICH)

Thanks for all the responses.  I previously thought (from reading a paper
on it) that MPIF (IBM's MPI implementation) was "ready," but it seems it
does not come with the system yet.  My best bet appears to be running
MPICH -- some benchmarks are available at:
	http://www.npac.syr.edu/users/alvin/homepage/mpich.html
(thanks Alvin) These show several megs/sec throughput with small messages,
which sounds great to me.

We will be getting the SP2 at the end of the month.  So, we'd like to
develope some programs on a single node (again using MPICH -- hurray the
portability of MPI) simulating parallelism (most likely using the p4
option).  I've been trying to get it to compile on an HP but to no avail.
Some stats on the machine:
	HP-UX cobra A.09.01 A 9000/735 2005328673 two-user license
I am trying to compile MPICH 1.0.7.
The error message (when making, I get up to compiling examples):
     gcc -g -o overtake overtake.o test.o -L/tmp/tmp/mpich/lib/hpux/ch_p4 -lmpi
/bin/ld: Unsatisfied symbols:
   sigrelse (code)
   MPIR_init_bottom (code)
   mpir_init_fdtes (code)
   sigset (code)
   mpir_init_fop (code)
   sighold (code)
   mpir_init_fcm (code)

I probably have to link in a library for sig*.  What about mpir*?  One note
is that I know the linker is case-sensitive (even in Fortran).. Is this a
problem?

Has anyone compiled MPICH on an HP sucessfully?

Any help would be appreciated!
Erik
--
Erik Demaine        || edemaine@ug.cs.dal.ca  || edemaine@fx2800.dal.ca
edemaine@cs.dal.ca  || 01ERIK@ac.dal.ca       || edemaine@is.dal.ca
URL: http://ug.cs.dal.ca:3400/~edemaine/edemaine.html
--> A glob is a round substance of thick liquid. A blob is an unformed object.

