mpiexec compilation with pgi compilers

[Ben Webb]

On Wed, Jun 19, 2002 at 10:09:24AM -0600, Brashers, Bart -- MFG, Inc. wrote:
> Thanks for your help, and work, Pete!  Version 0.68 compiled under pgcc just
> fine.  I'm still running into problems, though...
> > You can do without the "--with-pbssrc" since that's only for
> > ancient PBS.  We'll get rid of it someday.  Also read below for how
> > to change your --with-mpi arg.
> configure still complained when I tried it without the "--with-pbssrc".  But
> it worked if I added it back.  

	Perhaps you need --with-pbs instead. I have the PBS headers
installed in /usr/include, and the libraries in /usr/lib, so use
"--with-pbs=/usr".

> I had applied the mpich-sysvipc.patch to mpich-1.2.4, but I'm not sure why.

I think only you can explain why. ;)

> Now that I look for it on the web, I can't figure out when I got it.

http://bellatrix.pcl.ox.ac.uk/~ben/pbs/mpich-sysvipc.patch

> That caused errors referring to the P4_SYSV_FILE upon submission.

	Good! That's what it's supposed to do. Without the patch, MPICH
will work quite happily, creating and using SysV shared memory. But if
your job crashes (or you "qdel" it) then this memory isn't freed... and
eventually you'll chew it all up. You can use the "ipcs" program to list
the memory, and "ipcrm" to remove it, but you'll have no easy way of telling
which segments were created by MPICH (and can be deleted) and which are still
being used by MPICH jobs or other processes. _With_ the patch, the IDs
of any shared memory or semaphores are written into a file, so that you
can clear it up later. See http://bellatrix.pcl.ox.ac.uk/~ben/pbs/ for
details.

	Ben
-- 
ben at bellatrix.pcl.ox.ac.uk           http://bellatrix.pcl.ox.ac.uk/~ben/
"It isn't pollution that's harming the environment.  It's the
impurities in our air and water that are doing it."
	- Vice President Dan Quayle