Newsgroups: comp.parallel.mpi
From: Nicolas G Leshock <leshock@cse.psu.edu>
Subject: Spawning tasks using MPIRUN on Sun WS cluster
Organization: PSU CSE Dept; US Army Signal Corps
Date: 27 Aug 1995 01:18:08 GMT
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Message-ID: <41oh4g$3k4@psuvax1.cse.psu.edu>

Hi,
	I running tests to determine speedup of a parallel
program.  When I use MPIRUN to start the run (the servers are
already running on the machines) I get the following on a TOP
printout on each slave machine.

 2491 leshock   36    0  5652K  712K run     0:02 74.73% 14.06% nav_mpi540
 2478 leshock   25    0   608K  924K run     0:03  0.00%  0.00% top-sun4m-4.1.
 2465 leshock   15    0   504K    0K sleep   0:00  0.00%  0.00% tcsh
 2464 leshock    1    0   696K  572K sleep   0:00  0.00%  0.00% xterm
 2492 leshock    1    0  5652K  248K sleep   0:00  0.00%  0.00% nav_mpi540
 2476 leshock    1    0    48K   48K sleep   0:00  0.00%  0.00% serv_p4

Does anyone know why two copies of the code are spawned. Is there something
I'm doing wrong?  The second copy never runs, just sleeps and then dies when
the first (higher priority) running job completes.

A related question is whether the serv_p4 program can be run under inetd?

Thanks in advance for your help!
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~  Nick Leshock     email: ngl1@psu.edu,leshock@cse.psu.edu       ~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~  "Go, sir, gallop, and don't forget that the world was made in  ~
~   six days.  You can ask me for anything you like, except       ~
~   time."                                                        ~
~                 Napoleon Bonaparte to an aide (1803)            ~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


