From: Sierk Poetting <spoettin@soliton.physics.arizona.edu>
Newsgroups: comp.parallel
Subject: Parallel Computing, sleeping Processes
Date: 17 Oct 1998 07:42:30 GMT
Organization: The University of Arizona
Approved: bigrigg@cs.cmu.edu
Message-Id: <709hp6$9f0$1@encore.ece.cmu.edu>
Originator: bigrigg@ece.cmu.edu


Hi,

I am runing a program on an SGI-multiprocessor system (14 processors,
IRIX 6.5),  using the autoparallelization option of the latest MIPSpro
C++ compiler. If almost nobody is on the machine I get a very good
performance, the code is using up to 8 processors. But once more
processors are used by others, my performance goes down by a factor of
ca. 20, although there are still enough processors (>8) around that are
not doing anything. My child processes (created once a parallel section
is executed) are sleeping a lot.  I have the same priority level as all
the other users and there is still enough memory available.

Is there a way I can avoid this or  restrict the execution of my program

to certain processors , so that I can definitely avoid collisions with
users using other processors??

If somebody could help me, that would be really nice,

thanks already,
Sierk

--
Articles to bigrigg+parallel@cs.cmu.edu (Admin: bigrigg@cs.cmu.edu)
Archive: http://www.hensa.ac.uk/parallel/internet/usenet/comp.parallel

