Subject: ccm3 using both SMP and MPI
From: Norm Wood (firstname.lastname@example.org)
Date: Mon Oct 29 2001 - 09:25:33 MST
I'm interested in finding out if anyone has had success running
ccm3.10.11.brnchT.366physics.4 on dual-processor linux boxes using
Portland Group's PGF90 compiler.
I've been attempting to compile and run the model on a single, dual-
processor linux (kernel 2.4.12) machine. If I compile the code without
MPI and without SMP support, it seems to compile and run without a hitch.
If I add SMP support (via the "-mp" flag to the PGF90 compiler),
the compile proceeds without errors, but the run throws a segmentation
fault almost immediately. The atm.log.* file is written to the point
where almost all of FLDLST is reported (the fields included in the primary
history file), but it's interrupted with a line like:
max rss=0 shared mem=0 unshared data=0 unshared stack=0
then the segmentation fault occurs.
When I run the PG debugger on the code, it gets as far as the
"read(5,ccmexp)" statement in data.F90, then hangs. So I haven't
yet been able to identify the location in the code where the
segmentation fault is occurring.
Any suggestions would be appreciated.
-- Norm Wood
Department of Atmospheric Science Colorado State University email@example.com phone: 970-491-3334
This archive was generated by hypermail 2b27 : Tue Jan 22 2002 - 11:16:17 MST