From: Chris Samuel (csamuel_at_vpac.org)
Date: Wed Mar 31 2004 - 00:40:36 CST
-----BEGIN PGP SIGNED MESSAGE-----
On Wed, 31 Mar 2004 03:58 pm, David Chalmers wrote:
Hi David, :-)
> Thanks for your reply. We have Charm and NAMD compiled using the
> intel_cc_80 compiler and mpicc.
I've just joined the list and wanted to give a quick bit of background.
To get Charm++ to compile with the v8 Intel compilers I needed to add an extra
library search path in, I've done that as a quick hack by adding it into the
CORE_LIBS definition in the charmc script - see the attached patch.
It's not designed for general use as I patch the script itself, I'd be happier
if it could be in a config file but as a newcomer to Charm++ it's hard to
decipher what to set where!
Also I created two files to define an mpicc compiler type to use instead of
icc. From what I could tell if you try and build an MPI version of Charm++
at present (./build charm++ mpi-linux) it assumes that the MPICH you're using
is built with GCC as the build falls back to using gcc in a couple of places
leading to link failures if MPICH is really using the Intel compilers.
I've attached the two files, cc-mpicc.sh and cc-mpicc.h (attached) in the
directory charm/src/arch/common which at least let me build Charm++ cleanly
with the command line:
./build charm++ mpi-linux mpicc
Perhaps they should have a better name like intel-mpicc ?
Unfortunately, as David says, when I try and run the megatest program it runs
happily at first before bombing out with those errors. Once it's done that
the process hangs eating all the CPU it can until its walltime expires and
Torque terminates it. :-(
Christopher Samuel - (03)9925 4751 - VPAC Systems & Network Admin
Victorian Partnership for Advanced Computing http://www.vpac.org/
Bldg 91, 110 Victoria Street, Carlton South, VIC 3053, Australia
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)
-----END PGP SIGNATURE-----
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 05:18:09 CST