Re: [AMBER-Developers] Compilation of AMBER Tools

From: David A. Case <case.biomaps.rutgers.edu>
Date: Thu, 9 Jun 2011 02:42:31 -0400

On Wed, Jun 08, 2011, Ross Walker wrote:
>
> Can anybody actually compile AMBER Tools from the git tree anymore without
> using bleeding edge compilers?
>
> Gfortran in Redhat EL5.4 (4.1.2)

I don't think requiring gcc > 4.1 is "bleeding edge". Amber has not worked
correctly for some time at this level. The problem is not gcc, but gfortran:
the early versions of gfortran had many limitations -- back in those days, we
were using g95 as the "free" fortran compiler, since gfortran had so many
limitations.

Someone could try to find out what we used to do with the "allocatable"
statements that gfortran 4.1.2 was happy with. Then we could consider what to
do. But basically, since RedHat came out with an official gcc44 (and later)
packages, we've been telling people to install more recent compilers.

> eff.c(25): warning #161: unrecognized #pragma
> #pragma omp parallel reduction (+: e_cons) private (i, rx, ry, rz, rw)

This is just a warning, and I think all versions of the Intel compilers spew
them out. It just means that you are compiling a non-openmp version of NAB.
Maybe someone knows a flag to suppress this warning, but it's harmless.

....dac


_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Jun 09 2011 - 00:00:03 PDT
Custom Search