Re: [AMBER-Developers] Intel and At1.5/Amber11 combo (on rhel 4)

From: David A. Case <case.biomaps.rutgers.edu>
Date: Wed, 8 Jun 2011 17:33:34 -0400

On Wed, Jun 08, 2011, B. Lachele Foley wrote:
>
> When we try to build the AT1.5/Amber11 combo on RHEL4/Intel, there are
> errors. Building from a recent git repo is good (except the make clean
> thing in the earlier email). Building with AT1.4/Amber11 is good.
>
> The complaint concerns constants.f,

> cpp -traditional constants.f > _constants.f
> mpif90 -c -ip -O3 -xHost -o constants.o _constants.f
> : error #5149: Illegal character in statement label field [m]

The -FR (freeformat) flag is not being passed to the compiler. This should be
coming in through the FFLAGS field, which in turn should be picking up the
LOCALFLAGS variable in the amber11/src/sander/Makefile.

The error has nothing to do with constants.f; that is just the first file the
Makefile tries to compile. Nothing in sander will compile without the
free-format flag.

This *is* hard to debug, since no one else seems to be able to reproduce
it. Is it related to the parallel build? (That is, can you build sander
in serial mode?) I suppose you could edit amber11/src/sander/depend,
and in the rule to compile constants.f, have the Makefile echo FFLAGS,
LOCALFLAGS, etc. Note that there are lots of other things in LOCALFLAGS
that are not being picked up. Are you sure that you are running the
AT15_Amber11.py script after every invocation of configure? Basically,
each time you re-run configure, you also need to re-run the python script.

....dac


_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Wed Jun 08 2011 - 15:00:03 PDT
Custom Search