Re: [AMBER] ld: duplicate symbol _gauss in ../lib/sys.a(random.o) and rand2.o

From: Jason Swails <>
Date: Thu, 1 Jul 2010 11:00:16 -0400

I would also suggest using macports or fink to build gcc/gfortran 4.4. When
you type gcc --version and gfortran --version, what do you get? Do you see
anything that says "experimental" anywhere? In the past I've had problems
using gfortran-4.5 from the hpc website, since it was an experimental build
that had some bugs when it came to compiling amber. I believe gcc-4.5 was
only recently released officially, and I'm not sure if the binaries on that
site have caught up yet.

Good luck!

On Thu, Jul 1, 2010 at 10:52 AM, case <> wrote:

> On Thu, Jul 01, 2010, Yu Chen wrote:
> >
> > Thanks for quick replying and the patch. Seems the patch works, it got
> > rid of the duplicated symbol problem, but now I have got undefined
> > symbol problem as following, wondering if you have any thought about
> > this. I am wondering if this is related to the "fno-underscoring"
> > flag I have to pass to the MPICH compiling. For the gcc/gfortran
> > 4.5.0. I followed Mengjuei Hsieh's amberonmac blog, and downloaded the
> > pre-compiled package from Thanks in advance for your help!
> If you haven't done so, I strongly recommend you compile your own version
> of openmpi, rather than downloading some pre-compiled version. There
> is a "configure_openmpi" script in $AMBERHOME/AmberTools/src that makes
> sure
> that all the flags are set correctly, etc. This is especially true if you
> are giving different flags (like no_underscore) to MPI and to Amber -- that
> sounds like a recipe for problems.
> It takes a while for openmpi to compile, but you only have to do it once,
> and then you are sure that it is compiled in the same way as Amber.
> ....dac
> _______________________________________________
> AMBER mailing list

Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
AMBER mailing list
Received on Thu Jul 01 2010 - 08:30:04 PDT
Custom Search