Re: [AMBER] Fwd: some clues about yacc errors on OLCF rhea

From: Ada Sedova <ada.a.sedova.gmail.com>
Date: Wed, 18 Jan 2017 23:04:45 -0500

Wow, it worked. Looks like yacc was the only hangup with the openmpi. I
still have to run tests, but both builds finished.

The only other person at ORNL that I know who managed to build amber14
successfully on one of our intel commodity clusters had to install his own
version of openmpi in the home dir.

This fix was much easier.

Thanks, everyone!

~AS

On Wed, Jan 18, 2017 at 4:57 PM, Scott Brozell <sbrozell.rci.rutgers.edu>
wrote:

> Hi,
>
> As i already indicated in this thread, if you do this:
> export YACC=/usr/bin/yacc
> then invoke make this way:
> make -e ...
>
> for details see:
> man make
>
> scott
>
> On Wed, Jan 18, 2017 at 04:29:37PM -0500, Ada Sedova wrote:
> > Yes, I agree, I had considered following the long trail of config.h files
> > and Makefiles and trying to reset the YACC variable myself; but I
> imagined
> > I would maybe miss some as I don't know the entire workflow.
> >
> > And yes, it is possible that other libraries that amber tries to build
> will
> > also have an issue with our openmpi, or at least our mpicc wrapper.
> >
> > This is an especially difficult make for HPC systems exactly because of
> all
> > of the library building involved, as opposed to using variables that
> allow
> > the user to set the paths to the various libraries on the system.
> >
> > On Wed, Jan 18, 2017 at 3:25 PM, David A Case <david.case.rutgers.edu>
> > wrote:
> >
> > > On Wed, Jan 18, 2017, Hai Nguyen wrote:
> > >
> > > > may be try
> > > > export YACC=/usr/bin/yacc
> > > > to force amber to use your system yacc?
> > >
> > > I don't think this will work. But you could edit config.h and set the
> YACC
> > > variable there to /usr/bin/yacc. As far as I understand it, the value
> in
> > > config.h (*not* the value in the environment) is what determined which
> yacc
> > > is used.
> > >
> > > [There have indeed been some off-list emails about this. My
> reservation is
> > > this: if mpicc fails in building yacc, will it not also fail in
> building
> > > other pieces of Amber?]
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jan 18 2017 - 20:30:03 PST
Custom Search