On Wednesday 12 September 2012 17:16:48 Paul Hargrove wrote:
> Solaris and NetBSD platforms which displayed the VT Makefile error w/
> 1.6.2rc2 have completed successful builds w/ 1.6.2rc3.
> The PGI-8.0 platform which showed the other VT problem is down at the
> moment. So, I've tested that fix on the older (and thus slower) PGI-7.2-5
> platform which also displayed the original problem. That looks to be
> fixed too.
Strange. I've tested the fix with PGI version 7.2.5 as well - it works!
Could you please verify whether the preprocessor reaches the line 65 in
ompi/contrib/vt/vt/tools/vtunify/vt_unify.h (e.g. by inserting an #error
statement)? At this point the OpenMP parallelization will be disabled, if the
major version of the PGI compiler is less than 9.
> Finally, Jeff's fix to "loosen" the opal_nfs_path test to ignore
> type==none (e.g. for bind-mounts) doesn't appear to be in this rc -
> the file test/util/opal_path_nfs.c is unchanged relative to rc2.
> Perhaps a CMR is in order? Perhaps it just didn't rank as important
> -Paul [who doesn't mind creating more work for Jeff]
> On Wed, Sep 12, 2012 at 2:59 PM, Jeff Squyres <jsquyres_at_[hidden]> wrote:
> > Usual place:
> > http://www.open-mpi.org/software/ompi/v1.6/
> > Changes since rc2:
> > - Update README about IA64 and Intel compilers
> > - Fix VT compile problems
> > --
> > Jeff Squyres
> > jsquyres_at_[hidden]
> > For corporate legal information go to:
> > http://www.cisco.com/web/about/doing_business/legal/cri/
> > _______________________________________________
> > devel mailing list
> > devel_at_[hidden]
> > http://www.open-mpi.org/mailman/listinfo.cgi/devel
Technische Universität Dresden
Center for Information Services and High Performance Computing (ZIH)
01062 Dresden, Germany
Phone: +49 (351) 463-31945
Fax: +49 (351) 463-37773