[petsc-users] ERROR: Scalar value must be same on all processes

[Vijay Gopal Chilkuri]

I did what barry adviced.
I found that it's only the -xAVX option which causes problems.

If I compile the code with the above COPTFLAGS , *excluding* the *-xAVX*
everything works fine !

Thanks for the advice.

On Tue, May 26, 2015 at 2:25 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:

>
> > On May 26, 2015, at 2:14 AM, Jed Brown <jed at jedbrown.org> wrote:
> >
> > Vijay Gopal Chilkuri <vijay.gopal.c at gmail.com> writes:
> >> So somehow the optimization flags effect the code execution !?
> >> is this a known issue ?
> >
> > Either your code is invalid (e.g., undefined behavior), numerically
> > unstable, or your vendor has a bug that you should report.
>
>    So first run with valgrind to find any memory issues
> http://www.mcs.anl.gov/petsc/documentation/faq.html#valgrind
>
>    If/when that is clean slowly start putting back compiler optimization
> options (first just something like -O2) until you have a problem.
>
>    Barry
>
> Optimization compiler flags are not all that common but do happen, though
> most of the time it is a user or PETSc bug
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20150526/8e7e6bf1/attachment.html>