[petsc-dev] Deadlock VecScatterCreate

Pierre Jolivet pierre.jolivet at enseeiht.fr
Sun Sep 9 23:03:00 CDT 2018

OK, thanks to the both of you.
It now makes more sense to me why you do all these copying around in PCTELESCOPE, e.g., http://www.mcs.anl.gov/petsc/petsc-current/src/ksp/pc/impls/telescope/telescope.c.html#line469 <http://www.mcs.anl.gov/petsc/petsc-current/src/ksp/pc/impls/telescope/telescope.c.html#line469>.


> On 10 Sep 2018, at 5:14 AM, Jed Brown <jed at jedbrown.org> wrote:
> "Smith, Barry F." <bsmith at mcs.anl.gov> writes:
>>     It is not possible to do this. When communicating parallel to parallel the communicator for both vectors must be the same communicator. 
> Note that one can use VecPlaceArray as part of the communication so you
> don't need a copy to get it on a different subcommunicator.  Perhaps
> that could be made automatic at a future date, after the ongoing
> VecScatter refactoring has completed.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20180910/5e7707ba/attachment.html>

More information about the petsc-dev mailing list