For the third time, look at the title of the issue I linked. I'm not going to answer any more questions that are directly addressed in that issue because I don't have time to repeat myself so many times _and_ get useful things done.
This is higher priority now because we are using PetscSF in some more visible places. On Nov 2, 2012 5:57 PM, "Fande Kong" <fd.kong at siat.ac.cn> wrote: > Yes. I haven't tested my code on MPICH. Because MPICH don't > support infiniband in the supercomputer. But I have tested code on mvapich. > It also has bugs, when the finite mesh is larger than 50M. > > Now, I just want to know if you will change to use point-point. If you > would not, I will try to rewrite PetscSF for me. > > Thanks, > > On Fri, Nov 2, 2012 at 4:41 PM, Jed Brown <jedbrown at mcs.anl.gov> wrote: > >> >> On Nov 2, 2012 5:24 PM, "Fande Kong" <fd.kong at siat.ac.cn> wrote: >> > >> > I got it. You still didn't fix the bugs that I encountered a few months >> ago. All current MPI implementations couldn't support the function >> MPI_Accumulate or others well with one-sided communication when data is >> large. >> >> I have not had problems with MPICH. >> > > > > -- > Fande Kong > ShenZhen Institutes of Advanced Technology > Chinese Academy of Sciences > > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20121102/29e36d7a/attachment.html>
