[petsc-users] MPI bindings for PETSc local and global numbering

[Justin Chang]

"-map-by socket" was for the non-sequential ordering I had. I got
sequentially ordering with "-bind-to
user:0,1,2,3,4,5,6,7,10,11,12,13,14,15,16,17"

On Mon, May 11, 2015 at 4:07 PM, Jed Brown <jed at jedbrown.org> wrote:

> Justin Chang <jychang48 at gmail.com> writes:
> > ${MPIEXEC}.hydra -n <procs> -bind-to hwthread -map-by socket ./myprogram
> > <args>
>
> Why are you using "-map-by socket" if you want a sequential ordering?
>
> Abstractly, I expect a tradeoff between load balancing in case of
> correlated irregularity and cache locality, but would tend to prefer
> sequential ordering (so that nearby row numbers are more likely to share
> cache).
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20150511/f4db320c/attachment.html>