On 2021-06-28 18:45, John Peterson wrote:
Hi Edgar,
Including your previous results, the ex4 patch alone still has the
fastest
---8<--- snip
Great! it means that your change really works.
But since you rebuilt libmesh between then and now, I'm not sure we
should
really compare the two.
Precisely.
---8<--- snip
in parallel is probably counter productive to our goals since:
1.) It reduces the overall active time,
---8<-- snip
it is affected by the timing code
itself.
2.) It will introduce more variability in the results. Currently,
the coefficient of variation (mean divided by stddev) for these results
is
on the order of 15-18%, a fair bit larger than the differences in the
times
themselves.
Right.
If you are interested in investigating this further, I would suggest
that
---8<--- snip
Although it would be interesting, it sounds like a task for someone with
a tad more time than me :P (at the moment). At the end, I am quite sure
that the changes which you introduced make the code more efficient. This
task would surely bring more formality to it.
Thanks!
_______________________________________________
Libmesh-users mailing list
Libmesh-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/libmesh-users
