[ET Trac] #2626: Multipole is not OpenMP parallelized
Gabriele Bozzola
trac-noreply at einsteintoolkit.org
Mon Aug 8 10:55:39 CDT 2022
#2626: Multipole is not OpenMP parallelized
Reporter: Gabriele Bozzola
Status: new
Milestone:
Version:
Type: enhancement
Priority: trivial
Component:
Comment (by Gabriele Bozzola):
It turns out that the problem cannot be solved with one single pragma. It is still very easy to sprinkle pragmas in the for loops in the integrators. I did that in
[https://bitbucket.org/einsteintoolkit/einsteinanalysis/pull-requests/19/multipole-add-openmp-parallelization](https://bitbucket.org/einsteintoolkit/einsteinanalysis/pull-requests/19/multipole-add-openmp-parallelization)
All the test still pass.
I used the new code to run the next checkpoint in one of my BBH simulations and I found that approximately 50 % less time is spent in the multipole thorn when using 4 threads.
--
Ticket URL: https://bitbucket.org/einsteintoolkit/tickets/issues/2626/multipole-is-not-openmp-parallelized
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.einsteintoolkit.org/pipermail/trac/attachments/20220808/46c7afec/attachment.html
More information about the Trac
mailing list