Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Assert triggering when load-balancing a PABLO tree with user data #140

Open
arnodu opened this issue Feb 4, 2021 · 5 comments · Fixed by #141
Open

Assert triggering when load-balancing a PABLO tree with user data #140

arnodu opened this issue Feb 4, 2021 · 5 comments · Fixed by #141

Comments

@arnodu
Copy link

arnodu commented Feb 4, 2021

Hi,

This assert triggers when I load balance my PABLO octree with user data.

Here is a short reproducer to compile in debug and run with at least 2 MPI :
PABLO_test3.cpp.txt

@andrea-iob
Copy link
Member

I've tried your test case and I'm able to reproduce the problem. I'm looking into it.

@andrea-iob andrea-iob linked a pull request Feb 8, 2021 that will close this issue
@andrea-iob
Copy link
Member

Pull request #141 should fix the problem. It should also fix the compact-families partitioning on three dimensional trees. Could you please check if the problem is now fixed (the changes are on the "PABLO.fix.family.partitioning" branch)?

We have added your test case among the integrations test of PABLO, are you ok with that?

@arnodu
Copy link
Author

arnodu commented Feb 8, 2021

Of course you can use the test case.

Looks like the issue is fixed on 490648c , thanks.

@arnodu
Copy link
Author

arnodu commented Feb 8, 2021

Now that this issue is out of the way, we still have issues when load balancing results in MPI processes with 0 octants. Probably a totally different issue, though.

For instance when i try to adapt() when a process has 0 octants, I get a segfault at LocalTree.cpp#L1933

@andrea-iob
Copy link
Member

We found a problem with empty partitions, it think this is the same issue you are experiencing. I've updated the branch "PABLO.fix.family.partitioning" (I did a force updated, if you have already pulled from that branch you need to reset your local branch to the online version). We don't use that much the partitioning routine that keeps families compact, thanks for helping us fixing it.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants