[DFTB-Plus-User] DFTB+ 19.1 memory leakage

[Schwarz Dr., Paul (FTI)]

Dear Bálint,

thanks for your reply.
I am currently running the same simulation (same system, NVE MD) with the DFTB+ 19.1 version with OpenMP only (12 threads), and so far, I have no issues regarding memory leaks.
I will further try the MPI version with only 1 MPI process and multiple threads (standard solver) and keep you informed about the results.


Best regards
Paul



-----Ursprüngliche Nachricht-----
Von: DFTB-Plus-User <dftb-plus-user-bounces at mailman.zfn.uni-bremen.de> Im Auftrag von Bálint Aradi via DFTB-Plus-User
Gesendet: Donnerstag, 12. März 2020 17:11
An: dftb-plus-user at mailman.zfn.uni-bremen.de
Cc: Bálint Aradi <aradi at uni-bremen.de>
Betreff: Re: [DFTB-Plus-User] DFTB+ 19.1 memory leakage

Dear Paul,

I am really wondering where the memory leak comes from. We eliminated all allocated pointers in the main part of the code long ago to avoid memory leaks.

* Could you eventually try to compile a non-mpi DFTB+ without any extensions (e.g. no ELSI/ELPA) and try your system with it? Does it still show memory leaks? Or alternatively, could you try it with the binary from the website?

* If the official non-mpi binary also shows this behaviour, could you then send me (off-list) your geometry, so that I can try to reprocude the case myself?

Best regards,

Bálint

--
Dr. Bálint Aradi
Bremen Center for Computational Materials Science, University of Bremen http://www.bccms.uni-bremen.de/cms/people/b-aradi/