[QE-users] QE 7.2: performance degradation for isolated systems?

[Tom Demeyere]

Dear users,

Yesterday I wanted to perform a small calculation on a centred Pt13 nanoparticle in a 20 Å box, I tried QE 7.1 and QE 7.2 with the exact same input, hardware and parallelisation parameters. 7.2 end up being much slower:

QE 7.1:

15th iter: 100.5 secs
30th iter: 184.2 secs

QE 7.2:

15th iter: 463.5 secs
30th iter: 824.4 secs

But QE 7.2 converge the SCF in fewer steps (still ends up being much slower). I reproduced this behaviour on 2 other machines and playing with parallelisation parameters doesn’t change anything. How to explain this?

Best regards,
Tom Demeyere



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20230428/76ae458b/attachment.html>