[QE-users] Supercell Ph.x Slow Parallel Calculation

[Tang, Weilun]

Dear Quantum Espresso Community,

I am doing phonon calculation for a system of molecules absorbed on a metal surface. I used a 4x4 supercell (A = 20.38, B = 11.77, C = 14.41. This 4x4 supercell is chosen to minimize the interaction between periodic images. The metal surface lies on xy-plane.), and it has a total of 54 atoms. I used 8x8x1 k-point grid. This 8x8x1 k-point grid is the minimum and I don't use any symmetry (These two are required for my own purposes).

I found that ph.x is extremely slow with 2 nodes (128 processors each). It crashed with 2 nodes, so I used -pd .true. option. I used more nodes like 6 nodes, but it didn't speed up the calculation (maybe due to network in the cluster).

Is there any better way to handle such a large system to speed up the calculation?

Thank you for your help and your time in advance,

Weilun Tang

Department of Materials Science and Engineering
Iowa State University


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20260127/560648a6/attachment.html>