Optimizing Parallelization Parameters (NBANDS, NCORE, etc.) for DOS Calculation of a 99-Atom Nanocluster

[farah_shahzadi]

I'm currently performing a density of states (DOS) calculation for a nanocluster consisting of 99 atoms. I've set the k-points to 2x2x2. However, my calculations either get stuck or stop without completing. I'm seeking advice on how to properly set the parallelization parameters such as NBANDS, NCORE, and any other relevant tags to ensure that the calculation completes successfully and runs efficiently.
Here are some details about my setup:
Cluster with 40 cores available.
Current k-point setting: 2x2x2.

Could you please guide me on the optimal settings for NBANDS, NCORE, and other parameters to speed up the calculation and prevent it from getting stuck? Any insights or recommendations would be greatly appreciated.

Thank you!

[michael_wolloch]

Dear farah_shahzadi,

Please provide your input (and important output) files as a zipped folder, as stated in the forum guidlines. Without them, it is more difficult to recommend good settings.

Since you are writing about a nanocluster, I assume you have vacuum on all sides of your cell. In that case, the Γ-point is sufficient for k-point sampling and you would gain better performance using the gamma-only version of VASP.

Michael

[farah_shahzadi]

Thank you for your advice! I have prepared a zipped folder containing all the relevant input and important output files, which you can find attached. Please take a look and let me know if you have any recommendations for improving my settings.

[michael_wolloch]

Dear farah_shahzadi,

the POSCAR you sent is incomplete, and POTCAR, OUTCAR, and stdout are missing. I cannot help you properly if I do not have the information I need.

Cheers, Michael