VASP 6.4.3 produces oversized/corrupted WAVECAR

[muhammadsufyan_razman]

Dear all,

I am encountering a strange issue with WAVECAR files generated by VASP 6.4.3 on our national cluster and would appreciate any suggestions.

Issue:
Using the same VASP input files, I obtain dramatically different WAVECAR sizes on two clusters (national and local):

1. VASP 6.4.3 on our National cluster --> 31G May 19 00:44 WAVECAR
1 F= -.34948932E+03 E0= -.34948932E+03 d E =-.349489E+03 mag= 0.0002

2. VASP 6.4.2 on our local university cluster --> 6.8G May 19 02:02 WAVECAR
1 F= -.34948931E+03 E0= -.34948931E+03 d E =-.349489E+03 mag= 0.0002

Both HPCs use the same compiler/numerical libraries and MPI.
1) gcc 2) tbb/2021.13 3) compiler-rt/2024.0.2 4) mkl/2024.0 5) mpi/2021.14

The total energy of the system remains the same, but the WAVECAR file generated by VASP 6.4.3 is significantly larger and cannot be read by any postprocessing script to unfold the band structure.
Any help with this issue would be greatly appreciated.

Best,
Sufyan

[zahedzx]

Hi,

Could you please share the relevant input files (INCAR, KPOINTS, and POSCAR) as well as the OUTCAR files from both calculations? In particular, it would be useful to compare:

NBANDS, NKPTS, ISPIN, Any SOC or noncollinear settings
And the VASP build information reported in the OUTCAR

Please also share the job submission scripts, as differences in the execution environment or parallelization settings can sometimes influence the defaults chosen by VASP.

If possible, please provide all relevant files in a single ZIP archive. For guidance on preparing a minimal reproducible example, please see:
https://www.vasp.at/wiki/index.php/Mini ... le_example

Best regards,
Zahed