vasp 6.4.2 GW+SBE erro

[liang-wen_pi]

Hello,

I'm currently using VASP version 6.4.2 to calculate absorption spectra with GW+BSE. However, I keep encountering issues when calculating the GW part. The same problem occurs regardless of whether I use the one-step GW or two-step GW method. I've attached an image. Could an expert please advise on this? I have checked both memory and disk space, and there is no issue with them.

forrtl: severe (64): input conversion error, unit -5, file Internal List-Directed Read
Image PC Routine Line Source
vasp_std 0000000001A2A5D1 Unknown Unknown Unknown
vasp_std 0000000001A29B3F Unknown Unknown Unknown
vasp_std 00000000004D1284 Unknown Unknown Unknown
vasp_std 00000000004D0C84 Unknown Unknown Unknown
vasp_std 000000000158D411 Unknown Unknown Unknown
vasp_std 00000000019E6F0E Unknown Unknown Unknown
vasp_std 00000000019BAFCE Unknown Unknown Unknown
vasp_std 000000000040A40D Unknown Unknown Unknown
libc.so.6 00001535E7033D68 Unknown Unknown Unknown
libc.so.6 00001535E7033E25 __libc_start_main Unknown Unknown
vasp_std 000000000040A321 Unknown Unknown Unknown

THANKS!

[ferenc_karsai]

Please provide all necessary files according to the forum guidelines.

[liang-wen_pi]

Thank you for your reply. I have now provided all the files.
I couldn't find any incorrect explanations about it, so I'm not sure where the problem lies.

[ferenc_karsai]

I cannot reproduce the error with vasp 6.4.2.

I've tried with the following toolchain:
gfortrain 11.2
Open-mpi 4.1.2
Intel mkl
scalapack 2.1.0

Please try if possible with the same or similar toolchain.

Which toolchain have you used?

[liang-wen_pi]

THANKS FOR YOUR REPLY! I found it seems only happens in GW/BSE calculation.
The same erro appears in another case. My inputfiles are as follows.
There is the toolchain:
mpirun --version
Intel(R) MPI Library for Linux* OS, Version 2021.16 Build 20250513 (id: a7c135c)
Copyright 2003-2025, Intel Corporation.

gcc --version
gcc (Debian 14.2.0-19) 14.2.0
Copyright (C) 2024 Free Software Foundation, Inc.

and in makefile.include,it shows : # Intel MKL (FFTW, BLAS, LAPACK, and scaLAPACK)
# (Note: for Intel Parallel Studio's MKL use -mkl instead of -qmkl)
FCL += -qmkl=sequential
MKLROOT ?= /opt/intel/oneapi/mkl/2025.2
LLIBS += -L$(MKLROOT)/lib/intel64 -lmkl_scalapack_lp64 -lmkl_blacs_intelmpi_lp64
INCS =-I$(MKLROOT)/include/fftw

Is it a bug in this version? I have recompiled VASP twice already.

[ferenc_karsai]

So I've tried also your other example, works without any problems. There is no such bug known for VASP 6.4.2.
I used an intel toolchain, since you also seem to use an intel toolchain (at least your MPI is intel).
Here is my toolchain:
Intel-Fortran 2022.0.1
MKL 2022.0.1
Intel MPI 2021.5.0

The error message you use looks rather like an IO error. So maybe your file system has an issue.
So you should try first different toolchains (prefereably the one I used) and then maybe a different file system if you have access.

[andreas.singraber]

Hello!

I just realized that this is the same issue as posted here: https://www.vasp.at/forum/viewtopic.php?t=20555

I am currently also looking into this... let's continue the discussion in the new topic, closing this now...

All the best,
Andreas Singraber