[QE-users] SIGSEGV fault in phonon calculation

[Dominik Gresch]

Dear Ruochen Shi,

A segmentation fault generally does not occur because of lack of
available memory. In that case, you would see an error like "out of
memory, killed", or "killed (signal 9)" - the details of which depend on
your OS, MPI library, and possibly scheduler.

Segmentation faults instead occur when a program tries to access memory
that it is not allowed to - because it is not allocated for this
process, see https://en.wikipedia.org/wiki/Segmentation_fault

In general this is a "program error", meaning there is a fault in your
compiled executable. This does /not/ necessarily mean that it is a
problem with Quantum ESPRESSO - in fact, most of these problems occur
due to bugs in the underlying (MPI, BLAS, FFT) libraries or how they
were compiled and linked.

Since there are a million places in which a segfault can occur, it is
basically impossible to debug remotely. Your options are:

- try running the QE test suite. If you see segfaults there, it is
almost guaranteed to be a library / compilation issue
- compile the code with lower optimization level, enabling debug
statements, bounds checking, ...
- make sure the libraries used are compiled and linked correctly, or try
using different ones
- use a debugger, like gdb. This article
https://www.cprogramming.com/debugging/segfaults.html might be a decent
introduction

Best regards,

Dominik Gresch

---

Simulation Engineer, Microsoft Quantum

On 20.05.20 18:04, 时若晨 wrote:
>
> Dear all,
>
> I'm a new user to quantum-espresso, and recently i've encontered a
> problem when calculating phonon frequency in a heterojunction
> structure containing 24 atoms in an unit cell. A "forrtl: severe
> (174): SIGSEGV, segmentation fault occurred" error happens, which is
> frequently asked by users. I suppose this is beacause there's too many
> atoms in the unit cell, such that RAM need excesses the RAM provided,
> since I ran a calculation on a similiar structure but whit 12 atoms in
> the unit cell, and it ran all very well. The self-consistence-field
> calculation for this structure ran well,too. I've tried some ways,
> such as set 'verbosity' in the ph.in file to 'minimal', and add
> 'ulimit -s unlimited' in the .bashrc. However, no one works.
>
> So I wonder is there a way to overcome this problem?
>
> My input and output file for scf and ph at Gamma point is attached.
> The text for error is also attatched.
>
>
> Ruochen Shi
>
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200521/c1885352/attachment.html>