[QE-users] a possible bug in QE 6.2.2

[Paolo Giannozzi]

On Wed, Aug 26, 2020 at 5:05 PM Ranasinghe, Jayangani <jir520 at mail.usask.ca>
wrote:

I am using qe 6.2.2 for calculating electron relaxation time using with the
> epw package.
>

really? there is no such version: qe 6.2.1 maybe?

About your problem: the 6.2.1 version is almost 3 years old. Nobody has the
time to look for bugs in an old version.

By the way: memory leaks are often the result of sloppy coding, but
sometimes compilers or libraries are responsible
Also note that the true amount of memory used by a code is one of the
deepest mysteries of IT.

Paolo


> This calculation needs scf, ph, nscf and epw programs to run to the given
> order. Because of this calculation is computationally demanding and has
> many factors to optimize to get a convergent result, I wanted to use three
> different supercomputer clusters. In order to check the consistency of the
> result I ran the same calculation on the three computers. However, I
> noticed the memory usage on each cluster is totally different.
>
> Computer 1:
>
> Memory Utilized: 8.36 GB (estimated maximum)
> > Memory Efficiency: 2.23% of 375.00 GB (187.50 GB/node)
>
> Computer 2:
>
> Memory Utilized: 51.64 GB (estimated maximum)
> > Memory Efficiency: 183.62% of 28.12 GB (1.76 GB/core)
>
> Computer 3:
>
> >Memory Utilized: 174.14 GB
> > Memory Efficiency: 93.36% of 186.52 GB
>
>
>
> Further, ph.x calculation is crashed in the third computer. This program
> needs to create 10 madyn files for my input. However, the third computer
> generated only 4 matdyn files, but it finished the ph.out file without any
> crashing. In the other two computers, this did not happen (but still the
> memory usage is different).
>
>
>
> Our computer support says this could happen due to a bug in the QE
> program. Could you please let me know if there is a problem in this version
> of QE or any other reason for the issue. Thank you.
>
> _______________________________________________
> 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



-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200826/c84b5163/attachment.html>