[QE-users] segmentation fault when writing *.save, LDA+U
Paolo Giannozzi
p.giannozzi at gmail.com
Tue Jun 18 15:11:51 CEST 2019
On Mon, Jun 17, 2019 at 6:16 PM Sergey Lisenkov <proffess at yandex.ru> wrote:
I did apply both patches (this one and backports) before compiling. And I
> don't think ifort 18.0 is "old" intel compiler.
oh well. Maybe there is some additional problem with the DFT+U case.
On our cluster we have different QE installed. I tried 5.4 version and it
> works without "segmentation fault" crash.
the crash happens in the new code writing the xml file. In v.5.4 the xml
was different and the code writing it was also different
Paolo
> 17.06.2019, 16:50, "Paolo Giannozzi" <p.giannozzi at gmail.com>:
>
> Please apply the "patch_old_intel_to_6.4.1
> <https://github.com/QEF/q-e/releases/download/qe-6.4.1/patch_old_intel_to_6.4.1>"
> file on GitHub. I have the same crash, that disappears after applying the
> patch (but I do have a really old Intel compiler!)
>
> Paolo
>
>
> On Mon, Jun 17, 2019 at 1:19 PM Sergey Lisenkov <proffess at yandex.ru>
> wrote:
>
> Hi Paolo,
>
> Attached is my input file, pseudos, and output. I use 6.4.1 (latest)
> version.
>
> I tried different intel compilers (17.0 - 19.0), different optimization
> flags, but nothing helped.
>
> Thanks,
> Sergey
>
> 17.06.2019, 11:44, "Paolo Giannozzi" <p.giannozzi at gmail.com>:
>
> Which code version are you using? could you please provide the input?
>
> On Sat, Jun 15, 2019 at 4:42 PM Sergey Lisenkov <proffess at yandex.ru>
> wrote:
>
> Hello,
>
> I am getting this segmentation fault error, when pw.x reaches the last
> point of calculation (scf), writing files into prefix.save directory, only
> when LDA+U is used.
>
> I can reproduce this error on different type of computers (sgi hpe or cray
> xc), intel compiler (17.0 and 18.0). I ran my code in parallel.
>
> I thought it would be memory issue, but number of kpoints is small (4),
> and number of cores used is large (so, enough memory).
>
> Anybody else saw this?
>
>
> Thanks,
> Sergey
>
> ==================
> the Fermi energy is 3.6767 ev
>
> ! total energy = -2615.09121470 Ry
> Harris-Foulkes estimate = -2615.09121470 Ry
> estimated scf accuracy < 4.1E-11 Ry
>
> The total energy is the sum of the following terms:
>
> one-electron contribution = -2207.29565403 Ry
> hartree contribution = 1225.21145674 Ry
> xc contribution = -443.34772677 Ry
> ewald contribution = -887.42076673 Ry
> Hubbard energy = 0.11953763 Ry
> one-center paw contrib. = -302.35806154 Ry
> smearing contrib. (-TS) = 0.00000000 Ry
>
> total magnetization = 0.00 Bohr mag/cell
> absolute magnetization = 11.96 Bohr mag/cell
>
> convergence has been achieved in 14 iterations
>
> Writing output data file mof.save/
> forrtl: severe (174): SIGSEGV, segmentation fault occurred
> Image PC Routine Line
> Source
> pw641.x2 00000000044F705D Unknown Unknown
> Unknown
> pw641.x2 0000000004314C10 Unknown Unknown
> Unknown
> pw641.x2 0000000000983C57 Unknown Unknown
> Unknown
> pw641.x2 000000000054A7A0 Unknown Unknown
> Unknown
> pw641.x2 0000000000545336 Unknown Unknown
> Unknown
> pw641.x2 00000000005877D3 Unknown Unknown
> Unknown
> pw641.x2 0000000000401567 Unknown Unknown
> Unknown
> pw641.x2 00000000004013BE Unknown Unknown
> Unknown
> pw641.x2 00000000045CEBF1 Unknown Unknown
> Unknown
> pw641.x2 000000000040123D Unknown Unknown
> Unknown
> [NID 00016] 2019-06-15 08:37:28 Apid 41484316: initiated application
> termination
> Application 41484316 exit codes: 174
>
> _______________________________________________
> 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
>
> ,
>
> _______________________________________________
> 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
>
>
--
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/20190618/d9423d83/attachment.html>
More information about the users
mailing list