[Wien] Lapwso crashes with Intel 2018 update 1

[Osama Yassin]

Dear Gavin


- Compilation with the option -assume buffered_io did not solve the problem.

- By using the file get_nloat.f<https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16589/get_nloat.f> provided by Eamon McDermott, the nloat message disappeared but a new error appeared after lapwso end

L2main - QTL-B Error

<http://susi.theochem.tuwien.ac.at/reg_user/faq/qtlb.html>Thus I conclude that it is because of compilation issue related to Intel FC 2018 update 1.

Still working on it and looking for your help.

Osama

QTL-B - FAQ - WIEN2k<http://susi.theochem.tuwien.ac.at/reg_user/faq/qtlb.html>
susi.theochem.tuwien.ac.at
In some cases (mostly when the spheres are extremely small or you have heavy elements) it may even happen, that during the (first or later) scf-cycle LAPW2 stops with ...





Sent from Outlook<http://aka.ms/weboutlook>
________________________________
From: Wien <wien-bounces at zeus.theochem.tuwien.ac.at> on behalf of Gavin Abo <gsabo at crimson.ua.edu>
Sent: Saturday, January 13, 2018 7:04:36 PM
To: wien at zeus.theochem.tuwien.ac.at
Subject: Re: [Wien] Lapwso crashes with Intel 2018 update 1


It does look like that might be caused by the Intel compiler version (2018 update 1) that you are using.


Have you tried compiling with -assume nobuffered_io [ https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16802.html ]?


increase nloat in module -> This likely means the nloat value of the MODULE param block in SRC_lapwso/modules.F.  However, that comment might be old such that I don't recommend changing the value by hand, because it looks like the code in get_nloat.f now automatically adjusts the nloat value for you.


Have you tried Eamon's workaround fix for get_nloat.f [ https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16589.html ]?

On 1/12/2018 11:12 PM, Osama Yassin wrote:

Dear Gavin Abo


- I have installed Intel Compiler 2108 update 1. Then Wien2k 17.1 was complied successfully (serial and parallel).

- Normal SCF cycles completed successfully using SCAN meta-GGA or PBE-GGA.

- SOC was initiated from the command line using initiso_lapw. The case.inso is



WFFIL
4  0  0                 llmax,ipr,kpot
-10  1.5                Emin, Emax
    0 0 1                           h,k,l (direction of magnetization)
 0                       number of atoms with RLO
0 0      number of atoms without SO, atomnumbers

- the case.in1 is:



WFFIL  EF=.61137685760000000019   (WFFIL, WFPRI, ENFIL, SUPWF)
  7.00     11   4   ELPA pxq hm (R-MT*K-MAX,MAX L IN
  0.30    4  0      (GLOBAL E-PARAMETER WITH n OTHER CHOICES, global APW/LAPW)
 1    0.30     0.0000 CONT 1
 1   -3.77     0.0001 STOP 1
 2    0.30     0.0010 CONT 1
 0    0.30     0.0000 CONT 1
K-VECTORS FROM UNIT:4   -9.0       5.0    19   emin / de (emax=Ef+de) / nband

- The lapwso crashes

>   stop error

[1]  + Done                          ( cd $PWD; $t $exe ${def}_${loop}.def; rm -f .lock_$lockfile[$p] ) >> .timeso_$loop
 increase nloat in module  (atpar --- lo part)          53          52
[1]  + Done                          ( cd $PWD; $t $exe ${def}_${loop}.def; rm -f .lock_$lockfile[$p] ) >> .timeso_$loop
[2]  + Done                          ( cd $PWD; $t $exe ${def}_${loop}.def; rm -f .lock_$lockfile[$p] ) >> .timeso_$loop
TOO MANY BANDS:

- changing the number of bands does not succeed.

- On another run I get this: Here i used mBJ+soc. Previously this case was successful with intel 2017.


lapwso             00000000004043AA  Unknown               Unknown  Unknown
libc-2.24.so       00001460B54A0431  __libc_start_main     Unknown  Unknown
lapwso             00000000004044AE  Unknown               Unknown  Unknown
lapwso             000000000042043E  MAIN__                    591  lapwso.F
lapwso             0000000000414D18  hmsec_                    231  hmsec.F
lapwso             000000000040DDE7  abclm_                    187  abclm.F
lapwso             0000000000421B94  lomain_                    55  lomain.F
libpthread-2.24.s  00001460B611D5D0  Unknown               Unknown  Unknown
lapwso             000000000044490D  Unknown               Unknown  Unknown
Image              PC                Routine            Line        Source
forrtl: severe (174): SIGSEGV, segmentation fault occurred
TOO MANY KJS:

- What does increase nloat in module .....

best wishes






Sent from Outlook<http://aka.ms/weboutlook>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20180115/66a36114/attachment.html>