[Wien] Error in parallel LAPW2

Gavin Abo gsabo at crimson.ua.edu
Sat Mar 27 20:15:25 CET 2021


Below, you mention you are using Intel 2018.5.274.

Don't know if it helps or not, but what is the file size of the 
case.vector_1 for the failed case (+10) compared to the working ones 
(-10, -5, 0, 5)?

There is this Intel post:

https://community.intel.com/t5/Intel-Fortran-Compiler/IO-error-67-input-statement-requires-too-much-data/td-p/1151117?profile.language=en

It seems to suggest that with Intel 2018 (18.x) versions there was no 
problems with small files, but then the "input statement requires too 
much data" error would happen with larger file sizes.  There responding 
post hints a bug maybe not being fixed until the Intel 2019 (19.x) 
compilers with 19.0.2 being mentioned in particular.

On 3/27/2021 12:29 PM, Anupriya Nyayban wrote:
> Dear Prof. Marks, Fecher and Gavin
>
> ,Thanks a lot for your valuable additional comments and help!!
>
> I have checked the dayfile and case.output2 file but am not able to 
> find any clue about why the program has been stopped in my case. I 
> have attached the dayfile below.
>
> Yes, you are correct. P-1 is obtained over some refinements on lower 
> symmetry (P). I will surely consider this aspect and recalculate it.
>
> I have checked the scf files for -10, -5 , 0 and +5. Several warnings 
> can be found (for e.g.):
> warn: P(J, JATOM) almost zero (-0.975E-0.3) shifted energy by 0.01 
> down to 0.3583.
> warning: error in reading files restarted (after the convergence of 
> the difference in charges is done, I am wondering why this warning is!).
> ENE: warning total energy (even this can be seen in the last few 
> cycles for -10 and -5 wheras for +5 , error in reading files can be 
> found without the total energy warning).
>
> Still, I am worried why the calculation has been stopped for my case only.
>
> On Sat, 27 Mar 2021 at 21:13, Anupriya Nyayban <mamaniphy at gmail.com 
> <mailto:mamaniphy at gmail.com>> wrote:
>
>     Dear Prof. Marks, Fecher and Gavin
>
>     ,Thanks a lot for your valuable additional comments and help!!
>
>     I have checked the dayfile and case.output2 file but am not able
>     to find any clue about why the program has been stopped in my
>     case. I have attached these two files below.
>
>     Yes, you are correct. P-1 is obtained over some refinements on
>     lower symmetry (P). I will surely consider this aspect and
>     recalculate it.
>
>     I have checked the scf files for -10, -5 , 0 and +5. Several
>     warnings can be found (for e.g.):
>     warn: P(J, JATOM) almost zero (-0.975E-0.3) shifted energy by 0.01
>     down to 0.3583.
>     warning: error in reading files restarted (after the convergence
>     of the difference in charges is done, I am wondering why this
>     warning is!).
>     ENE: warning total energy (even this can be seen in the last few
>     cycles for -10 and -5 wheras for +5 , error in reading files can
>     be found without the total energy warning).
>
>     Still, I am worried why the calculation has been stopped for my
>     case only.
>
>
>
>     On Fri, 26 Mar 2021 at 12:10, Anupriya Nyayban
>     <mamaniphy at gmail.com <mailto:mamaniphy at gmail.com>> wrote:
>
>         Dear Prof. Blaha,
>
>         Previously, I have followed the steps as:
>         deleted the case.struct file
>         copied the struct file for +10 as case.struct
>         x dstart
>         run_lapw -I -fc 10.0 -p
>         And, I have got the message as "forrtl: severe (67): input
>         statement requires too much data, unit 10,
>         file/case/./case.vector_1" at the first cycle.
>
>         Now, I have created a new case directory and saved the
>         +10.struct as case.struct. Initialization has been done with
>         RKmax = 7.0 and Kmesh = 150. The same message could be seen at
>         the beginning when ""run_lapw -p -fc 10.0" has been executed.
>
>         Here, the struct file for +10 is attached below.
>
>
>
>         On Thu, 25 Mar 2021 at 12:34, Anupriya Nyayban
>         <mamaniphy at gmail.com <mailto:mamaniphy at gmail.com>> wrote:
>
>             Dear Prof. Blaha,
>
>
>             Thank you very much for the help!!
>
>             First, I have activated both min and run_lapw in
>             optimize.job to find the energy of the relaxed one, and
>             could realize the serious mistake now.
>
>             Second, yes, the calculation crashes in the first cycle
>             for +10.
>
>             Third, I have run x dstart, run_lapw -I -fc 10.0 -p for
>             +10 and found the following message at the first cycle:
>              "forrtl: severe (67): input statement requires too much
>             data, unit 10, file/case/./case.vector_1".
>
>             May I find the volume optimization with a smaller RKmax
>             value to avoid the large data error and later I can have
>             scf with the optimized lattice parameters. converged RKmax
>             and Kmesh?
>
>
>
>
>             On Wed, 24 Mar 2021 at 17:42, Anupriya Nyayban
>             <mamaniphy at gmail.com <mailto:mamaniphy at gmail.com>> wrote:
>
>                 Dear experts and users,
>
>                 In addition to the above information, I want to
>                 mention that commands used in optimize.job script are
>                 "min -I -j "run_lapw -I -fc 1.0 -i 40 -p"" and
>                 "run_lapw -p -ec 0.0001". The RKmax and kmesh are set
>                 to 7.0 and 150 respectively.  The energy versus volume
>                 graph (fitted to Murnaghan equation of state) looks
>                 very different from the usual. I am not getting any
>                 idea why lapw2 crashes (error in paralle lapw2 is
>                 shown in lapw2.error) for +10% of change in volume. I
>                 need your valuable suggestions to proceed with the
>                 calculation.
>
>
>
>
>                 On Fri, 19 Mar 2021 at 00:39, Anupriya Nyayban
>                 <mamaniphy at gmail.com <mailto:mamaniphy at gmail.com>> wrote:
>
>                     Dear experts and users,
>
>                     I was calculating the volume optimization in
>                     parallel (with 8 cores) of an orthorhombic 2*2*1
>                     supercell having 80 atoms (in the supercell) in a
>                     HPC (Processor: dual socket 18 core per socket
>                     intel skylake processor, RAM: 96 GB ECC DDR4 2133
>                     MHz RAM in balanced configuration, Operating
>                     system: CentOS-7.3, using compiler/intel
>                     2018.5.274). The changes in volume were set to
>                     -10, -5, 0, 5, 10 (in %). I could find error only
>                     in lapw2.erro which states "error in parallel
>                     lapw2". The scf calculations have been completed
>                     for the volume changes of -10, -5, 0, 5%.
>
>                     Looking forward for your suggestion.
>                     If you need any additional information please let
>                     me know.
>
>                     Thank you in advance.
>
>                     -- 
>                     With regards
>                     Anupriya Nyayban
>                     Ph.D. Scholar
>                     Department of Physics
>                     NIT Silchar
>
>
>
>                 -- 
>                 With regards
>                 Anupriya Nyayban
>                 Ph.D. Scholar
>                 Department of Physics
>                 NIT Silchar
>
>
>
>             -- 
>             With regards
>             Anupriya Nyayban
>             Ph.D. Scholar
>             Department of Physics
>             NIT Silchar
>
>
>
>         -- 
>         With regards
>         Anupriya Nyayban
>         Ph.D. Scholar
>         Department of Physics
>         NIT Silchar
>
>
>
>     -- 
>     With regards
>     Anupriya Nyayban
>     Ph.D. Scholar
>     Department of Physics
>     NIT Silchar
>
>
>
> -- 
> With regards
> Anupriya Nyayban
> Ph.D. Scholar
> Department of Physics
> NIT Silchar
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20210327/3b08c058/attachment.htm>


More information about the Wien mailing list