Running CDFT Tutorial Calculation on Cluster

Nico Holmberg holmbe... at gmail.com
Fri Jul 27 10:30:09 UTC 2018


Hi Brian,

The cube files you attached seem to be quite a bit smaller than I get when 
I run the tutorial (12 vs 25 Mb), which is likely the cause of the error 
message you received in question 1. For some reason, likely related to the 
error you described in question 2, the cube file is only partly written to 
disk. I don't understand what you mean by "It would simply output a blank 
file, and then break. I was able to fix this by passing in the blank file 
of the appropriate name (giving me the files above), but it seemed like an 
odd error."

In any case, an MPI parallelized cube file writer was introduced in CP2K 
6.1. It has been tested on different platforms and compilers (including 
Intel ifort), though not with this particular input file, but it is 
possible that some bug still remains. In order to debug this behavior, it 
would be helpful if you could post more details about how the CP2K 6.1 
binary was compiled and how you ran the calculation. In particular, the 
following would be helpful

   1. Compiling environment including version numbers
   2. Number of MPI processes used in the calculation. Does the error 
   related to printing out the cube file disappear if you use just 1 MPI 
   process?'
   3. Detailed output of the crashing simulation with debugging symbols 
   turned on in the binary (your computing center might not provide one by 
   default)

I have different Intel compiler versions available to me on a cluster so I 
can try to reproduce the error you're seeing if you can figure out the 
compiler version.


BR, 

Nico

  

keskiviikko 25. heinäkuuta 2018 22.59.27 UTC+3 Brian Day kirjoitti:
>
> Hi Nico,
>
> I was able to finish the tutorial by generating those electron density 
> cube files in a separate step with an older version of CP2K, and then using 
> them to complete the tutorial. For whatever reason, v6.1 seems to have 
> issues with generating these cube files. If this is an error you are 
> familiar with and have any compiling advice I can pass on to our computing 
> center, that would be great, as they seem unsure of how to handle it. 
> Regardless, Thanks again for your help, and appreciate you updating the 
> tutorial for the new version. 
>
> Best,
>      Brian
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180727/b877fc2a/attachment.htm>


More information about the CP2K-user mailing list