[CP2K-user] cp2k gets stuck in high performance computer

Alfio Lazzaro alfio.... at gmail.com
Thu May 30 19:24:42 UTC 2019


I cannot judge on your inputs (I'm not an expert), however I see that you 
are using a quite old CP2K version (3.0?). Furthermore, you are not saying 
anything about how CP2K is compiled (I see POPT with Intel compiler).
A way to see where it is hanging is to add TRACE keyword to your input, 
inside the GLOBAL section. My suspicious is that there is something wrong 
with an MPI call, likely too big messages and that's why you don't see the 
problem when you use more nodes. In any case, I suggest to to use a new 
CP2K version...

Alfio



Il giorno giovedì 30 maggio 2019 20:54:49 UTC+2, Xiaoliu ha scritto:
>
> Hi all,
>
> I met a problem when running cp2k in Louisiana optical Network 
> Infrastructure (Loni).When submitting input file, the output file gets 
> stuck after a certain line while the job is not killed if 4 or 6 nodes are 
> used. However, if I use 1,2, 8, 11, 12, 16, 24, and even 85 nodes, it works.
>
> Also, when starting from restart file, cp2k gets stuck with 12 nodes (I'm 
> testing other numbers of nodes also, and they're in queue).
>
> My system is acetic acid and methylimidazole mixture in a 19 A cubic box 
> under Quickstart, nvt.
>
> Does anyone have any idea how to solve this problem?
>
> Regards,
> Xiaoliu
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20190530/7deaa9a0/attachment.htm>


More information about the CP2K-user mailing list