intel bug with CP2K 5.1 version?

[Be Brave]

Hi Alfio, 

we I asked my HPC administrator to instal the new cp2k 5.1, they also used 
*easybuild*, but they obtained several errors after running the *make test. 
*So I installed the cp2k by myself using the toolchain script which will 
basically do everything. You can find toolchain in the instalation package 
you download from cp2k repo (cp2k-5.1/tools/toolchain/) - use stable 
release and not the current trunk. After that I got 0 error in the test 
results.

Except MPI and Intel MKL libraries (which I loaded via modules), the 
toolchain installed all other libraries.
So my installation on HPC cluster was like: 

module load MPICH/3.2-GCC-5.3.1-snapshot-20160419-2.25
module load imkl/2017.1.132-iimpi-2017a
./install_cp2k_toolchain.sh -j 8 --install-all --mpi-mode=mpich 
--math-mode=mkl --enable-omp=no --with-binutils=system --with-mkl=system 
--with-gcc=system --with-mpich=system --with-binutils=system  
--with-cmake=system

Best.
 Jan.


Dne čtvrtek 18. ledna 2018 14:06:19 UTC+1 Alfio Lazzaro napsal(a):
>
> Hello Phil,
> then, I would suggest to ask to your HPC system administrator to test CP2K 
> installation. There is nothing we can do in this forum for you :(
>
> Best regards,
>
> Alfio
>
>
>
> Il giorno mercoledì 17 gennaio 2018 14:43:22 UTC+1, Phil G. ha scritto:
>>
>> Hello Alfio,
>>
>> thank you for your reply. I think it is a problem with the new module. 
>> Our HPC is a slurm-type computer system (HRSK-II) and it has the EasyBuild 
>> environment. 
>> I use batch-file (*.sh) in order to run the DFT calculation or MD 
>> simulation on HPC after submitting input and batch files via WinSCP and 
>> PuTTY program (SSH connection with X11 forwarding). 
>>
>> Kind regards,
>>
>> Phil
>>
>> Am Dienstag, 16. Januar 2018 14:50:12 UTC+1 schrieb Alfio Lazzaro:
>>>
>>> Hello Phil,
>>> I tried your input with CP2K 5.1 compiled with GNU, PSMP version, 
>>> running with 288 ranks and 1 thread, and I cannot get your message.
>>> The only difference is that I changed the 
>>>
>>> <     BASIS_SET_FILE_NAME 
>>> /lustre/scratch2/s2047290/2018/LNpos4x4_H2O_MD_201801_1/BASIS_MOLOPT 
>>> <     POTENTIAL_FILE_NAME 
>>> /lustre/scratch2/s2047290/2018/LNpos4x4_H2O_MD_201801_1/POTENTIAL 
>>> --- 
>>> >     BASIS_SET_FILE_NAME BASIS_MOLOPT 
>>> >     POTENTIAL_FILE_NAME POTENTIAL
>>>
>>> But I don't think this is the reason for the error...
>>> Therefore it looks like a problem with your new module 5.1-intel-2017b. 
>>> However, it is not clear to me where you are getting this module and how it 
>>> is compiled (is it easybuild? which arch file are you using? which compiler 
>>> version?).
>>> I would recommend running the regtest (if you haven't already) to check 
>>> if the CP2K installation is correct (see 
>>> https://www.cp2k.org/dev:regtesting)
>>>
>>> Alfio
>>>
>>>
>>> Il giorno lunedì 15 gennaio 2018 16:43:56 UTC+1, Phil G. ha scritto:
>>>>
>>>> Dear CP2K fans,
>>>>
>>>> since our HPC gets the new module version of CP2K (5.1-intel-2017b), 
>>>> after submitting the batch file and starting the MD, the simulation was 
>>>> stopped/aborted due to the error message: "Dipole correction needs more 
>>>> vacuum space above the surface"
>>>> This error does not occur if the MD simulation was run with older 
>>>> module version of CP2K (3.0). Is there a bug in the new module file of CP2K 
>>>> 5.1-intel-2017b ?
>>>>
>>>> Here the files are attached here.
>>>>
>>>> Kind regards,
>>>>
>>>> Phil
>>>>
>>>>  
>>>>
>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180124/77c06ad2/attachment.htm>