CP2K Regtesting on Intel Compiler

Matthias Krack matthia... at psi.ch
Thu Oct 24 08:56:24 CEST 2013


Hi Rafa,

you can download the reference outputs of the most recent CP2K trunk 
version used by the automatic regression tester here<http://people.web.psi.ch/krack/cp2k/regtest/trunk_LAST-Linux-x86-64-g95-popt.tar.bz2>and the reference outputs for the CP2K version 2.3 and 2.4 are also 
available via http://people.web.psi.ch/krack/cp2k/regtest/regtest.html
These outputs will differ from your results, because they were generated 
using a different compiler (g95) and hardware combination and the requested 
numerical tolerance for many test inputs is quite tight. Therefore I would 
focus on the test inputs which are testing features of CP2K you are 
interested in.

Matthias


On Wednesday, October 23, 2013 9:56:28 PM UTC+2, Rafael Soler-Crespo wrote:
>
> Hi Matthias,
>
> As you mentioned, I was patient and the process occurred normally. Now, I 
> am getting that some tests failed and some tests were NEW. Of course, this 
> is because I don't have a reference to test against. Where can I find this 
> reference to check if my code is compiled correctly? I tried looking at the 
> header in *do_regtest* but I didn't find anything.
>
> Thanks,
> Rafa
>
> On Tuesday, October 22, 2013 1:50:48 AM UTC-5, Matthias Krack wrote:
>>
>> Hi Rafa,
>>
>> the generation of the first Changelog may take 10-20 min and in this time 
>> no change occurs in the folders. It is simply the runtime of the svncl.plscript building the Changelog for the full revision history of CP2K. The 
>> next update run will require only a few seconds. So just try to be patient 
>> for the first time as the printed comment suggests.
>>
>> Matthias
>>
>>
>> On Monday, October 21, 2013 11:50:02 PM UTC+2, Rafael Soler-Crespo wrote:
>>>
>>> Hi all,
>>>
>>> I've been trying to setup regtesting with the Intel Compiler, MPI 
>>> Version. However, for some reason, the testing script (do_regtest) hangs on 
>>> writing the Changelog for the first time. To make sure all is fine, I am 
>>> tracking the process using *$ ls -l* but no change of time occurs in 
>>> the folders associated with the regtesting. Is there some specific reason 
>>> as to why this might be happening?
>>>
>>> Thanks,
>>> Rafa
>>>
>>> P.S: I have attached my regtesting script (sections to be changed) below.
>>>
>>> # The following variable definitions can now be loaded from a
>>> # site-specific configuration using the -config option. Create one
>>> # configuration file for each architecture/compiler using these as a
>>> # template (and don't forget to include a modified cp2k_prefix for MPI
>>> # runs as well!).
>>> export LC_ALL=C
>>> export FORT_C_NAME=intel
>>> # Might consider following export to be done, too
>>> #export MPI_F90=mpif90
>>> export MPI_F90=mpiifort
>>> dir_base=$PWD
>>> cp2k_version=popt
>>> dir_triplet=Linux-x86-64-${FORT_C_NAME}
>>> export ARCH=${dir_triplet}
>>> cp2k_dir=cp2k/cp2k # Since SVN creates a cp2k folder with another cp2k 
>>> folder within it. The second one has 3 folders: potentials, basis_sets, 
>>> cp2k.
>>> maxtasks=1
>>> numprocs=1
>>> export OMP_NUM_THREADS=1
>>> emptycheck="NO"
>>> leakcheck="NO"
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20131023/813fe91b/attachment.html>


More information about the CP2K-user mailing list