[CP2K-user] Regression Test With multi-gpu

[vitesse lin]

Hi Yang,

I am using master branch.

How did you make the binary file and what is the test that you can run pass?
You can post more information about yours.


jt Yang於 2019年12月11日星期三 UTC+8下午2時26分44秒寫道：
>
> Hi lin,
>
> I encountered with similar question with CP2K(7.0 development) and still 
> looking for help,
> which version you're using?
>
> Though I made a binary file could pass all test, I can't run it with conf 
> file,
>
> Besides, I have two gpus, when run without conf file, only one of two 
> doing job and I see in your conf file, no keywords specify how many gpus 
> should be used, which is another question confusing me.
>
> Good luck,
>
> J.T.Yang 
>
> On Thursday, December 5, 2019 at 2:49:41 PM UTC+8, vitesse lin wrote:
>>
>> Hi everyone,
>>
>> Can CP2K regression test run on multiple-GPU or what configuration I used 
>> is incorrect?
>>
>> "local_cuda.psmp.conf" is as follow.
>> export FORT_C_NAME=local_cuda
>> dir_base=$PWD
>> cp2k_version=psmp
>> dir_triplet=${FORT_C_NAME}
>> export ARCH=${dir_triplet}
>> cp2k_dir=../cp2k
>>
>> numprocs=4
>> numthreads=2
>> maxtasks=8
>>
>> leakcheck="NO"
>> default_err_tolerance="1.0E-14"
>> cp2k_run_prefix="mpirun -np ${numprocs}"
>> job_max_time="900"
>>
>> Failed tests are as follow.
>>
>>    1. libcp2k_unittest                
>>    2. ice.inp                         
>>    3. H2O-qmmm-none-force-mixing-1.inp
>>    4. RI_MP2_H2O_NONORTHO_MME.inp     
>>    5. farming-1.inp                   
>>    6. N3_dp_md.inp                    
>>    7. farming.inp                     
>>    8. dbcsr_types_01.inp              
>>    9. farming-1.inp                   
>>    10. farming-2.inp                   
>>    11. farming-3.inp                   
>>    12. farming-4.inp                   
>>    13. farming-5.inp                   
>>    14. farming-6.inp                   
>>    15. farming-7.inp                   
>>    16. farming-8.inp                   
>>    17. farming-9.inp
>>
>> The partial log of regression output.
>>
>> *************************** testing started ******************************
>>  started on  Thu Dec 5 00:28:56 CST 2019
>>  configuration: local_cuda-psmp
>>  regtesting location summary file: 
>> /home/result/regress-4-2-8/TEST-local_cuda-psmp-2019-12-05_00-28-56/summary.txt
>>  regtesting location error_summary file: 
>> /home/result/regress-4-2-8/TEST-local_cuda-psmp-2019-12-05_00-28-56/error_summary
>>  regtesting location memory_summary file: 
>> /home/result/regress-4-2-8/TEST-local_cuda-psmp-2019-12-05_00-28-56/memory_summary
>>  regtesting location output dir: 
>> /home/result/regress-4-2-8/TEST-local_cuda-psmp-2019-12-05_00-28-56
>>  regtesting location last dir: 
>> /home/result/regress-4-2-8/LAST-local_cuda-psmp
>> ---------------------------- Settings ------------------------------------
>> maxtasks         = 8
>> numprocs         = 4
>> OMP_NUM_THREADS  = 2
>> cp2k_run_prefix  = mpirun -np 4
>> cp2k_run_postfix =
>> cp2k_prefix      = mpirun -np 4 
>> /home/result/regress-4-2-8/../cp2k/exe/local_cuda/cp2k.psmp
>> cp2k_postfix     =
>> cp2k_version     = psmp
>> dir_triplet      = local_cuda
>> job_max_time     = 900
>> leakcheck        = NO
>> doretest         = no
>> nobuild          = nobuild
>> quick            = quick
>> noreset          = reset
>> shard            = 1 / 1
>> skiptest         = noskiptest
>> do_unit_test     = yes
>> farming          = no
>> maxbuildtasks    = 8
>>
>> Thanks,
>> Vitesse.
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20191210/b55e18ca/attachment.htm>