[CP2K:6407] Re: CP2K Hangs

Manav Kumar mk3... at nyu.edu
Sat Apr 25 11:26:16 CEST 2015


Hi James,

No, I haven't tried that yet. I will pass on the information to the system 
administrators at our HPC and see what they say. Did that fix the issue 
with you?

Best,
Manav Kumar


On Saturday, April 25, 2015 at 5:05:16 AM UTC-4, JamesChen wrote:
>
> Hi Manav,
> I have a similar issue with mvapich2 and some buggy version of Intel 
> compilers. It seems that you are using rather standard software stack. Have 
> you ever downgraded your openmpi to 1.6.5 (prior stable version)? 
> James
>
> On Sat, Apr 25, 2015 at 6:08 AM, Manav Kumar <mk3... at nyu.edu <javascript:>
> > wrote:
>
>> Hi James,
>>
>> I'm assuming when you say the software environment, you mean the 
>> dependencies for CP2K. Here are the dependencies that are added to my 
>> environment when I use the CP2K module:
>>
>> 'gcc/4.9.1'
>>
>> 'openmpi/gcc_4.9.1/1.8.3'
>>
>> 'mpiblacs/gcc_4.9.1/openmpi_1.8.3/2.0.2'
>>
>> 'lapack/gcc_4.9.1/3.5.0'
>>
>> 'scalapack/gcc_4.9.1/openmpi_1.8.3/2.0.2'
>>
>> 'fftw3/gcc_4.9.1/3.3.4'
>>
>>
>> Let me know if there is any other information that might help!
>>
>>
>> Best,
>>
>> MK
>>
>> On Friday, April 24, 2015 at 5:56:32 PM UTC-4, JamesChen wrote:
>>>
>>> Hi,
>>> Could you provide us your software environment? Are you using mvapich2 
>>> as MPI? I will suggest you follow the CP2K manual and choose Gfortran/Gnu 
>>> compiler, Netlib Math library and Openmpi as the reference software stack.
>>> http://www.cp2k.org/howto:compile 
>>>
>>> Cheers,
>>> James
>>>
>>> On Sat, Apr 25, 2015 at 5:08 AM, Manav Kumar <mk3... at nyu.edu> wrote:
>>>
>>>> Hi Matt,
>>>>
>>>> Thank you for your response!
>>>>
>>>> I'm sorry I mis-spoke earlier, I meant to say that NVT not NVE that's 
>>>> my mistake. When I ran NVT with CP2K module it worked fine.
>>>>
>>>> I ran a cell optimization using the CP2K module, and it went through 
>>>> fine.
>>>>
>>>> I have talked to some of the other members in my group, and they 
>>>> mentioned that it is possible that when I am running the simulation on 
>>>> multiple nodes that one of them is faulty and therefore causing the 
>>>> calculations to hang. However, it seems like the probability of me hitting 
>>>> the same faulty node every time I submit a job seems highly unlikely. Also, 
>>>> if this was the case then wouldn't the jobs that I submit using my own 
>>>> compiled version of CP2K also hang?
>>>>
>>>> When you said that the it was a problem with the cell size, what did 
>>>> you mean exactly?
>>>>
>>>> Best,
>>>> Manav Kumar
>>>>
>>>>
>>>> On Friday, April 24, 2015 at 11:51:29 AM UTC-4, Matt W wrote:
>>>>>
>>>>> Hi,
>>>>>
>>>>> nothing is obviously wrong with your input that I can see. Maybe (i) 
>>>>> problem with the executable (ii) problem with cell size.
>>>>>
>>>>>
>>>>> To start debugging -
>>>>> You say you can run NVE - can you run NVT as you've got it set up?
>>>>> Can you run a cell optimization, maybe just a few steps, to check that 
>>>>> the stress tensor/cell volume change is working OK.
>>>>>
>>>>> Matt
>>>>>
>>>>> On Friday, April 24, 2015 at 12:52:48 PM UTC+1, Manav Kumar wrote:
>>>>>>
>>>>>> Hi users and developers,
>>>>>>
>>>>>> This is my first project using CP2K, so I'm sorry in advance if some 
>>>>>> of the things I say might be wrong.
>>>>>>
>>>>>> I am currently trying to generate a trajectory for a simulation of 
>>>>>> diacetone diperoxide (DADP) using a NPT ensemble. I am attaching the input 
>>>>>> file, and the submit script below.
>>>>>>
>>>>>> The HPC that I am using has a CP2K module. I performed a 
>>>>>> wave-function optimization, geometry optimization, cell optimization, and a 
>>>>>> NVE ensemble calculation all using the module. However, now that I am 
>>>>>> running the NPT calculation using the module it seems to hang. It reads in 
>>>>>> all the parameters from a previous restart file, it starts to initiate the 
>>>>>> process but gets stuck in the SCF optimization routine. It doesn't print an 
>>>>>> error message, it just stays at that position until the wall time comes 
>>>>>> along and kills the process. Sometimes it completes a lot of SCF 
>>>>>> optimization steps, sometimes it gets through none.
>>>>>>
>>>>>> What I've done so far to try to remedy the problem is that I've tried 
>>>>>> compiling CP2K myself and running the calculation using that executable 
>>>>>> instead of the module. I've had a bit more success with my own compiled 
>>>>>> executable, but even with that it sometimes hangs. The system 
>>>>>> administrators responsible for the HPC checked the binaries and said that 
>>>>>> everything was fine.
>>>>>>
>>>>>> Has anyone else had an issue similar to this?
>>>>>>
>>>>>> Best,
>>>>>> Manav Kumar
>>>>>>
>>>>>> I will attach my input file, output file, and submit script below so 
>>>>>> that you can see what I have been doing.
>>>>>>
>>>>>  -- 
>>>> You received this message because you are subscribed to the Google 
>>>> Groups "cp2k" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send 
>>>> an email to cp2k+... at googlegroups.com.
>>>> To post to this group, send email to cp... at googlegroups.com.
>>>> Visit this group at http://groups.google.com/group/cp2k.
>>>> For more options, visit https://groups.google.com/d/optout.
>>>>
>>>
>>>  -- 
>> You received this message because you are subscribed to the Google Groups 
>> "cp2k" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to cp2k+... at googlegroups.com <javascript:>.
>> To post to this group, send email to cp... at googlegroups.com <javascript:>
>> .
>> Visit this group at http://groups.google.com/group/cp2k.
>> For more options, visit https://groups.google.com/d/optout.
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150425/268cddbe/attachment.html>


More information about the CP2K-user mailing list