[gmx-users] Problems with g_energy

[Justin A. Lemkul]

Gavin Melaugh wrote:
> Hi justin
> 
> Basically the gmxdump produces the same as below as does gmxcheck.
> gmxcheck also produces the following:
> 
> Checking energy file ener.edr
> 
> Opened ener.edr as single precision energy file
> frame:      0 (index      0), t:      0.000        
> Reading energy frame   4000 time 40000.000          
> WARNING: there may be something wrong with energy file ener.edr
> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.             
> Trying to skip frame expect a crash though                    
> 
> Timesteps at t=43010 don't match (10, -43010)
> 
> WARNING: there may be something wrong with energy file ener.edr
> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.             
> Trying to skip frame expect a crash though                    
> 
> Timesteps at t=0 don't match (-43010, 0)
> 
> WARNING: there may be something wrong with energy file ener.edr
> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.
> Trying to skip frame expect a crash though
> 
> 
> It is at the point 43010 ps where the simulation had more than likely
> exceeeded the queue limit.
> Do you think it would be safer to use the -maxh option of mdrun ? as my
> jobs several days in duration.
> 

That's not a bad idea, if you know there's a limit.  It looks like the file 
somehow got corrupted.  You might be able to parse out meaningful data from 
frames before the corruption using eneconv for the .edr file and trjconv for 
your trajectories.  Then, if you have a .cpt file from an earlier time point, 
you can restart.  If you've been continually over-writing the .cpt files, 
though, you won't be able to get an exact restart.

-Justin

> Cheers
> 
> Gavin
> 
> 
> 
> Justin A. Lemkul wrote:
>>
>> Gavin Melaugh wrote:
>>> Hi folks
>>>
>>> I have recently encountered some problems with g_energy which I have not
>>> encountered before. I am running simulations for 32 molecular cages on
>>> 16 cpus in our local cluster. The maximum time for any one job is 24
>>> hrs. Therfore after 24 hrs the jobs are killed so I countinue the job
>>> using the following script:
>>>
>>> #!/bin/bash
>>> #PBS -N mdsubhexcge
>>> #PBS -q long_eth
>>> #PBS -l nodes=2:ppn=8
>>> #PBS -e test.e
>>> #PBS -o test.o
>>> #PBS -l walltime=23:59:59
>>>
>>> . /etc/profile.d/modules.sh
>>> . ~/.bashrc
>>>
>>> cd  /home/gmelaugh/phd/subhexyl/long600to200
>>> cat $PBS_NODEFILE > machinefile
>>>
>>> module load maui torque
>>>
>>> mpirun -machinefile machinefile -np 16
>>> /home/gmelaugh/gromacs-4.0.7-parallel/bin/mdrun -pd -s topol.tpr -cpi
>>> state.cpt -append
>>>
>>>
>>> The bottom command as I gather continues to write the relevant
>>> information to the relevant files starting from the last checkpoint.
>>> This seems to be the case as the new data is written to both the traj
>>> and log files. I am however encountering the following errors when
>>> trying to run g_energy:
>>>
>>> Reading energy frame   4000 time 40000.000              WARNING:
>>> there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.             Trying
>>> to skip frame expect a crash though                   
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.             Trying
>>> to skip frame expect a crash though                   
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.             Trying
>>> to skip frame expect a crash though                   
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.             Trying
>>> to skip frame expect a crash though                   
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.
>>> Trying to skip frame expect a crash though
>>>
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.
>>> Trying to skip frame expect a crash though
>>>
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.
>>> Trying to skip frame expect a crash though
>>>
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.
>>> Trying to skip frame expect a crash though
>>>
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.
>>> Trying to skip frame expect a crash though
>>>
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.
>>> Trying to skip frame expect a crash though
>>>
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.
>>> Trying to skip frame expect a crash though
>>>
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=0, time=0.
>>> Trying to skip frame expect a crash though
>>>
>>> WARNING: there may be something wrong with energy file ener.edr
>>> Found: step=0, nre=0, ndisre=0, nblock=-1125995535, time=0.
>>> Trying to skip frame expect a crash though
>>>
>>> -------------------------------------------------------
>>> Program g_energy, VERSION 4.0.7
>>> Source code file: smalloc.c, line: 179
>>>
>>> Fatal error:
>>> Not enough memory. Failed to realloc 846361600 bytes for fr->ener,
>>> fr->ener=0x7b0b40
>>> (called from file enxio.c, line 354)
>>> -------------------------------------------------------
>>>
>>> "I'd Like Monday Mornings Better If They Started Later" (Garfield)
>>> : Cannot allocate memory
>>>
>>> There is ample memory space on my machine so I can't see where the
>>> problem lies. Is there a way of viewing the energy file in nonbinary
>>> format? or has anybody ever come across this problem before?
>>>
>> You can run both gmxcheck and gmxdump on the .edr file.  What does
>> that show?
>>
>> -Justin
>>
>>> Cheers
>>>
>>> Gavin
> 
> 

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================