[gmx-users] Large output files and limited disk space. How do I handle them?

Justin A. Lemkul jalemkul at vt.edu
Mon Aug 9 05:21:59 CEST 2010

NG HUI WEN wrote:
> Dear gmxusers,
> I have a very basic question here which I hope someone could help me 
> with. I was running a couple of simulations over the weekend on a shared 
> cluster and both came to a stop for the same reasons:
> Program mdrun_mpi, VERSION 4.0.7
> Source code file: trnio.c, line: 252
> File input/output error:
> Cannot write trajectory frame; maybe you are out of quota?
> Indeed there were quite a number of large files in my user 
> directory (e.g. the .trr files and etc). I think the problems probably 
> arise from the fact that (i) I am storing my trajectory as full 
> precision .trr files and (ii) setting too small a value for nstxout and 
> nstvout in .mdp. 
> I have seen some tutorials that suggested using "trjconv" to derive the 
> reduced precision .xtc files from the .trr files and then discard the 
> latter. My question is, would this be wise as I am not sure whether I 
> would find myself needing these .trr files in the future.

Then consider your objectives.  Does your intended analysis require full 
precision?  Do you need the velocities?  How many data points will you really 
need?  For the majority of analyses that one might perform, .xtc files suffice, 
so you can skip the .trr file altogether and write an .xtc file using nstxtcout.

Also of interest:


> Here's a snippet of my .mdp file. Am I saving my coordinates and 
> velocities too frequently? If I were to increase this, are there any 
> compelling factors that I need to take into considerations?
> ; Run parameters
> integrator      = md            ; leap-frog integrator
> nsteps          = 500000        ; 2 * 500000 = 1000 ps (1 ns)
> dt              = 0.002         ; 2 fs
> ; Output control
> nstxout         = 100           ; save coordinates every 0.2 ps
> nstvout         = 100           ; save velocities every 0.2 ps
> nstenergy       = 100           ; save energies every 0.2 ps
> nstlog          = 100           ; update log file every 0.2 ps

I really see no need for such output every 0.2 ps.  I would consider 5000 data 
points per ns overkill for most applications.


> Many thanks for your help!
> <<
> Email has been scanned for viruses by UNMC email management service 
> <http://www.nottingham.edu.my>
>  >>


Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080


More information about the gromacs.org_gmx-users mailing list