[gmx-users] simulation failure

[Erik Lindahl]

Hi,

On Jul 9, 2004, at 3:40 PM, herbst at fhi-berlin.mpg.de wrote:

>> The disk is not full, but both files are larger than 2GB. Is that the
> upper boundary for edr files? What shoudl I do now? Can I restart the
> simulation from the trajectory files?
>

The Gromacs code supports files larger than 2Gb, but it requires 
"LARGEFILE" support in the operating system when you compile it. Still, 
even if the operating system supports large files NFS version 2 doesn't 
(version 3 does), and some vendor versions of tar/gzip have problems 
too. For that reason we usually prefer to keep the files smaller than 
2Gb....

You don't need to write the energy frames very often, by the way. The 
average and drift values are calculated from partial sums by g_energy, 
and they are always based on the statistics from ALL frames, not only 
the actual values written to the energy file.

If your trajectory file is OK it should work fine to restart the 
simulation.

Cheers,

Erik