Re: [AMBER] ptraj.MPI in Ambertools1.4 : Compression is not currently supported for parallel ptraj

From: Daniel Roe <>
Date: Wed, 30 Jun 2010 10:16:14 -0400


On Sun, Jun 27, 2010 at 5:10 PM, M. Reza Ganjalikhany
<> wrote:
> Thank you very much for your help, but could you tell me about the Bugfix.4
> in Ambertools1.4 which is mentioned:
> bugfix.4: Fix to parallel *ptraj* when *compressed trajectories* are input.
> Is this compressed trajectory corresponds to that netCDF format you have
> mentioned?

The brief description is actually a little unclear. Bugfix 4 gives
ptraj the ability to determine the number of frames in a compressed
file, which is necessary for memory allocation. The information at the
head of the bugfix file is more accurate:

********> bugfix 4
Fixes bug where ptraj actions would attempt to allocate a negative number of
bytes of memory when the number of frames to be read is unknown, i.e. file is
compressed or corrupted. The number of frames for compressed files can now be
calculated ahead of time. For gzip files this is simple, but for bzip2 and
zip files this adds to the processing time since uncompressed file size cannot
be determined for these file types without external programs. For corrupted
files the trajectory filesize is used to predict the number of frames.
After applying the patch ptraj should be completely recompiled (i.e. perform
a make clean in the ptraj directory).

Since ptraj doesn't natively support compression but uses external
programs, parallelizing compressed file access would be a real pain.
NetCDF files are not compressed, but they are unformatted and so use
up less space than ASCII formatted files (like the native Amber
Trajectory format). If you want to use ptraj.MPI I would second
Jason's recommendation and use NetCDF files (via ioutfm=1 in sander
and pmemd). You can use ptraj to convert formatted amber trajectories
to NetCDF files by adding the 'netcdf' keyword to the 'trajout'


Daniel R. Roe
Postdoctoral Associate
SAS - Chemistry & Chemical Biology
610 Taylor Road
Piscataway, NJ   08854
AMBER mailing list
Received on Wed Jun 30 2010 - 07:30:13 PDT
Custom Search