[gmx-developers] XML files

David van der Spoel spoel at xray.bmc.uu.se
Thu Mar 21 15:13:12 CET 2002

On Thu, 21 Mar 2002, Anton Feenstra wrote:

>Sounds good. Are you also already implementing the forcefield
>independent topology, that was discussed before?
Not really, the forcefield is now part of the same XML file (as you can 
see in the DTD). However it could be conceived to have several XML files 
and have a tool that combines them, take the inputrec from one, the 
topology from another, the coordinates from a third.

>There is one issue, though, that I don't recall was mentioned
>before. If you have a large system (around 10000 atoms), reading
>in the .tpr file can be quite slow (even from the local disk on
>the r12k octane I'm at), in the order of 10-20 seconds. This will
>get worse since people are going towards even larger systems (isn't
>that right Erik?). Do you have any idea what would be the 
>performance impact of moving from .tpr to XML in that respect?
In a sense it will be better because the molecule information will not be 
replicated (only one water topology...) But of course everything is ASCII.
As it works now, one does a library call that reads in the whole file in a 
tree structure, which you then read out to a topology etc.

>Another sideways related thing now comes to mind also. After
>getting rid of the old non-portable binary topology .tpb, we still
>have the old non-portable trajectory format (.trj). I don't know
>of anyone still using it, but a problem is that you can not tell
>on which type of machine it was written. So I would suggest to
>either drop it, or add an identification of the machine type, so
>we can issue a warning.
We don't want to create more such files, so changing the format is not 
necessary. I suggest we allow reading as is, and whether or not you get 
something sensible is easy enough to determine.

>> Although conversion tools will be necessary of course.
>gmx2xml? ;-)
That should be a good idea, to have a central tool that reads all old 
stuff and spits out an XML file.

Groeten, David.
Dr. David van der Spoel, 	Biomedical center, Dept. of Biochemistry
Husargatan 3, Box 576,  	75123 Uppsala, Sweden
phone:	46 18 471 4205		fax: 46 18 511 755
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://zorn.bmc.uu.se/~spoel

More information about the gromacs.org_gmx-developers mailing list