[gmx-users] Announcing: Gromacs 4.0, release candidate 1

Erik Lindahl lindahl at cbr.su.se
Mon Sep 22 16:43:00 CEST 2008


One extra comment: By default, checkpointing will result in separate  
output files to avoid filling up your disk,  or just in case something  
horrible happens just when we are about to resume the simulation.  
There are also some old strange versions of the GPFS file system where  
append operations are not allowed (no idea what happens if we were to  
try it anyway).

However...  it IS possible to have Gromacs 4 automatically checkpoint  
and just append to the previous output files if you (think you) know  
what you're doing. We haven't experienced any bugs with this yet, but  
this is one of the things we'd like you to test!

So, to enable all this stuff, simply execute mdrun e.g. as:

mdrun -v -deffnm test -cpi test -append

* The -cpi option will tell mdrun to look for the checkpoint file (the  
first time when we don't have it, you'll just get a normal run)
* The -append option will tell mdrun to append to the previous output  

Thus: the command line should preferrably be identical in all runs.  
This far we've only tested it when we're nice and don't move files,  
but feel free to stress-test it and report bugs to us. Don't complain  
if your trajectories are eaten after you've moved them in the middle  
of a simulation, though ;-)



On Sep 22, 2008, at 8:46 AM, Erik Lindahl wrote:

> Stockholm, September 22 2008
> In a bold move today, the Gromacs developers finally decided not to  
> wait for Duke Nukem Forever before releasing Gromacs 4.0, and just  
> put out release candidate 1 together with a new manual at
> ftp://ftp.gromacs.org:/pub/beta/
> "We realize it could be a big disappointment to many users not  
> having g_dukenukemforever built-in, but sometimes you have to make  
> hard decisions. Had we waited further, we might even have had to  
> release Gromacs 5 before Gromacs 4", an anonymous developer said.
> If you haven't already been playing around with CVS for the last  
> year we think you'll be pretty impressed with the scalability of the  
> new code, in addition to full checkpointing support, REMD, and  
> various optimizations.
> Having said that, since the code has changed a lot we decided to  
> have a fairly short cycle of "release candidates" before putting out  
> the real version. Please download the code above, try to compile and  
> run with it (see manual and html pages in the distro for  
> documentation of new stuff), and submit any bugs you find to  
> bugzilla ("If it's not in bugzilla, it's not a bug").
> Have fun!
> Erik, Berk, and David.
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before  
> posting!
> Please don't post (un)subscribe requests to the list. Use thewww  
> interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php

Erik Lindahl   <lindahl at cbr.su.se>  Backup: <erik.lindahl at gmail.com>
Associate Professor, Computational Structural Biology
Center for Biomembrane Research, Dept. Biochemistry & Biophysics
Stockholm University, SE-106 91 Stockholm, Sweden
Tel: +46(0)8164675  Mobile: +46(0)703844534  Fax: mail a PDF instead

More information about the gromacs.org_gmx-users mailing list