[gmx-users] NaN error using mdrun-gpu

Adam Jion adamjion at yahoo.com
Wed Feb 29 03:19:16 CET 2012


Hi!

I have a Palit 3GB GTX 580 GPU card to simulate about ~ 140 000 atoms.
However, after an mdrun-gpu simulation of 5000 timesteps (each time step = 0.002), the log file starts to show NaN in the energy values.

How do I fix this?

Regards,
Adam

ps. The same system runs well in a conventional CPU-based mdrun. So I don't think its due to the system blowing up.
pps. The GPU card reaches a temperature of ~ 88 degrees celsius after 5 minutes of simulation. Could this be the problem? However, I have another GPU card (MSI 1.5GB GTX 580) that works well with a smaller system (100 000 atoms) even at 85 degrees celsius.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20120228/3c8c714f/attachment.html>


More information about the gromacs.org_gmx-users mailing list