On Tue, 2014-04-08 at 14:34 +0900, KangMooseok wrote:
> Dear amber users. When I run pmemd.cuda, sometimes computer will
> be rebooted automatically, without error message. Normally the mdout
> file shows the gpu information. But in this case mdout files do not
> show the gpu information. And hang then after automatically rebooted.
> It happens randomly in each 105 nodes. What is the problem? Is it
> unknown bug,
It is _definitely_ not an Amber bug. Maybe a driver bug. Maybe a
kernel bug. Maybe buggy hardware. FWIW, I've been using Nvidia driver
334.21 for awhile on my machine with no problems.
Does it happen to _every_ node? If so, try downgrading the Nvidia
driver (maybe to one of the stable drivers in the 331 line). If it only
happens to a small subset of the nodes, look into the possibility of a
hardware problem.
HTH,
Jason
P.S. Your entire email came crammed on 2 lines (for me). That makes it
VERY hard to understand your mdout file since the formatting is
completely messed up. Please try to make sure your emails are formatted
correctly in the future so we can make more sense of whatever output you
paste there.
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Apr 08 2014 - 05:00:02 PDT
