namd crashes after 300K steps

From: Peter Jones (pm-jones_at_bigpond.com)
Date: Thu Jul 26 2012 - 03:06:17 CDT

Hello All,

I'm running NAMD 2.9 on a machine with GPU acceleration, recently compiled by the machine's administrators. I find that for two simulations of different systems, (i.e. different protein in water box of significantly different sizes- 45K and 200K atoms) that the simulations each run for 325000 steps and then exit with errors to do with atoms moving too fast, and RATTLE constraints. However, if I stop the simulations before this point, i.e. run for 300000 steps, and then restart, without changing anything in the protocol, the runs go on indefinitely (100s of ns), without any errors and without the trajectories appearing odd in any way. I have no idea what could be causing this and any help would be greatly appreciated, thankyou,

Best Regards,
PJ

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:21:50 CST