Hi,
I've noticed some strange behavior while running Langevin dynamics
simulations with GB on a quad-core 2.66 GHz, 64-bit linux workstation with 2
GB memory. I have no problems running two independent simulations at the
same time using sander classic, but when I run four simulations, the CPU
usage for two of the simulations is above 90% while the CPU usage for the
other two are below 50%. While running the four simulations at the same
time, each simulation using only 0.2% memory, other tasks on the computer
become unusually slow e.g. it takes ~30 seconds to bring up a terminal
window.
Does anybody know why I'm getting this poor performance? I have run four
independent GROMACS simulations before on this workstation and had no
problems.
The size of the system I am simulating is 2948 atoms. I am using AMBER 9.0,
which was compiled using an intel compiler and Red Hat Enterprise Linux
version 4.0.
With thanks,
Lillian
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Mar 14 2007 - 06:07:29 PDT
