Re: [AMBER] Regarding low %CPU efficiency while running production MD

From: case <>
Date: Sat, 17 Dec 2011 10:15:18 -0500

On Sat, Dec 17, 2011, wrote:

> I tried using pmemd.MPI instead of sander.MPI but it does
> not seem to have any effect on the time taken for one run to
> finish.
> My input file looks like this:
> &cntrl
> imin=0, irest=1, ntx=5,
> nstlim=2500000, dt=0.002,
> ntc=2, ntf=2,
> ntt=1, tautp=0.5,
> tempi=325.0, temp0=325.0,
> ntpr=500, ntwx=500,
> ntb=0, igb=1,
> cut=999.,rgbmax=999.
> /

pmemd is much faster than sander for solvated simulations, but the difference
in peformance (and in parallel scaling) is much smaller for generalized Born

Are you sure you need to have a cutoff of 999? Also, I would strongly
recommed using ntt=3 (not ntt=1) for these simulations (see discussion in the
Users' Manual under the ntt variable: "ntt=1 is especially dangerous for GB


AMBER mailing list
Received on Sat Dec 17 2011 - 07:30:04 PST
Custom Search