[AMBER] annoying elapsed time in mdout of amber18

From: Song-Ho Chong <songho.chong.gmail.com>
Date: Thu, 27 Sep 2018 21:57:00 +0900

Dear Amber developers,

I recently started to use pmemd.cuda of amber18, and found
that the elapsed time in mdout file is problematic.

I usually do short equilibration (20 ps NVT + 200 NPT) before
production runs, and save output files every 1 ns.
With a 2 fs time interval, the last part of mdout file, say, after
335 ns production run with NVT should be like

 NSTEP = 500000 TIME(PS) = 335220.000 TEMP(K) = 309.42 PRESS =

and this output line is exactly as expected.

But, for some reason, right after 336 ns production run, I see

 NSTEP = 500000 TIME(PS) = 336219.999 TEMP(K) = 309.97 PRESS =

instead of TIME(PS) = 336220.000 that I expected.

I'm running more than 20 simulations (for various mutants), and
this always occur between 335 and 336 ns in all of them.

After that, something similar happens between 429 and 430 ns,
that is

  TIME(PS) = 429219.999 at the end of 429 ns production run
  TIME(PS) = 430219.998 at the end of 430 ns production run

and this is continuing.

This is very annoying since I'm checking this number to see
something wrong has happened or not, and I have never
experienced this with the previous versions (Amber12-16).

What could be the reason for this?

Best wishes,

Song-Ho Chong
Department of Chemistry, Sookmyung Women's University
AMBER mailing list
Received on Thu Sep 27 2018 - 06:00:04 PDT
Custom Search