On Thu, Jun 4, 2015 at 9:59 PM, Chinh Su Tran To <chinh.sutranto.gmail.com>
wrote:
> Thanks Jason and Prof. Monard.
>
> One more question. Sometimes I got the results regarding "Infinity".
> For example, the results below I got from the .out file (running on GPU
> installed in a server, using ALPB Implicit, amber14, ff12SB), as below:
>
>
> --------------------------------------------------------------------------------
> 5. TIMINGS
>
> --------------------------------------------------------------------------------
>
> | NonSetup CPU Time in Major Routines:
> |
> | Routine Sec %
> | ------------------------------
> | Nonbond 223686.99 79.15
> | Bond 0.00 0.00
> | Angle 0.00 0.00
> | Dihedral 0.00 0.00
> | Shake 2049.98 0.73
> | RunMD 56857.13 20.12
> | Other 0.00 0.00
> | ------------------------------
> | Total 282594.10
>
> | Generalized Born CPU Time:
> |
> | Routine Sec %
> | ------------------------------------
> | Radii Calc 0.00 0.00
> | Diagonal Calc 0.00 0.00
> | Off Diagonal Calc 0.00 0.00
> | GBSA LCPO Calc 0.00 0.00
> | ---------------------------------
> | Total 0.00 0.00
>
> | Final Performance Info:
> | -----------------------------------------------------
> | Average timings for last 0 steps:
> | Elapsed(s) = 0.03 Per Step(ms) = +Infinity
> | ns/day = 0.00 seconds/ns = +Infinity
>
This is just performance (timing) information. Not related to results at
all. If you look at the mdinfo file during the course of the simulation,
it shows how fast the last chunk of simulation (between mdinfo prints) ran,
and the overall efficiency. Then below it shows the timings for every step.
It just so happens that the timing info was printed on the last step to the
mdinfo file, so the end of the mdout file had the timing for the last "0"
steps (as it says). Well, the time per step, when the total number of
steps is 0, is infinity (because you're dividing a number very close to
zero by zero). Same with the ns/day.
These numbers are meaningless, and not a cause for concern.
>
> |
>
> | Average timings for all steps:
>
> | Elapsed(s) = 283573.63 Per Step(ms) = 5.67
>
> | ns/day = 30.47 seconds/ns = 2835.74
>
> | -----------------------------------------------------
>
And here you see the average timing over the whole run. These are more
reliable performance measurements ;).
HTH,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jun 04 2015 - 21:00:02 PDT
