The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> Stsrt individual Monte-Carlo runs efficiently
https://designers-guide.org/forum/YaBB.pl?num=1140090246

Message started by achim.graupner on Feb 16th, 2006, 3:44am

Title: Stsrt individual Monte-Carlo runs efficiently
Post by achim.graupner on Feb 16th, 2006, 3:44am

Hello all,

I am still a little busy with a trimming simulation. I do the following (also see my previous post):
1. run a single MC-simulation (1 run only)
2. compute trim value
3. repeat the abouve MC simulation with the correct trim value at various operating points

The trouble is the following: I do single monte carlo runs with different start values:
mc1 montecarlo \
 numruns=1 \
 firstrun=500 \
 ...


I have observed that with increasing number of "firstrun" the MC-simulation becomes more and more time consuming. It takes considerable time to skip MC runs, see output:

************************************************
Monte Carlo Analysis `mc1': iteration = (1 -> 1)
************************************************
awaiting artil process initialization ....
**** Performing monte carlo loop for `mc1'
...
Total time required for dc analysis `mc1-001_dc' was 420 ms.
Total time required for montecarlo analysis `mc1' was 650 ms.

...

****************************************************
Monte Carlo Analysis `mc1': iteration = (500 -> 500)
****************************************************
awaiting artil process initialization ....
**** Skipping runs 1 to 499 for `mc1'
...
Total time required for dc analysis `mc1-500_dc' was 460 ms.
Total time required for montecarlo analysis `mc1' was 13.89 s.


I suppose that  the computation of the random values take significant time.
Andrew, do you, or does someone else has an idea how to speed up this process? Is it somehow  possible to store a monte-carlo state which can be reloaded later?

Thanks, regards,
achim

Title: Re: Stsrt individual Monte-Carlo runs efficiently
Post by Andrew Beckett on Feb 18th, 2006, 6:27am

Hi Achim,

I filed an enhancement request precisely for this a while ago. The trouble is at the moment, it has to sequence through the whole random number series in order to get to the final setting of that Monte Carlo run.

I'll create a service request for you, and associate it with the PCR.

Regards,

Andrew.

The Designer's Guide Community Forum » Powered by YaBB 2.2.2!
YaBB © 2000-2008. All Rights Reserved.