Stats

  • Locked Locked
  • Replies 4
  • Subscribers 125
  • Views 17716
  • Members are here 0
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

How do I load parametric analysis results in ADE L again to plot

Nader Fathy

Hello,

I am doing a simple parametric analysis from ADE L and sweeping on 4 values: L (20 points), Vd (10 points), Vb (10 points), and Vg (50 steps); this gives 100,000 simulations. I am interested in DC simulation only to gather information on (id, gm, gmb, vth, various junction capacitance, ... etc). I am expecting to get various plots of the mentioned parameters on the y-axis and Vg on the x-axis (since Vg is set as the last variable in parametric analysis). My settings are shown below:

My problem is: Cadence finished the 100k simulations in about a week for the schematics below and kept hanging on the last run for 3 days (I assume it was trying to finalize or save the results or something..), but then it crashed & I didn't get to see the results.

My questions:

1) Now I have 100k directories in the simulation directory with each run in a separate folder, but I don't know how to load them back in ADE and get the same plots if the simulation finished normally. I know how to load a single run in the results browser only by loading the psf directory, but I don't know how to do it in the parametric analysis multiple run. Is there an easy way to restore the results without running the simulations again? My final goal is to get plots similar to the image below (Not an exact result image but similar)

2) Normally after simulation finish, I plot the curves then select each bundle and export it to a .matlab file. Is there a way to do this automatically after simulation finishes? It takes a long time to group all curves as shown above and save it in a file separately; then repeat this for every device in my schematics. 

3) I noticed that parametric analysis is very fast at the beginning of simulations; each DC simulation takes less than a second. Then after a day or two in simulating, a single run takes 10 seconds, and towards the end it takes up to 20-30 seconds to do a single run. I tried APS (High performance simulation), but it is even slower with the first runs, so I expect it will be worse towards then end. Is there any other way to run this kind of simulation in parallel or more efficiently?

I am using the following Cadence versions:

MMSIM Version: 13.1.1.660.isr18

Virtuoso Version: IC6.1.8-64b.500.1

irun Version: 14.10-s039

Spectre Version: 18.1.0.421.isr9

Many thanks in advance!

Kindest Regards,

Nader Sh.

  • Dear Nader,

    Nader Fathy said:
    Now I have 100k directories in the simulation directory with each run in a separate folder, but I don't know how to load them back in ADE and get the same plots if the simulation finished normally. I know how to load a single run in the results browser only by loading the psf directory, but I don't know how to do it in the parametric analysis multiple run. Is there an easy way to restore the results without running the simulations again?

    Unless Andrew has a different experience or wants to correct me, I will provide my experience with the ADE-L Parametric Analysis tool. I must preface my remarks, however, to note that the use of this tool - and ADE-L - is no longer supported by Cadence and there are more efficient means of performing the analyses you describe which I will detail later. Three points to consider...

    1. If the Parametric analysis was the victim of a Virtuoso crash, I fear that the compiled psf directory that points to all the various simulations performed may either not be created or may be corrupted. This may explain why you can not load the full set of simulations in ViVA. However, many years ago when I actively used the Parametric Analysis tool, I filed several Cadence support tickets as occasionally ViVA would not recognize the compiled psf directory - even if it completed successfully.

    I described it in the post at:

    community.cadence.com/.../1335421

    I found a workaround to this issue was often to run a second Parametric Analysis (it need not be as long as the simulation set you submitted or even the same run). When it finished, for some reason, ViVA was able to recognize the compiled psf directory of a prior Parametric Analysis set.  You might try this to see if after you run a short second Parametric analysis ViVA recognizes the compiled psf directory for your initial run of 100K simulations. If not, I fear that the compiled directory is not readable.

    2. If the compiled psf directory is damaged, you might try to use an ocean script to retrieve the analysis results for those analyses that completed. There are a few examples within Cadence Support. One example is at URL:

    support.cadence.com/.../ArticleAttachmentPortal

    3. Finally, with respect to your comments regarding the increased time required for simulations as you approach the end of the set of Parametric Analyses, I would suggest you consider using ADE-Explorer or even ADE-XL to perform your sweeps. They are far more efficient in parsing jobs to a compute farm and options exist to allow simulations to complete even in Virtusuo crashes. There are also means to request either tool re-run unfinished or errored points of a simulation set. I am sure there will be a learning curve for you, but I think in the long run you might consider this.

    Shawn

  • in reply to ShawnLogan

    Hi Shawn,

    Thanks a lot for your feedback! Highly appreciated.

    Nader Sh.

  • in reply to Nader Fathy

    Hi Nader,

    Thank you for your nice comment - but I only hope you can retrieve most of your simulation data! I am sure you must be frustrated after a week of simulation work. It may be too basic for you, but I've attached an ocean script I've used to load the individual waveforms from a parametric analysis where two variables are swept. The script finds the lengths of the two sweeps and then iterates over the combination of the two to extract each waveform. This script happens to use an AC analysis and determines the real and imaginary parts of an impedance. Both are written to a comma-separated variable file.

    I only include it as a possible help in case you have not written an ocean script to extract data from a parametric analysis.

    Shawn

    z_param.zip

  • in reply to ShawnLogan

    Thank you so much Shawn! This is very useful Slight smile

Community Guidelines

The Cadence Design Communities support Cadence users and technologists interacting to exchange ideas, news, technical information, and best practices to solve problems and get the most from Cadence technology. The community is open to everyone, and to provide the most value, we require participants to follow our Community Guidelines that facilitate a quality exchange of ideas and information. By accessing, contributing, using or downloading any materials from the site, you agree to be bound by the full Community Guidelines.