Stats

  • Locked Locked
  • Replies 2
  • Subscribers 125
  • Views 14986
  • 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

[Spectre] Extracting model parameters for each Monte-Carlo iteration

gilsoriano

Dear wizards,

 

I need help to extract information about model parameters for each Monte-Carlo iteration in a Spectre simulation.

 The scenario is as follows:

  • Several Monte-Carlo iterations of a transient. 
  • The netlist contains bsim4-based models.
  • Mismatch parameters are extracted to a file via montecarlo parameters: savemismatchparams, mismatchscalarfile and mismatchparamfile.
  • I'm using Cadence-Spectre 5.1.4.

For now, I have circumvented the issue by adding a line just before the transient analysis:

   paramInfo info what=all where=file file="MC_MODEL_PARAM" save=all

it does what I want but it has two drawbacks:

  1. It slows down simulations A LOT.
  2. It appends all the model parameters one after the next one, for each MC iteration, in the same file.

Basically, I got the problem solved, but I would be interested in knowing whether it can be done in a similar way as mismatch parameters (which results in different files for each MC iteration)

 

 Thanks for your help!

 

Cheers,

Carlos

  • Carlos,

    If you use what=models  it will save less information. Secondly, by using where=rawfile, it would save into the psf directory - and that way the data is organized by iteration number. It's binary, but easily processed using OCEAN functions.

    Regards,

    Andrew.

  • Andrew,

     thanks a lot for your help. I have used what=models and what=inst to relax the amount of information to be saved, as well as where=rawfile. I tinkered around with data but I'm still experiencing problems.

    I would like to process the MC data into MATLAB and then stumbled upon Spectre/RF MATLAB library, which looks great. I'm able to run the cds_srr command, however I'm not getting the results expected.

    The picture of my simulation structure is as follows:

    .
    `-- monteCarlo
        |-- MC3_separate
        |   |-- MC3-001
        |   |-- MC3-002
        |   |-- MC3-003
        |   |-- MC3-004
        |   |-- MC3-005
        |   `-- MC3-006
        `-- params
     
    • In the first level (.) I have the netlist *.scs files of both the testbench and the top design.
    • ./montecarlo Contains the results of the simulation of the testbench. At this level I can find both latest transient analysis and parameters (inst, in this particular case): TRAN1.tran and paramInfo.info. Then, two more files appear: MC3_TRAN1.montecarlo and MC3_paramInfo.montecarlo. I guess they are the files that point to the rest of MC simulations which are in the MC3_separate subfolders. Finally, logFile (attached, as well) is the file which MATLAB reads to know which files to evaluate.
    • ./params I have all the MC parameters specified in testbench file (which is attached in this post). 
    • ./MC3_separate contains all the MC runs specified. Inside the subfolder I can find both transient and paramaters information for each run.

    The problem arises whenever I try to load my data in MATLAB. cds_srr(dirname) is unable to find the MC simulations, even though they appear in logFile (to whom folder I point to via dirname). I can only access to the latest transient and parameter info files: TRAN1.tran and paramInfo.info. I have noticed that changing from psfbin to sst2 exporting options makes montecarlo simulations appear when calling cds_srr. However I'm bogged down again when trying to access this simulation: it is empty.

    To summarize everything up, whenever I run cds_srr I'm not able to load MC simulations. 

     In case you need further information do not hesitate to ask.

     Thanks a lot for your help.

     Regards,

    Carlos 

    MCproblem.zip

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.