Home > Community > Forums > Custom IC SKILL > Memory issues with ocean script (Monte-Carlo+Parametric analysis)

Email

* Required Fields

Recipients email * (separate multiple addresses with commas)

Your name *

Your email *

Message *

Contact Us

* Required Fields
First Name *

Last Name *

Email *

Company / Institution *

Comments: *

 Memory issues with ocean script (Monte-Carlo+Parametric analysis) 

Last post Wed, Jul 24 2013 8:21 AM by Martin A. 1 replies.
Started by Martin A 23 Jul 2013 05:53 AM. Topic has 1 replies and 619 views
Page 1 of 1 (2 items)
Sort Posts:
  • Tue, Jul 23 2013 5:53 AM

    • Martin A
    • Not Ranked
    • Joined on Tue, Jul 23 2013
    • Posts 2
    • Points 10
    Memory issues with ocean script (Monte-Carlo+Parametric analysis) Reply

    Hi all, (sorry for this loooong first post :) )

    - I need to do a Monte-Carlo simulation (Virtuoso 6.1.5) for a RF circuit with N runs for different combinaisons of parameters (voltages V1 to V4)

    - For each of the 256 combinaisons I want to run a Monte-Carlo to evaluate 4 different parameters (like gain, linearity... Perf1 to Perf4) for each iteration ( N*256 runs totally)

    - For each iteration of this Monte-Carlo I want to store the data into a .csv file like this:

    V1,V2,V3,V4,Perf1,Perf2,Perf3,Perf4 

    - I did a script in OCEAN XL which is basically a loop in which I create a Monte-carlo script with the right parameters, run it and write the results into a .csv file

    for each combinaison(

                    write MC Run (with good parameters)

                    Perform MC Run

                    write Results into the csv file()

    )

    Everything works fine (the MC is done with good parameters, results are well-written)

    The problem is the following:

    After approx. 50 or 60 of the 256 combinaisons a memory warning appears saying than my RAM available is low. The memory continues to go down until the simulation crashes by lack of RAM memory. I run Cadence on a 24GB-RAM server (shared with other users) so I guess than the issue is in my script because I did not have any problems before...

    How can I be sure than after eah iteration of the loop (MC+writing results) my MC run is deleted from RAM memory?

    Leaving a file open for all the simulation process is it "RAM-consuming"?

    What can be the problem?

    If someone has an idea it would be very helpfull... 

     

    • Post Points: 5
  • Wed, Jul 24 2013 8:21 AM

    • Martin A
    • Not Ranked
    • Joined on Tue, Jul 23 2013
    • Posts 2
    • Points 10
    Re: Memory issues with ocean script (Monte-Carlo+Parametric analysis) Reply

     Investing it in more details, I think that for each iteration time it creates a new job. 

    When I lauch my script no jobs are running. If I look into the jobs ".cdslck" folder I see that the number of locked jobs is increasing xith the number of loop iterations. I guess that all those cdslck files are staying in the memory, filling it up until the memory is completely full...

    The problem is so (I guess) to kill the finnisehd job(s) after each iteration but as I am in oceanXL and I don't know how to do this...

    • Post Points: 5
Page 1 of 1 (2 items)
Sort Posts:
Started by Martin A at 23 Jul 2013 05:53 AM. Topic has 1 replies.