Stats

  • Locked Locked
  • Replies 8
  • Subscribers 143
  • Views 14446
  • 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

ocean run corner on the termination of an error is encountered!!!!

xianweng

hello,

Run with the ocean corner, write the results in txt, the error is encountered in some corner on the next stop, and how to display an error and then continue to run forever? Just use soon, thank you advice 

for example:

if I want to seek GB, GAIN
PM = phaseMargin (VF ("/ V"))
GB = gainProd (VF ("/ V"))
GAIN = value(dB20(VF("/V")) 1)
if(PM&&GB&&GAIN fprintf(p, "%f %f %f \n"  PM GB GAIN) fprintf(p, "%s\n" "error"))
It is an error "Gain dose not cross one" when i run ocean,then  the simulation stops, but when I comment GB GAIN ,It is still an error "Gain dose not cross one", but the simulation can continue

How to make the simulation continue with errors

  • Well your code has typing errors in it (spaces after function names before the open parenthesis) but hopefully you don't have that in reality.

    Is there an error message? Maybe setting _stacktrace=50 will give some clues as to the error?

    The message is probably coming from phaseMargin, but I'm presuming the error is actually coming from gainProd or value. Hard to tell without seeing the data.

    Andrew.

  • Dear Andrew,

    I have a similar problem. I am running a set of corner simulations and in some cases due to convergence problem. As soon as convergence problem raises, everything stops and the remaining corner simulations won't be started. May I ask if there is any way that even in case there is convergence issue the thread of corner simulations continue ? I appreciate your help.

     

    Armin

  • Since there are several ways that corners can be run, can you post your script?

    The answer will probably depend on how you're doing it.

    Andrew

  • Dear Andrew,

     

    To simplify, my script is something like this:

     

    foreach(myVDD '(1.0 1.1 1.2)

    foreach(myTEMO '(-40 37 90)

    foreach(myMOS '("tt" "ss" "ff")

    analysis('tran ?stop 10u)

    run()

    fprintf()

    )))

     

    It starts runing and just in one of the corners the simulator stops because of convergence problem. So, it doesn't continue the rest of corner simulations. In parallel I am trying to understand why it doesn't converge however still I would like to know how could I make the simulation could be continued for the rest of corners.

     

    Thanks,

     

    Armin

     

     

  • It's probably due to something failing in your fprintf after the run() command. Most likely, the results are not present, and so any results access you're doing is returning nil, resulting in an error in either a calculation or in the expected type in the arguments to printf.

    I normally recommend using constructs like:

    when(member('dcOp results())
    ; do calculations and fprintf
    )

    The first argument to member might be 'tran or whatever analysis you're trying to get the results from - see what they're called by calling results() in a successful run.

    Regards,

    Andrew

  • Thanks for your reply. From your explanation,  what I can understanding is that since some of the outputs of my measurements are "nil", so the loop stops, right? Otherwise, if I use the construction that you introduced, then the loop continues even if tran simulation stops due to convergence. I am correct? 

     

     

    Thanks again,

     

    Armin

  • Yes, that's right. Put another way, it's not the failure of the simulation that causes the SKILL code to stop, but a lack of error detection and an assumption that a valid result is available.

    When coding, it's always wise to put in sufficient error checking - and the when() is just a simple way of checking it.

    Strictly speaking, you might need more thorough error checking - for example, if you were using, say, cross() to find the time when a signal crossed a threshold, and in a particular corner it didn't, then cross would return nil. If the code that processes the result of the cross function doesn't have something to specifically handle this unexpected value, that could also cause it to fail.

    You could also be lazy and do:

    run()
    errset(
    {
    ; your calculations and fprintf statements
    }
    t
    ) ; errset

    The errset will trap any errors in the body - and stop it from halting - but it is not very elegant. The "t" as the second argument is there to ensure it prints out the error message.

    Regards,

    Andrew

  • Now everything is quite clear ! Thank you very much for your help.

     

     

    Armin

     

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.