Stats

  • Locked Locked
  • Replies 3
  • Subscribers 143
  • Views 1001
  • 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

run() & monteRun()

imagesensor123

 Hi all,

     for the same simulation, i have two process. the first, i set my model to the mc section, and i launched a simulation by clicking the "netlist and run" icon, we can get the result1(output curves). the second i also started a simulation from tools->monte Carlo, after configure the form, i triggered the "run" from the analog statistical analysis menus. we will get another result2. in special case, if i set the startIter "1" and numIter "1", can we espect that the result2 is the same with the result1? in another word, we choose "mc" section , not "tm" section in a classic simulation( from ADE window directly), is it a monte carlo analysis (just for one time)? one more question, could we change distribution method during monte carlo analysis? i didn't find any entrance in the analog statistical form.

 welcome to any discussion!

regards,

zfeng

  • Zfeng,

    You would not expect that a single point monte carlo would be the same as a non-monte carlo run. The non-monte carlo run will use the mean value of all parameters (those which have variations defined in a statistics block in the model files). The monte carlo run will be doing a randomization using the distributions defined in that statistics block, but only doing a single sample. The values won't be at their mean, but some random variation within the distribution.

    Assuming you're using IC5141 (you didn't say), you set up the distributed processing for ADE as a whole, not for just monte carlo. So you'd set it under Setup->Simulator/Directory/Host. If using IC61, you'd do this in ADE XL under Setup->Job Options.

    Andrew.

  •  Hi Andrew,

          yes, i agree with you. as we know, the montecarlo analysis use some random values controled by distribution method, and what i puzzled about is that if we can change only the model section to perform a monte-carlo analysis? that means i do not trigger the simulation from tools->montecarlo->simulatio->run, just from the ADE main window. if not, what's the difference between the normal simulation(section tc) and this one (section mc)? by the way, i do use IC5141. thanks a lot.

    regards,

    zhenfu

  • Without seeing your models, I can't really say what is different between the two sections. Typically though some design kits have the statistics block only enabled when you pick a section specifically for monte carlo, and also sometimes they then use subckt wrappers around the models which is needed to allow mismatch. In normal cases maybe they have a direct instantiation of the model, but for mismatch in monte carlo, you must have a parameterized model within a subckt. Doing that normally shouldn't cause a problem either, but it can cause a simulation penalty with other (non-Cadence) simulators, so sometimes foundries organise their models two ways to cope with that (even though it probably doesn't matter with spectre).

    Sometimes it's also because the typical, fast, slow models have hard values of all the model parameters, whereas the models used in the monte carlo sections are parameterized instead.

    All depends on how it's been organised by the model creators.

    Regards,

    Andrew.

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.