Stats

  • Locked Locked
  • Replies 3
  • Subscribers 125
  • Views 14814
  • 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

multiple instances for monte carlo

yizh

Hi,

I design a small block with a few transistors.

I want to see and measure, in my simulations, 1000 instances of this block, as they will be fabricated in Si (i.e. no intentional extreme cases, but true distribution).

For various reasons, I prefer not to run 1000 different splits, but instead to instantiate the small block 10 times in my testbench and run only 100 splits.

This way, instead of one measurement, I have 10 measurements in my Maestro, and I extract the required statistics manually based on the 100X10=1000 measurement results I have.

The question is, is my thinking correct. 

I'm afraid that for some reason, due to the way the Monte Carlo in implemented, I don't really have results of 1000 blocks, as the 10 different instances may be correlated somehow. 

If it matters, the sampling method I'm using is "random", because I want to see a true distribution of 1000 blocks and not to see extreme end cases.

Thanks!

Parents
  • Andrew Beckett

    From what you've described, that should be OK. You can also use one of the newer (more efficient) sampling methods such as "Low Discrepancy Sequence" which tends to avoid wasted samples which just repeat the previous experiments. 

    In essence this is a common test for repeated blocks such as memories. You simulate a single SRAM bit cell and then use this to determine the overall failure rate. Often in such case you might use some of the Virtuoso Variation Option methods such as "Scaled Sigma Sampling" to do things like estimate the yield at 6 sigma, but do so by only running a relatively small (typically 7000) simulations.

    Regards,

    Andrew.

  • yizh
    in reply to Andrew Beckett

    Thanks Andrew.

    This is very interesting!

    Is there any manual in which I can find more data on Low Discrepancy Sequence and Scaled Sigma Sampling?

  • Andrew Beckett
    in reply to yizh

    Yes, In the CIW (or any Virtuoso window), bring up the Help menu and at the top there's a type-in field. Type "Low-Discrepancy Sequence" (with the quotes) or "Scaled-Sigma Sampling" (also with the quotes). I'm giving you the search string that best finds the relevant manuals.

    In essence it's covered in the Virtuoso ADE Explorer User Guide (for Low-Discrepancy Sequence) or the Virtuoso Variation Option User Guide (for Scaled-Sigma Sampling).

    Regards,

    Andrew.

Reply
  • Andrew Beckett
    in reply to yizh

    Yes, In the CIW (or any Virtuoso window), bring up the Help menu and at the top there's a type-in field. Type "Low-Discrepancy Sequence" (with the quotes) or "Scaled-Sigma Sampling" (also with the quotes). I'm giving you the search string that best finds the relevant manuals.

    In essence it's covered in the Virtuoso ADE Explorer User Guide (for Low-Discrepancy Sequence) or the Virtuoso Variation Option User Guide (for Scaled-Sigma Sampling).

    Regards,

    Andrew.

Children
No Data

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.