Abstract
- Verification of custom circuit designs in sub-100nm technologies poses a specific challenge to the designer because simulation runtime can be long.
- Speeding up Monte Carlo simulation is an important challenge for characterizing circuit performance and for verifying circuits of various sizes, simulation runtimes, and robustness requirements.
- This paper shows various methods implemented in MunEDA WiCkeD™ to solve the problem efficiently and reliably.
1. Quasi Monte Carlo
A popular method to improve Monte Carlo sampling accu-racy are quasi-Monte Carlo methods, such as Latin Hyper-cube Sampling (LHS). Due to the random nature of sampling, some sampling points are very close to each other, which is inefficient. Secondly, the arithmetic mean value of standard Monte Carlo sampling points is not exactly 0, creating an error with estimating the spec mean value. Quasi-Monte Carlo methods distribute sample points more evenly, and put the mean value of data points closer to its expected value 0.
For more information on this topic, send us a request
To get the selected information or request for a solutions demo please enter your data with the following form.