Solido
What does AFS control when running Monte-Carlo simulations ?
2024-06-23T23:38:34.000-0400
Solido Simulation Suite
Summary
This article describes AFS Monte-Carlo behavior in batch mode, Solido DE and other environments.
Details
Analog and mixed-signal designs are pervasive throughout the technology landscape today. The rapid growth in data expansion drives new designs and requirements across various technology sectors and pushes companies to smaller process geometries to address performance, size, and cost. This push introduces new challenges to analog verification. The main simulation challenge is to correlate to actual silicon measurements. Now, designers cannot tape out without performing statistical Monte Carlo and corner simulations for IP and cell-level characterization to account for global and local process variation with high sigma limits.It is essential to know what controls your Monte Carlo simulation parameters and how they are generated for each simulator and environment setup. Each simulator (i.e., AFS, Spectre, hspice,..etc.) or environment setup (i.e., ADE, Solido Design Environment,..etc.) will generate a different set of Monte Carlo parameters by their internal algorithms. Thus, the results won't match unless the same Monte Carlo parameters and netlist are used in both simulators. Below is a brief explanation of different use models you may choose to adopt:
Using AFS in Batch Mode/ADE
AFS generates its own Monte Carlo parameters and then runs the netlist.
The AFS-generated Monte Carlo parameters won't match those generated by other simulators in batch mode, ADE mode, or Solido DE because AFS uses a different Monte Carlo generation parameter algorithm than the other simulators/setups.
Hence, AFS batch mode simulation results will not match the Spectre batch mode or SDE results.
Using Solido DE
Solido DE generates the Monte Carlo parameters, creates the required netlists, and runs the different simulators with the same netlist.
In this scenario, the AFS and other simulators results should be similar
In conclusion, running your Monte Carlo simulations via the Solido Design Environment is the best and most efficient setup. This way, you can compare the different simulator results with each other since the Solido DE Monte Carlo-generated parameters and netlists are the same for each simulator.