ABSTRACT
This chapter discusses when optimization is beneficial, and how the most important optimizers work. It shows the best practices for setting up an optimization and how to respond in case of problems. Sometimes designers claim that optimization results are not well suited for layout reasons. One problem is often that some goals are simply not set up, such as the block area; another reason could be the style of parameterization. Simulated annealing works fine on most combinatorial optimization problems like the traveling salesman problem. An optimization usually requires much more simulation effort than a pure verification, because many different design parameterizations must be simulated till the optimizer reaches really an optimized version. In a transistor-level testbench and running simulation, we have many options to decide how to check stability, like via transient analysis or with an AC loop gain analysis.
