IMPROVING THE PERFORMANCE OF PARALLEL RELAXATION-BASED CIRCUIT SIMULATORS

In this paper, we describe methods of increasing parallelism, thereby improving the performance, of waveform relaxation-based parallel circu it simulators. The key contribution is the use of parallel nonlinear r elaxation and parallel model evaluation to solve large subcircuits tha t may lead to load balancing problems. These large subcircuits are fur ther partitioned and solved on clusters of tightly-coupled multiproces sors. This paper describes a general hybrid/hierarchical approach for waveform relaxation, and then focuses on the implementation issues reg arding parallel nonlinear relaxation and parallel model evaluation. A number of benchmark circuits are used to demonstrate the improved perf ormance of this approach and its range of applicability in parallel ci rcuit simulation.