PowerRush: An Efficient Simulator for Static Power Grid Analysis

摘要: Efficient power grid analysis is critical for modern very large scale integration design but computationally challenging in runtime and memory consumption because of the increasing size grids. PowerRush proposed as an efficient IR-drop simulator, which includes SPICE parser, a robust circuit builder, linear solver Algebraic MultiGrid Preconditioned Conjugate Gradient. The AMG-PCG pure algebraic method, can provide stable convergence without geometric information. Aggregation-based AMG with K-cycle acceleration adopted preconditioner to improve scalability iterative method. In multigrid scheme, double pairwise aggregation technique applied matrix graph coarsening ensure low setup cost requirement. Furthermore, scheme Krylov subspace at each level guarantee enhanced robustness scalability. experimental results large-scale grids have shown that has remarkable both consumption. DC 60-million nodes be solved by 0.01 $mV$ accuracy within 150 s 21.99 GB total used. Moreover, perform much better than widely used direct Cholmod well-developed Hybrid on