SRAM-PG: Power Delivery Network Benchmarks from SRAM Circuits
Shan Shen, Zhiqiang Liu, Wenjian Yu
TL;DR
PDN analysis in modern ICs requires realistic, diverse benchmarks with full RC parasitics. The authors present SRAM-PG, a public benchmark suite derived from four state-of-the-art SRAM designs at 28 nm, constructed from post-layout netlists and load-current measurements to support both DC and transient PDN analyses. The benchmarks feature complete RC networks and accurate load profiles, enabling realistic IR drop and EM assessments and offering templates for DC/transient SPICE simulations. This resource improves comparability and scalability for PDN solvers and aids in IR drop mitigation and RC-reduction research across diverse memory-centric designs.
Abstract
Designing the power delivery network (PDN) in very large-scale integrated (VLSI) circuits is increasingly important, especially for nowadays low-power integrated circuit (IC) design. In order to ensure that the designed PDN enables a low level of voltage drop and noise which is required for the success of IC design, accurate analysis of PDN is largely demanded and brings a challenge of computation during the whole process of IC design. This promotes the research of efficient and scalable simulation methods for PDN. However, the lack of sufficient public PDN benchmarks hinders the relevant research. % on this aspect since it is hard to conduct a rapid and clear comparison between different approaches to solving this problem. To this end, we construct and release a set of PDN benchmarks (named \emph{SRAM-PG}) from SRAM circuit design in this work. The benchmarks are obtained from realistic and state-of-the-art SRAM designs, following a workflow for generating the post-layout PDN netlists with full RC parasitics. With careful modeling of load currents, the benchmarks reflect the dynamic work mode of the IC and can be used for both transient and DC analysis. The benchmarks are derived from the designs for diverse applications. And, sharing them in the public domain with detailed descriptions would largely benefit the relevant research. The whole set of benchmarks is available at \href{github}{https://github.com/ShenShan123/SRAM-PG}.
