GPU Acceleration in VLSI Back-end Design: Overview and Case Studies

The semiconductor industry keeps seeking for reducing the design time and efforts in modern integrated circuit implementation, which often incorporates billions of transistors. Among the entire design flow, back-end design involving the physical implementation takes a significant portion of the design time. Recent advances in GPU acceleration bring new opportunities to speedup the design closure. In this tutorial, we review the efforts in literature and current status on accelerating the back-end design automation algorithms. We summarize the challenges in the key design stages such as placement, routing, and timing anaylsis, and provide several case studies on how to enable massive parallelism in practice.