Fluidic Cooling and Gate Size Co-optimization in 3D-ICs: Pushing the Power-Performance Limits

Abstract

The performance improvement of modern computer systems is usually accompanied by increased computational power and thermal hotspots, which in turn limit the further improvement of system performance. In 3D-ICs, this thermal problem is significantly exacerbated, necessitating the need for active cooling approaches such as micro-fluidic cooling. This paper investigates a co-optimization approach for 3D-IC electric (gate sizing) and cooling design that fully exploits the interdependency between power, temperature and circuit delay to push the powerperformance tradeoff beyond conventional limits. We propose a unified formulation to model this co-optimization problem and use an iterative optimization approach to solve the problem. The experimental results show a fundamental power-performance improvement, with 12% power saving and 16% circuit speedup.