SHF: Small: GOALI: Addressing the Challenges of Parameter Variation in the Design of Ultra-Low Power Chip Multiprocessors Using Near-Threshold Technology

点击翻译按钮获取中文摘要

The computing revolution of the last few decades has been driven in large part by a rapid growth in the performance of microprocessor chips. Unfortunately, this growth is now severely restricted by hard limits on power consumption. To ensure that computing technology will continue to evolve, solutions that dramatically increase the energy efficiency of computation must be created. This proposal focuses on developing techniques for building microprocessors that are an order of magnitude more efficient than the current state of the art. This will be achieved by designing and testing technologies for making chips more intelligent in the way they manage power. These systems will dynamically monitor their power consumption, allocate power intelligently to critical tasks and coordinate application execution with the power regulation mechanism, all with the goal of reducing energy waste. The proposed solutions span multiple technology layers, bringing together experts from different areas of chip design from both academia and industry.Beyond its technological and commercial potential, this research will further strengthen multidisciplinary teaching and research in energy efficient design at Ohio State University. The PIs will pilot a joint graduate-level course focusing on ultra-low power microprocessor design. This proposal will also help foster a long-term collaboration between Ohio State and Mentor Graphics that is committed to sending engineers and researchers to visit the Ohio State campus, advice graduate students involved in this project and be intellectually involved. This will encourage students to seek careers in computer aided VLSI design and microprocessor technology, thus helping fuel an industry that is vital to the growth of the US economy.