Removing the Golden Handcuffs

This is the most exciting time since the days of Turing and Shannon to be working in the field of information. For the past five decades, computing has been driven by Moore’s Law and the exponential advances that were enabled by semiconductor device scaling. However, in order to turn those advances into products as quickly as possible, computing was trapped by the von Neumann architecture and all of the limitations that come with it. We find ourselves confronting the end of traditional transistor scaling, but we also understand from basic physics and the biological example of the brain that our present computing machines are still orders of magnitude away from any fundamental limits. There have been a large number of creative ideas that have not been fully explored because of resource limitations, and new proposals for re-inventing computing are appearing at a rapid pace. Necessity drives creativity, which means that new exponential advances in computing capability will likely arise from the combination of different computational paradigms, architectures, structures, materials and physics. I will describe some of the grass roots and governmental efforts to organize research programs to address these issues, including IEEE Rebooting Computing (http://rebootingcomputing.ieee.org/), the International Roadmap for Devices and Systems (http://irds.ieee.org/) and the Grand Challenge on Future Computing (https://www.nano.gov/node/1517). In addition, I will provide specific examples that implement new approaches, including memory-driven computing (https://www.labs.hpe.com/next-next/mdc), photonic fabrics for transporting data (http://genzconsortium.org/faq/gen-z-technology/) and biologically inspired accelerators (http://svivek.com/research/publications/isca16.pdf) to dramatically accelerate specific applications.