A hot topic in the news recently revolves around the problems with heat dissipation for all new designs. Dr. Michael Frank takes a look at some of them and tells you what he thinks. Interesting reading.
A lot the technology news this week (for example, this EETimes article or this CNET article) seems to be centered on the International Solid-State Circuits (ISSC) conference going on this week in San Francisco, and about how one chipmaking giant or computer manufacturer after another is reporting on their increasingly desperate-seeming attempts to control processor power dissipation, and/or aggressively cool down the systems built out of them. If, last year, the hand was already writing on the wall for traditional power-performance scaling of irreversible FET-based logic, this year, that hand seems to be madly plastering down layer after layer of thick, chaotic graffiti. Watch out!
As for me, I find myself shaking my head in amazement, wondering exactly how long industry will continue to bury its collective head in the sand regarding the fundamental underlying cause of the problem, namely the inherent thermodynamic wastefulness of the traditional digital logic paradigm, while it continues to blithely ignore the enormous and well-demonstrated promise and potential that reversible computing offers, for blowing away the severe energy-efficiency barriers that are inherent not only to standard CMOS logic, but to all possible irreversible logic schemes. For example, Hewlett-Packard can play around with its cute little nanowire latch lattices all it wants, but, if it doesn't also buy into thoroughly logically and physically reversible switching principles, this and all similar efforts are inevitably doomed to meet embarrassing failure to get a very long way beyond the power-performance characteristics of ordinary FET-based logic.
Fortunately, a precious few bright folks in industry are finally beginning to sit up and take notice of reversible computing. For example, next Monday at MIT, the Microelectronics Advanced Research Corporation's Focus Center Research Program (MARCO/FCRP), a spinoff of Semiconductor Research Corporation (SRC), is hosting a small private workshop on reversible computing at their Center for Materials, Structures, and Devices at MIT, organized by Vwani Roychowdhury of UCLA and Rick Kiehl of U. Minnesota, and including such respected physicists as Seth Lloyd of MIT, Konstantin Likharev of SUNY Stonybrook, and Craig Lent of Notre Dame, together with such experienced industry representatives as David Frank of IBM. I have the good fortune to be offered the opportunity to attend, observe, and join this discussion, and I have high hopes that this well-qualified group will help to set the record straight, for the benefit of the brain trust at SRC and the architects of the ITRS, regarding both the absolute necessity for (and the physical non-impossibility of) reversible computing, as well as the major, serious research investments that will be required in order to make it a reality. Next week, I will report on the outcome of the workshop.
Regardless of the result, the interested public is invited to learn more by attending the 1st International Workshop on Reversible Computing in Ischia, Italy next May. We have (finally) selected our papers, and published the preliminary workshop program on our website. Our speakers will be addressing reversible computing from all angles: Basic theoretical principles, high-performance applications, new device technologies, and approximate CMOS-based realizations. Skeptics have been invited as well, to prod our community with constructive criticism. Working together, I am optimistic that we can overcome the doubts and skepticism, and show the world that reversible computing is not only a possible, but the only possible way out of the power-dissipation crisis which is already upon us. By which I mean, it is truly the only solution that is more than just a temporary, short-sighted, stop-gap measure, but that, instead, offers the potential for indefinite continued improvement.
Dr. Michael P. Frank, Ph.D., Assistant Professor
Florida A&M University and Florida State University
FAMU-FSU College of Engineering,
Department of Electrical & Computer Engineering,
2525 Pottsdamer St. Room 341, Tallahassee, FL 32310
email@example.com, Tel:(850) 410-6463, cell (850) 597-2046