Introduction to reversible computing: motivation, progress, and challenges

• Authors: Michael P. Frank
• Venue: Proceedings of the 2nd conference on Computing frontiers
• Year: 2005

Reversible computing is motivated by the von Neumann-Landauer (VNL) principle, a theorem of modern physics telling us that ordinary irreversible logic operations (which destructively overwrite previous outputs) incur a fundamental minimum energy cost. Such operations typically dissipate roughly the logic signal energy, itself irreducible due to thermal noise. This fact threatens to end improvements in practical computer performance within the next few decades. However, computers based mainly on reversible logic operations can reuse a fraction of the signal energy that theoretically can approach arbitrarily near to 100% as the quality of the hardware is improved, reopening the door to arbitrarily high computer performance at a given level of power dissipation. In the 32 years since the theoretical possibility of this approach was first shown by Bennett, our understanding of how to design and engineer practical machines based on reversible logic has improved dramatically, but a number of significant research challenges remain, e.g., (1) the development of fast and cheap switching devices with adiabatic energy coefficients well below those of transistors, (2) and of clocking systems that are themselves of very high reversible quality; and (3) the design of highly-optimized reversible logic circuits and algorithms. Finally, the field faces an uphill social battle in overcoming the enormous inertia of the established semiconductor industry, with its extreme resistance to revolutionary change. A more evolutionary strategy that aims to introduce reversible computing concepts only very gradually might well turn out to be more successful. This talk explains these basic issues, to set the stage for the rest of the workshop, which aims to address them in more detail