Energy efficiency and heat generation have become primary concerns in the design of computing systems. At their root are the energy inefficiencies imposed by the second law of thermodynamics. Yet these inefficiencies may not be as fundamental as previously thought. Over the past 15 years, more than two dozen distinct challenges to the second law have been published in the mainstream scientific literature – over 60 papers, and multiple books and conference proceedings – suggesting that the second law might fail under certain circumstances. If this law can be circumvented, in principle, computing systems could be designed that run on the energy from ambient heat, and remain cool in the process.
This talk gives a brief review of the second law, its relation to the heat generation problem, and an overview of recent second law challenges. One promising silicon-based technology will be considered in detail, one that has recently received strong experimental corroboration in the laboratory. Prospects for its implementation toward perpetual computing with perfect cooling will be discussed.