[ad_1]
As computing devices become more powerful, the amount of power dissipated must remain at a certain level. The kT barrier, which represents an amount of energy, will be hit around 2015. To penetrate this barrier, reversible computers that do not generate entropy and therefore do not dissipate as much heat as conventional computers must be developed. Research into reversible computing is indispensable for the power and economics of our computers to continue to increase.
As the densities and switching speeds of our computing devices continue to increase exponentially, the amount of power dissipated by these devices must remain at a certain level or else economically impractical cooling apparatus is required. Conventional computers perform thermodynamically irreversible logic operations, i.e. it is not possible to extrapolate previous machine states based solely on information from future states. The information, in the form of bits, is erased. This bit erasure represents entropy, which is related to heat dissipation.
As we use more and more advanced techniques to design our integrated circuits, the energy dissipation per logic operation is continuously decreasing. But around 2015, development will hit a key barrier – the kT barrier – which represents an amount of energy calculated by multiplying the temperature of the computing environment (usually room temperature, or ~300 Kelvin) by Boltzmann’s constant. The only way to penetrate this barrier is to lower the temperature of our computers or to develop thermodynamically reversible computers that do not generate entropy and therefore do not dissipate as much heat as conventional, irreversible computers.
Building reversible computers is a significantly more attractive option than cooling because lowering the computing environment to the lowest achievable temperature (~0 Kelvin) only reduces energy dissipation per unit volume by two orders of magnitude, while building of reversible computers allows you to arbitrarily reduced.
By building computers that perform reversible logic operations, arbitrarily low levels of heat dissipation can be achieved. The downside is that reversible architectures can get quite complicated. As 2015 approaches and the computing industry begins to approach the kT barrier, it is likely that compilers will be designed to maximize the number of thermodynamically reversible operations within conventional computing architectures. When we begin to consider computers built with very small and fast logic gates, as in nanocomputers, reversibility becomes an essential feature to keep energy dissipation at tolerable levels.
Research into reversible computing today was initiated by MIT, whose Pendulum project was created specifically to devise a fully reversible computing architecture. Since the maximum achievable computing efficiencies necessarily consist of reversible architectures, this area of research is indispensable if the power and economics of our computers are to continue to increase.
[ad_2]