asynchronous logic

A data-driven circuit design technique where, instead of the components sharing a common clock and exchanging data on clock edges, data is passed on as soon as it is available. This removes the need to distribute a common clock signal throughout the circuit with acceptable clock skew. It also helps to reduce power dissipation in CMOS circuits because gates only switch when they are doing useful work rather than on every clock edge.

There are many kinds of asynchronous logic. Data signals may use either "dual rail encoding" or "data bundling". Each dual rail encoded Boolean is implemented as two wires. This allows the value and the timing information to be communicated for each data bit. Bundled data has one wire for each data bit and another for timing. Level sensitive circuits typically represent a logic one by a high voltage and a logic zero by a low voltage whereas transition signalling uses a change in the signal level to convey information. A speed independent design is tolerant to variations in gate speeds but not to propagation delays in wires; a delay insensitive circuit is tolerant to variations in wire delays as well.

The purest form of circuit is delay-insensitive and uses dual-rail encoding with transition signalling. A transition on one wire indicates the arrival of a zero, a transition on the other the arrival of a one. The levels on the wires are of no significance. Such an approach enables the design of fully delay-insensitive circuits and automatic layout as the delays introduced by the layout compiler can't affect the functionality (only the performance). Level sensitive designs can use simpler, stateless logic gates but require a "return to zero" phase in each transition.

http://cs.man.ac.uk/amulet/async/.