Fault-tolerant Distributed Systems in Hardware

Very large-scale integrated (VLSI) hardware designs can be seen as distributed systems at several levels of abstraction: from the cores in a multicore architecture down to the Boolean gates in its circuit implementation, hardware designs com-prise of interacting computing nodes with non-negligible communication delays. The resulting similarities to classic large-scale distributed systems become even more accented in mission critical hardware designs that are required to operate correctly in the presence of component failures. We advocate to act on this observation and treat fault-tolerant hardware de-sign as the task of devising suitable distributed algorithms. By means of problems related to clock generation and distribution, we show that (i) design and analy-sis techniques from distributed computing can provide new and provably correct mission critical hardware solutions and (ii) studying such systems reveals many interesting and challenging open problems for distributed computing. 2