Advanced architectures for astrophysical supercomputing

This thesis explores the substantial benefits offered to astronomy research by advanced 'many-core' computing architectures, which can provide up to ten times more computing power than traditional processors. It begins by analysing the computations that are best suited to massively parallel computing and advocates a powerful, general approach to the use of many-core devices. These concepts are then put into practice to develop a fast data processing pipeline, with which new science outcomes are achieved in the field of pulsar astronomy, including the discovery of a new star. The work demonstrates how technology originally developed for the consumer market can now be used to accelerate the rate of scientific discovery.