Abstract
High precision integer arithmetic and rational computation algorithms, are targeted to loosely coupled systems. The realisability and possible advantages of parallel implementations are investigated. The parallel algorithms under investigation make use of higher-radix methods.A new type of high precision parallel multiplier is developed, along with a thorough performance analysis. It is also outlined how to achieve savings in signed-digit multipliers. A new type of parallel divider is proposed. New methods are defined to extract coarse-grain parallelism from rational computation, and fine-grain methods are examined.