Architecture Independent Massive Parallelization of Divide-and-Conquer Algorithms

We present a strategy to develop in a functional setting cor rect e cient and portable Divide and Conquer DC programs for mas sively parallel architectures Starting from an operational DC program mapping sequences to sequences we apply a set of semantics preserving transformation rules which transform the parallel control structure of DC into a sequential control ow thereby making the implicit data par allelism in a DC scheme explicit In the next phase of our strategy the parallel architecture is fully expressed where architecture dependent higher order functions are introduced Then due to the rising commu nication complexities on particular architectures topology dependent communication patterns are optimized in order to reduce the overall communication costs The advantages of this approach are manifold and are demonstrated with a set of non trivial examples