Abstract:
This paper explores some of the performance-enhancing features of supercompilation in the context of the Reduceron --- a special-purpose graph-reduction machine. Two small examples are discussed in detail, highlighting areas where the two technologies interact. A strategy is introduced for countering a situation where supercompilation adversely affects Reduceron execution time. Performance results and other metrics are presented across a range of nineteen benchmarks highlighting the synergistic properties of supercompilation on the Reduceron. This paper represents work in progress.