Vectorization of an applicative language: Current results and future directions

摘要

The most important and challenging problem facing researchers today is the effective use of parallel processors. Few would deny that the biggest obstacle is the inadequacy of current software tools -- an inadequacy born of imperative programming. As an alternative, applicative and functional programming languages provide a cleaner and simpler parallel programming model. They hide architectural idiosyncrasies, guarantee determinancy, enforce software engineering principles, and in many ways simplify compilation. Regrettably these languages have acquired a reputation for inefficiency. In this report, we show that applicative programs do not require special hardware for efficient execution, and with little effort can automatically exploit concurrent and vector processors. 17 refs., 1 fig., 5 tabs.