Efficient Stream Processing in GHC and Its Evaluation on a Parallel Inference Machine

概要

論文の詳細を見る
A set of primitives to implement efficient stream processing in GHC, Guarded Horn Clauses, and its evaluation results on PIM-D, Parallel Inference Machine based on the Dataflow model, are presented. The language is efficient as the conventional procedural languages, because stream processing, the basic operation in GHC, is implemented as machine primitives rather than creating enormous merge processes. In order to support such primitives with preserving "1ogically correctness," only one extra bit shared flag added to every pointer is sufficient. The evaluation results show that performance is improved by introducing such primitives
一般社団法人情報処理学会の論文
1988-03-10