Optimisng parallel logic programming systems for scalable machines

摘要

Parallel logic programmign (PLP) systems have obtianed good performance on traditional bus-based shared-memory architectures.However,the scalable multiprocessors,being devleoped today pose new challenges.our experience with a sophisticated PLP system,Andorra-I,demonstrates that inded performance suffers greatly on modern architercures.In order to imporver perforance,we performm a detailed analysis of the cache beahivour of all Andorra-I data structures via execution driven simulation of a DASH-like multiprocessor.Based on this anlysis we optimise the Andorra-I code using 5 different techniques.Our results show that the techniques provide signhificant performance imporvements,leading to the conclusion that PLP systems can and should perform well on modern scalabel multiprocessors.