摘要
给出一个主要针对Java高性能应用的Java自适应优化编译框架.它具有分布式系统结构,采用一个启发式算法搜索一个很大而复杂的优化空间以提高被编译程序的运行性能.在两个测试环境下测试16个测试程序的结果表明:它在测试100个点后能取得平均加速比为1.14和1.10,并且其中的78%和89%在头20次测试中即可获得.
Modem hetemgenous computing environment puts an increasing demand on not only high performance but also its le but is frequently unable to deliver high performance. An adaptive optimization framework for Java applications in high performance computing is presented. It has a distributed architecture and employs a heuristic algorithm to explore a large and complex optimization space for performance improvement. Experimental resuhs on 16 benchmarks and two environments show that it can achieve an average speedup of 1.14 and 1.10 respectively, and 78% and 89% of which can be obtained within the first 20 attempts.
出处
《暨南大学学报(自然科学与医学版)》
CAS
CSCD
北大核心
2006年第5期676-682,共7页
Journal of Jinan University(Natural Science & Medicine Edition)