摘要： 探讨了如何提高CC-NUMA结构下共享变量程序的并行效率。主要介绍了几种有效的负载均衡策略和减少共享存储访问延迟的优化 手段。通过分析可以看出,通过合适的优化方法,CC-NUMA结构下共享变量的应用程序可以取得好的并行效率。

关键词: Cache一致性非均匀存储器访问, 负载均衡, 任务粒度, 访问延迟

Abstract: This paper discusses how to improve the shared variable program performance on CC-NUMA parallel computer. It provides several effective methods of load balancing and reducing shared memory access latency. Through the analysis, it can conclude that the shared variable program will has a high performance if it uses a properly optimization method.

中图分类号: 

• TP311.1