Adaptive prefetching technique for shared virtual memory

被引：2

作者：

Lee, SK ^{[1
]}

Yun, HC ^{[1
]}

Lee, JW ^{[1
]}

Maeng, S ^{[1
]}

机构：

[1] Korea Adv Inst Sci & Technol, Comp Architecture Lab, Yusong Gu, Taejon 305701, South Korea

来源：

FIRST IEEE/ACM INTERNATIONAL SYMPOSIUM ON CLUSTER COMPUTING AND THE GRID, PROCEEDINGS | 2001年

关键词：

D O I：

10.1109/CCGRID.2001.923237

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

Though shared virtual memory, (SVM) systems promise low cost solutions for high performance computing, they suffer from long memory latencies. These latencies are usually caused by repetitive invalidations on shared data. Since shared data are accessed through synchronizations and the patterns by which threads synchronizes are repetitive, a prefetching scheme based on such repetitiveness would reduce memory, latencies. Based on this observation, we propose a prefetching technique which predicts future access behavior by analyzing access history per synchronization variable. Our technique was evaluated on an 8-node SVM system using the SPLASH-2 benchmark. The results show that our technique could achieve 34% - 45% reduction in memory access latencies.

引用

页码：521 / 526

页数：6