摘要
在可重构计算领域,布局硬件任务所产生的碎片是影响系统资源利用率和任务拒绝率的关键因素之一.本文提出了一种对布局碎片进行量化的方法QFOAC,该方法将每个已占用的可重构计算单元对总碎片量化值的贡献求和,得到可重构芯片布局碎片量化值.QFOAC方法对碎片的抽象准确直观,且运算时间复杂度为常数.实验表明,将QFOAC碎片量化方法应用于布局器,可显著提高芯片利用率,降低任务拒绝率,且基本不增加系统开销,适用于动态重构和实时系统.
The fragmentation of resource placement is a critical factor to affect CU (Chip Utilization) and TRR (Task Rejection Ratio) in the field of reconfigurable computing. A method QFOAC that quantifies the fragments of placement was proposed in this paper, QFOAC sums the fragment contribution of every occupied reconfigurable computing unit to get the fragment value of the chip. QFOAC abstracts the placement fragments manifestly and precisely, and it has a constant time complexity. Simulation results showed that QFOAC can remarkably improve CO and TRR with little overhead, and it is suitable for dynamic reconfigurable system and real-time system.
出处
《小型微型计算机系统》
CSCD
北大核心
2007年第5期944-947,共4页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60273042)资助
中国科学院创新基金
安徽省自然科学基金项目(03042203)资助
关键词
可重构
布局碎片
资源碎片量化
碎片整理
reconfigurable
fragment of placement
quantification of resource fragments
defragmentation