| 引用本文: | 欧钢.一种应用于数据通道综合的定向搜索流水调度算法.[J].国防科技大学学报,2002,24(2):36-43.[点击复制] |
| OU Gang.A Directed Searching Pipeline Scheduling Algorithm for Datapath Synthesis[J].Journal of National University of Defense Technology,2002,24(2):36-43[点击复制] |
|
| |
|
|
| 本文已被：浏览 6328次 下载 5948次 |
| 一种应用于数据通道综合的定向搜索流水调度算法 |
| 欧钢 |
|
(国防科技大学 ATR重点实验室，湖南 长沙 410073)
|
| 摘要: |
| 流水调度是专用数字信号处理器高层综合中一个困难而急待解决的问题。给出了一种定向搜索流水调度算法，目标是使全面考虑了运算单元、寄存器和互连的硬件代价最小化。它作为一种利用启发信息的迭代算法,一方面克服了确定性算法爬峰能力差、易于陷于局部极值的缺点,另一方面启发信息的利用加快了搜索过程。典型设计实例显示算法性能达到或超过了目前流水调度文献报道的最好性能。 |
| 关键词: 高层综合 流水时序调度 数字信号处理 |
| DOI： |
| 投稿日期：2001-10-15 |
| 基金项目: |
|
| A Directed Searching Pipeline Scheduling Algorithm for Datapath Synthesis |
| OU Gang |
|
(ATR National Lab, National Univ. of Defense Technology ,Changsha 410073,China)
|
| Abstract: |
| In the high-level synthesis of application-specific DSP, pipeline scheduling is a difficult and urgent problem Concerning the global optimization of functional units, registers and interconnections, we present a novel algorithm called directed searching pipeline scheduling, which is an iterative algorithm utilizing heuristic information. The heuristic information speeds up the searching process, and the weakness of deterministic algorithms, which are vulnerable to trip in local optimal, poor at hill-climbing, is overcome in this algorithm. Typical design examples show that the performance of this transformation-based algorithm has reached or surpassed the best performance of pipeline scheduling algorithms reported so far. |
| Keywords: high-level synthesis scheduling digital signal processing |
|
|
|
|
|
