基于嵌套循环指令分析的片上存储器分配策略

Nested loops analysis-based allocation scheme for on-chip memory

本文提出了一种无Cache情况下基于嵌套循环指令分析的片上存储器(On-chip memory)分配策略。该策略分析程序中循环指令,控制划分粒度将所有函数切割成块,然后使用背包算法和优先级算法组合的分配算法,选择合适块放入片上存储器,达到优化程序性能的目的。实验结果表明,该策略能够显著提高程序性能,平均提高一倍,甚至更高,同时它能够预知优化后程序执行时间的变化,最高误差为2%。

An allocation scheme for on-chip memory based on nested loops analysis is presented. The scheme analyzes the loop instructions in the whole application, splits all the functions into blocks under partition granularity control, and allocates some blocks in on-chip memory with the combined algorithm consisting of knapsack and priority algorithms to improve performance. Experimental results show that the scheme has ability to improve significant performance significantly, and predict the application＇s execution time after optimization with a maximum error of 2%.