基于分布函数的WCET快速估计

在实时软件系统中,软件时间性能的分析与评估技术是一个重要的课题,然而随着CPU的结构越来越复杂,采用传统的模拟底层硬件执行的方法越来越困难。而基于分布函数的最坏执行时间(Worst Case Execution Time,WCET)估计方法从概率角度出发,可以绕过复杂的底层硬件建模,估计程序的最坏执行时间。首先对TI TMS320C6713 DSP汇编代码进行基本块的划分,以基本块为结点构建程序流图；然后用贝塔分布模拟每条指令的运行时间并采用改进的计划评审技术(Program Evaluation and Review Technique,PERT)确定贝塔分布相关参数,指令叠加后用正态分布模拟每个基本块的执行时间；最后利用基于路径的方法得到整个程序的最坏执行时间。实验结果表明此方法是可行的和合理的。

Fast Estimation of WCET Based on Distribution Function

In the real-time software systems,the software-time performance analysis and evaluation techniques are important.But with the structure of CPU being more and more complex,traditional methods based on underlying hardware simulations are becoming increasingly difficult.Based on distribution function,the worst-case execution time(WCET) estimation method that is from a probabilistic perspective,can bypass the complex underlying hardware modeling and estimate the worst-case execution time.This article first divided TI TMS320C6713 DSP assembly code into basic blocks to build the program flow diagram.Then the run time of each instruction was simulated based on beta distribution whose parameters are determined according to the improved program evaluation and review technique(PERT).The instruction execution time was superimposed based on normal distribution to simulate the run time of each basic block.Finally, the worst execution time of the whole program was gotten by using a path based approach.Experimental results show that this method is feasible and reasonable.