Analysis and Calculation of Software Residual Time for Scheduling Applications

Often a software program’s execution time is characterized by a single value (either mean or worst-case). When multiple tasks are concurrently scheduled with time quanta such that scheduling is based on partial execution (a common scenario in multimedia systems) a more accurate estimation of residual time, R(t), is desired than can be obtained from just the initial mean value, in order to have effective scheduling decisions. Using the additional information that some programs are yet active after an initial execution period, the residual time approach can provide more accurate estimation, and therefore more effective scheduling, in time-sensitive situations. Our approach of residual time calculation achieves better system utilization in multimedia applications when the variance in task execution time results in tasks completing earlier than worst-case. We developed an analytic model for computing expected residual execution time, R(t), of software programs from their execution time and probability distributions. Furthermore, we developed and evaluated several methods to reduce the size of the execution time and probability vectors. The approach goes as beyond whatever a mean value can provide, while being efficient in CPU usage.