The key factor in reducing the consumed energy when dependent-tasks applications are scheduled on DVFS-enabled computing platforms is task execution time slots. The unique and axiomatic approach to reduce the energy consumption on such platforms involves scaling down the execution frequency of each task within its execution time slot, provided a suitable scaling-down frequency is available. Regrettably, scheduling algorithms often shrink task execution time slots due to minimizing task completion times. This paper presents BlueMoon, a mechanism that reschedules the application tasks to extend the execution time slot of each task while ensuring that the overall completion time of the application tasks remains unaffected. BlueMoon is implemented and tested on numerous schedules of application graphs. The experimental results, conducted through computer simulations, demonstrate that BlueMoon substantially extends the execution time slots of tasks when compared to other mechanisms.

当依赖任务应用程序在支持 DVFS 的计算平台上进行调度时，减少能耗的关键因素是任务执行时隙。减少此类平台上的能耗的独特且公理化的方法涉及在其执行时隙内缩减每个任务的执行频率，前提是有合适的缩减频率可用。遗憾的是，调度算法经常由于最小化任务完成时间而缩短任务执行时隙。本文提出了BlueMoon，这是一种重新调度应用程序任务的机制，以延长每个任务的执行时隙，同时确保应用程序任务的整体完成时间不受影响。BlueMoon在大量应用程序图表上进行了实施和测试。通过计算机模拟进行的实验结果表明，与其他机制相比， BlueMoon大大延长了任务的执行时隙。