Multiprocessor scheduling by reduction to uniprocessor: an original optimal approach

Abstract

Optimal multiprocessor real-time schedulers incur significant overhead for preemptions and migrations. We present RUN, an efficient scheduler that reduces the multiprocessor problem to a series of uniprocessor problems. RUN significantly outperforms existing optimal algorithms with an upper bound of O(log m) average preemptions per job on m processors (fewer than 3 per job in all of our simulated task sets) and reduces to Partitioned EDF whenever a proper partitioning is found.

An unfair semi-greedy real-time multiprocessor scheduling algorithm

“…With regard to optimal real-time multiprocessor scheduling algorithms, both RUN and U-EDF outperform all other existing algorithms [2,19] in terms of the generated scheduling overheads. In this paper, we show how our novel algorithm achieves better performance in terms of the number of preemptions and migrations.…”

“…In [19], a new optimal algorithm, namely, RUN (Reduction to Uniprocessor), has been presented. RUN introduces a new approach to multiprocessor scheduling in which the scheduling problem is reduced to a series of uniprocessor ones using a dualization technique.…”

“…The first problem is that sometimes, we have consecutive deadlines with only one unit of time, such as [14,15), [15,16), [16,17), [19,20), and [20,21). This will result in numerous preemptions and migrations.…”

“…time slices as mentioned previously, which are bounded by two successive deadlines, and the end of each TL-plane corresponds to the deadline of a task in the system. Hence, tasks are marshalled in the intervals [0, 5), [5,7), [7,10), [10,14), [14,15), [15,16), [16,17), [17,19), [19,20), [20,21), [21,25), [25,26), [26,28), and [28,29), which correspond to the first 14 TL-planes, after which all tasks would finish at least one period of their executions. This means that at the beginning of each TL-plane, all tasks have to be allocated local executions proportional to their utilizations and marshalled until the end of the time slice at which they must all be preempted.…”

A Two-Level Task Scheduler on Multiple DSP System for OpenCL

“…is totally free and has perfect support for its own DSP processors. DSP/BIOS kernel is a priority based preemptive [10] scheduler. It is designed to be used by applications that require real-time scheduling and synchronization.…”