Comparison of empirical success rates of global vs. partitioned fixed-priority and EDF scheduling for hard real time

Improvements in schedulability tests for global fixed-priority and EDF scheduling in a homogeneous multiprocessor (symmetric multiprocessing) environment have shown that the worst-case guaranteed achievable utilization levels for global EDF scheduling equals what can be achieved with partitioned scheduling, and both ways of applying EDF scheduling out-perform fixed-priority scheduling, for sets of independent periodic or sporadic hard-deadline tasks with deadline equal to period. However, less is known about the comparative performance of the partitioned vs. global and EDF vs. fixed-priority approaches in the average and without the restriction that deadline equal period, and particular which of the known combinations of a scheduling algorithm and a sufficient a priori test of schedulability is more likely to succeed in verifiably scheduling a set of tasks to meet all deadlines. This paper compares the performance of several such combinations on a variety of pseudo-randomly chosen sets of sporadic tasks.