On-line scheduling of imprecise computations to minimize error

Abstract

This paper describes three algorithms for scheduling preemptive, imprecise tasks on a processor to minimize the total error. Each imprecise task consists of a mandatory task followed by an optional task. Some of the tasks are on-line; they arrive after the processor begins execution. The algorithms assume that when each new on-line task arrives, its mandatory task and the portions of all the mandatory tasks yet to be completed at the time can be feasibly scheduled to complete by their deadlines. The algorithms produce for such tasks feasible schedules whose total errors are as small as possible. The three algorithms are designed for three types of task systems: (1) when every task is on-line and is ready upon its arrival, (2) when every on-line task is ready upon arrival but there are also off-line tasks with arbitrary ready times, and (3) when on-line tasks have arbitrary ready times. Their running times are 0 (n logn ), 0 (n log n ), and 0 ( n log2 n ), respectively.

Power-Aware Scheduling for Parallel Security Processors with Analytical Models

“…To optimize the power or energy consumption in real-time systems, we propose a heuristic algorithm to schedule tasks under the condition of variable supply voltage. The proposed scheduler maintains a list, called the reservation list [30], in which these tasks are sorted by deadlines. Since each periodic task arrives with a certain periodicity, we can get the information about the arrivals and deadlines of tasks in a given interval.…”

Optimal reward-based scheduling for periodic real-time tasks

“…Observe that since f i t i is non-decreasing and concave by assumption, both w i t i and f 0 i t i is non-increasing and positive valued. Equation (18) implies that the marginal returns When it is not possible to find a closed formula, following exactly the approach presented in [10,13], we solve in the equation…”

“…The algorithms proposed for imprecise computation applications concentrate on a model that has an upper bound on the execution time that could be assigned to the optional part [5,15,18]. The aim is usually to minimize the (weighted) sum of errors.…”

“…The aim is usually to minimize the (weighted) sum of errors. Several efficient algorithms are proposed to solve optimally aperiodic scheduling problem of imprecise computation tasks [15,18]. A common assumption in these studies is that the quality of the results produced is a linear function of the precision error; consequently, the possibility of having more general error functions is usually not addressed.…”

“…Reward functions with 0/1 constraints, where no reward is accrued unless the entire optional part is executed, or step reward functions have also received some interest in the literature. Unfortunately, this problem has been shown to be NP-Complete in [18].…”