Energy-Quality-Time Optimized Task Mapping on DVFS-Enabled Multicores

Abstract

Abstract-Multicore architectures have great potential for energy-constrained embedded systems, such as energy-harvesting wireless sensor networks. Some embedded applications, especially the real-time ones, can be modeled as imprecise computation tasks. A task is divided into a mandatory subtask that provides a baseline Quality-of-Service (QoS) and an optional subtask that refines the result to increase the QoS. Combining dynamic voltage and frequency scaling, task allocation and task adjustment, we can maximize the system QoS under real-time and energy supply constraints. However, the nonlinear and combinatorial nature of this problem makes it difficult to solve. This work first formulates a mixed-integer non-linear programming problem to concurrently carry out task-to-processor allocation, frequencyto-task assignment and optional task adjustment. We provide a mixed-integer linear programming form of this formulation without performance degradation and we propose a novel decomposition algorithm to provide an optimal solution with reduced computation time compared to state-of-the-art optimal approaches (22.6% in average). We also propose a heuristic version that has negligible computation time.

Approximation-aware Task Deployment on Asymmetric Multicore Processors

“…Not all the optional subtasks can be fully executed, i.e., {O l } (∀l ∈ N ), as this violates the real-time and energy constraints. The QoS function of the tasks is adopted from [12], [13]. The use of arbitrary task graphs does not affect our problem formulation, as different task graphs only change the values of the problem parameters {A, C, D, f , b 1 , b 2 }.…”

“…The extension of the optimal deployment methods from dependent to independent tasks and from SMPs to AMPs is not straightforward, as additional nonlinear constraints are usually introduced into the problem. On the other hand, the heuristics of previous works usually adopt multi-step optimizations, e.g., [4], [12], [13] and they require that the variables can be solved separately. Hence, they need to be re-designed when the problem formulation changes.…”

“…, n), where m + n = k.Remark Since the objective function of the PP2 only contains the continuous variables y, and the MP only considers the binary variables x, an auxiliary variable (continuous)Φ is introduced into the MP as the objective function.Φ and Φ have the same physical meaning, as shown in the Proof 5.1. Although the MP includes the continuous variableΦ, this problem can be solved by only considering the binary variables x[13].Lemma 5.1: Solving the MP (SP), we obtain a lower bound Φ l (k) (upper bound Φ u (k)) of the optimal value of PP2, Φ * .2) MP and SP Iterations: At the initial iteration, i.e., k = 0, we set {x(0)|Ax(0) b 1 }, Φ l (0) = −∞ and Φ u (0) = +∞. Denote ε as a small positive tolerance.…”

Mapping imprecise computation tasks on cyber-physical systems

“…Other works consider the QoS-aware task mapping problem using the IC-task model and having as a goal to maximize the QoS under a set of real-time and/or energy supply constraints, e.g., [4,6,[12][13][14][15][16]. The target platform considered in [12] is a single core platform.…”

“…Table 1 provides a summary of some representative papers from the literature. Task mapping is a well-known problem in embedded system community, no matter the tasks are precise or imprecise, and the multicore platforms are homogeneous [4,7,8,[12][13][14][15] or heterogeneous [3,[9][10][11]16]. However, the work deals with task mapping problem on networked system is rare, and most of them consider precise tasks [2,6,[17][18][19][20].…”