Abstract Dynamic Voltage and Frequency Scaling (DVFS) computing platforms are highly effective in reducing energy consumption by dynamically adjusting the operating frequency and voltage of processing units within predefined operating pairs. By selectively scaling down the execution frequency of application tasks, significant energy savings can be achieved while preserving timing constraints. For applications composed of dependent tasks, energy-aware frequency scaling is predominantly addressed through the Scaling Axiomatic Approach ( SAA ), which exploits task slack to enable safe frequency reduction but incurs a high computational cost due to repeated global timing recalculations. To mitigate this limitation, the GinGa approach was proposed to reduce the computational complexity, albeit with a degradation in energy optimization effectiveness. This paper introduces the Scaling Axiomatic Approach Replacement ( SaaR ), a low-complexity, compile-time mechanism designed as a principled replacement for SAA . While preserving the axiomatic foundation of slack-based frequency scaling, SaaR restructures the computation through bounded and localized timing-update mechanisms and a dedicated time-updating criterion, thereby eliminating repeated global recomputation. As a result, SaaR achieves energy savings comparable to those of SAA while significantly reducing the computational complexity. Experimental results confirm that SaaR outperforms GinGa and provides an effective balance between energy optimization and execution efficiency on DVFS-enabled computing platforms.

Summary form only given. The heterogeneous computing environment is an interesting computing platform due to the fact that a single parallel architecture may not be adequate for exploiting all of a program's available parallelism. In some cases, heterogeneous systems have been shown to produce higher performance for lower cost than a single large machine. Task scheduling is the key issue when aiming at high performance in this kind of environment. A large number of scheduling heuristics have been presented in the literature, most of them target only homogeneous computing systems. We present a simple scheduling algorithm based on list-scheduling and task-duplication on a bounded number of heterogeneous machines called heterogeneous critical parents with fast duplicator (HCPFD). The analysis and experiments have shown that HCPFD outperforms on average all other higher complexity algorithms.

Efficient task scheduling of computationally intensive applications is one of the most essential and difficult issues when aiming at high performance in heterogeneous computing environments. Although a large number of scheduling heuristics have been presented in the literature, most of them target only homogeneous computing systems. In this paper we present a simple list-scheduling heuristic for a bounded number of heterogeneous machines called Heterogenous Critical Parent Trees (HCPT). The analysis and experiments have shown that HCPT provides comparable or even better results together with low complexity.