Scholarlink

<p>In recent years, heterogeneous multiprocessor computing systems have been widely adopted due to their powerful capability to accelerate task execution. However, the improvement of task execution efficiency is usually accompanied by a significant increase in energy consumption. Targeting the minimization of scheduling length for parallel applications under energy constraints, this paper proposes a novel scheduling algorithm based on the time–energy coefficient. Firstly, we establish mathematical models for tasks and heterogeneous processors, and formalize the scheduling problem as a constrained optimization problem. Then, the proposed algorithm is used to obtain the near-optimal start time and processor allocation for each task, so as to guarantee efficient task execution. Different from existing studies that ignore the differentiated influence of unit energy on the execution time of different tasks, the proposed algorithm takes both task execution time and the impact of unit energy on task execution time into account in energy allocation, and further designs an innovative weighted energy preallocation strategy. We conduct experiments on randomly generated DAGs and two practical application graphs. Experimental results show that the proposed algorithm reduces the scheduling length by 2–9% compared with representative existing algorithms, which fully verifies its effectiveness and superiority.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

TECSA: an energy-aware heuristic scheduling algorithm based on time–energy coefficients in heterogeneous multiprocessor systems

  • 期刊论文

  • 发表日期：

    2026年6月23日

查看全文
  • Yi Lu,
  • Jing Wu,
  • Jianhua Lu,
  • Wei Hu,
  • Dawei Jiang

摘要

In recent years, heterogeneous multiprocessor computing systems have been widely adopted due to their powerful capability to accelerate task execution. However, the improvement of task execution efficiency is usually accompanied by a significant increase in energy consumption. Targeting the minimization of scheduling length for parallel applications under energy constraints, this paper proposes a novel scheduling algorithm based on the time–energy coefficient. Firstly, we establish mathematical models for tasks and heterogeneous processors, and formalize the scheduling problem as a constrained optimization problem. Then, the proposed algorithm is used to obtain the near-optimal start time and processor allocation for each task, so as to guarantee efficient task execution. Different from existing studies that ignore the differentiated influence of unit energy on the execution time of different tasks, the proposed algorithm takes both task execution time and the impact of unit energy on task execution time into account in energy allocation, and further designs an innovative weighted energy preallocation strategy. We conduct experiments on randomly generated DAGs and two practical application graphs. Experimental results show that the proposed algorithm reduces the scheduling length by 2–9% compared with representative existing algorithms, which fully verifies its effectiveness and superiority.