Multi-objective scheduling of heterogeneous parallel systems using the VITS algorithm

Heterogeneous systems are widely used to implement a wide range of critical services. Resource management is a key challenge in parallel systems and becomes more complicated when system resources are heterogeneous. The issue of resource heterogeneity from different aspects simultaneously has not rec...

Full description

Saved in:
Bibliographic Details
Published in:The Journal of supercomputing Vol. 81; no. 7; p. 842
Main Authors: Bakhoda, Saeedeh, Abdollahi Azgomi, Mohammad, Ebrahimi Dishabi, Mohammad Reza
Format: Journal Article
Language:English
Published: New York Springer US 08.05.2025
Springer Nature B.V
Subjects:
Compilers
Completion time
Computer Science
Efficiency
Energy consumption
Genetic algorithms
Heterogeneity
Interpreters
Multiple objective analysis
Objectives
Optimization
Processor Architectures
Profitability
Profits
Programming Languages
Resource management
Run time (computers)
Scheduling
Simulated annealing
Tabu search
Throughput and profitability
Total completion time and runtime
Multi-objective scheduling
Improved allocation vector
Heterogeneous parallel systems
Extended tabu search
ISSN:1573-0484, 0920-8542, 1573-0484
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Heterogeneous systems are widely used to implement a wide range of critical services. Resource management is a key challenge in parallel systems and becomes more complicated when system resources are heterogeneous. The issue of resource heterogeneity from different aspects simultaneously has not received much attention in the literature. This study addresses this issue and investigates multi-objective scheduling in heterogeneous parallel environments regarding processing speed and cost. The main objectives are increasing the system's throughput by completing more tasks, improving the system's profitability, and reducing the total completion time and runtime. Proper task allocation and scheduling on heterogeneous resources are effective in achieving goals. This paper introduces a vector allocation and scheduling approach improved by an extended tabu search-based strategy. In the proposed methodology, abbreviated as the VITS, a vector approach is first used to allocate and schedule tasks on heterogeneous resources. Then, the vector is improved using an extended tabu search-based strategy to obtain better results for the objectives. The proposed methodology utilizes several efficient genetic algorithm mutation methods to produce better and high-quality solutions. The proposed algorithm was simulated and evaluated on several benchmark files of different sizes, containing a minimum of 100 tasks and 10 heterogeneous resources and a maximum of 500 tasks and 80 heterogeneous resources. The simulation results are compared with other selected algorithms, including a vector allocation method improved by genetic (VIGA) and simulated annealing (VISA) algorithms. The evaluation of the results verifies the superiority of the proposed algorithm. Comparing the results on the various test files demonstrates that the proposed VITS compared to VIGA and VISA have respectively an average percentage improvement of 6.9 and 7.7 in task completion rate, 3.8 and 5.3 in profitability, 4 and 2.9 in total task completion time, and 66 and 500 in the algorithm's runtime.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1573-0484
0920-8542
1573-0484
DOI:10.1007/s11227-025-07293-9