Design of Dynamic Scheduling Algorithm for Parallel Data Processing in Multi-Core Chips

Multi-core processors are widely used in modern distributed systems, but traditional scheduling algorithms face challenges in task response time and resource utilization under high-load scenarios. This paper aims to optimize task scheduling efficiency in multi-core environments through a novel dynam...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2025 2nd International Conference on Intelligent Computing and Robotics (ICICR) s. 432 - 436
Hlavní autoři: Kang, Boce, Cheng, Yu, Hao, Zihe, Tian, Zhen
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 16.05.2025
Témata:
distributed systems
Dynamic scheduling
dynamic scheduling algorithm
Heuristic algorithms
load balancing
Load management
multi-core processing
Multicore processing
Optimization
parallel computing
performance optimization
resource allocation
Resource management
Scalability
Scheduling algorithms
task priority evaluation
Throughput
Time factors
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Multi-core processors are widely used in modern distributed systems, but traditional scheduling algorithms face challenges in task response time and resource utilization under high-load scenarios. This paper aims to optimize task scheduling efficiency in multi-core environments through a novel dynamic scheduling approach. We design a dynamic scheduling algorithm based on multi-factor evaluation, which integrates three key components: a task priority evaluation model using weighted calculations of task urgency, resource demands, and dependency relationships; an improved minimum completion time (IMCT) strategy for load balancing; and an adaptive resource allocation method with threshold control mechanism. The experimental evaluation is conducted on the Intel i712700 K processor platform using the PARSEC benchmark suite. Results demonstrate significant performance improvements: system throughput increases by 35.7 \%, CPU utilization reaches 87.6 \%, and average task response time reduces by 42.3 \% compared to traditional scheduling algorithms. In high-load environments, the algorithm exhibits excellent scalability, maintaining scheduling overhead within 0.8 ms while processing 1000 concurrent tasks. The proposed algorithm provides an effective solution for task scheduling optimization in large-scale distributed systems, particularly suitable for scenarios with dynamic workload characteristics.
DOI:10.1109/ICICR65456.2025.00081