Knowledge Driven Dynamic Frequency Scaling
1 Reviews
•
5.0
Abstract
This paper proposes a novel, knowledge driven dynamic frequency scaling approach for dynamic power management. It uses a neural network model for learning the behavior of various applications and then uses this knowledge for predicting the optimal processor frequency at any instance of time. The experimental results show the elegance of the approach. The approach may be suitably modified for application in multicore processors and SoCs(System on Chips).
Details
| Title: | Knowledge Driven Dynamic Frequency Scaling |
| Subjects: | Engineering |
| More Details: | View PDF |
| Report Article: | Report |
Submission History
From:
Anirban Lahiri
[View Profile]
Date of Publication:
July 15, 2025, 8:21 a.m. UTC
Sunreeta Sen(Sunreeta_Sen)
Great work on knowledge-driven dynamic frequency scaling
This paper presents a robust knowledge-driven dynamic frequency scaling technique using neural network–based workload characterization. The proposed approach is well designed, efficiently predicts optimal processor frequencies at runtime, and demonstrates strong potential for scalable power management in multicore processors and SoC architectures.