Design of a Cross-Heterogeneous Hardware Computility Scheduling Platform for Heterogeneous Computing Systems
With the development of underwater acoustic signal processing technology and the ever-increasing business demands, the traditional single computing platform can no longer meet the growing computing needs of the business, heterogeneous computing platforms have emerged accordingly. However, systems are constrained by factors such as size and power consumption In practical scenarios, and the hardware platform has limited computing nodes, making it difficult to increase computilityby expanding nodes. Therefore, it becomes increasingly urgent to improve the performance of the heterogeneous computing platform. This article proposes a cross-heterogeneous hardware computility scheduling platform suitable for heterogeneous computing systems. Through the heterogeneous computing interfaces, the computility scheduling platform and the heterogeneous computing agents, masks the differences between heterogeneous computing software and hardware, achieving dynamic scheduling of cross-heterogeneous computility. The experiments demonstrate that the computility scheduling platform proposed in this paper, integrated with underwater acoustic signal processing applications for scheduling simulation, has validated its effectiveness in heterogeneous computing systems.
Authors
- Cheng Jiang (ORCID: https://orcid.org/0000-0002-6682-7161)
- Jie Pan (ORCID: https://orcid.org/0000-0002-9993-1379)
- Yalong He
- Jun Chai (ORCID: https://orcid.org/0000-0002-7133-3481)
Institutions
- Hangzhou Institute of Applied Acoustics (CN)
Publication Details
- Published
- 2024-10-11
- DOI
- https://doi.org/10.1109/cbase64041.2024.10824645
- Primary Topic
- Distributed and Parallel Computing Systems
- Type
- article
- Field-Weighted Citation Impact
- 0.00