Reconfigurable computing and task scheduling for active storage service processing

  • Authors:
  • Yu Zhang;Dan Feng

  • Affiliations:
  • Wuhan National Lab for Optoelectronics, School of Computer Science, Huazhong University of Science and Technology, Wuhan, China;Wuhan National Lab for Optoelectronics, School of Computer Science, Huazhong University of Science and Technology, Wuhan, China

  • Venue:
  • ARC'10 Proceedings of the 6th international conference on Reconfigurable Computing: architectures, Tools and Applications
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Active storage is a promising solution for data processing. Many existing active storage systems utilize spare CPU processing power on the storage controllers to execute active service. The software approach is very flexible for various applications, but the performance is quite low for computational-intensive service tasks. This paper presents a low power SoC-based hardware solution for high performance active storage. The basic idea is incorporating flexible reconfigurable accelerators in storage controllers for efficient active service processing. In order to reduce the reconfiguration latency, we also proposed hybrid configuration prefetching and configuration caching algorithms according to the task pattern. In the experiments, we presented results on applications such as data security, data compression and image processing with our FPGA prototype for an active storage processor.