Allocating resources based on a model of coarse‐grained reconfigurable architecture. Issue 10 (12th July 2019)

• Record Type: Journal Article
• Authors: Zhang, Huizhen; Pan, Yubiao; Zhang, Yiwen; Wang, Cheng
• Is Part Of: Journal of engineering. Volume 2019:Issue 10(2019)
• Page Start: 7272
• Page End: 7278
• Publication Date: 2019-07-12
• Subjects: reconfigurable architectures -- instruction sets -- field programmable gate arrays; allocating resources -- coarse‐grained reconfigurable architecture -- reconfigurable resource utilisation -- reconfigurable instruction set processors -- RISPs -- field programmable gate array chips -- graph multicolouring allocating algorithm -- simplified reconfigurable resource model -- FPGA‐based coarse‐grained reconfigurable resources -- graph multicolouring method -- state‐of‐the‐art resource utilisation method; Engineering -- Periodicals; Engineering; Electronic journals; Periodicals; 620.005
• DOI: 10.1049/joe.2018.5230
• Languages: English
• ISSNs: 2051-3305

Abstract

Abstract : Reconfigurable resource utilisation is a key factor that impacts the performance of reconfigurable instruction set processors (RISPs), which can customise their instruction set for special applications. In this study, the authors propose a novel method to improve reconfigurable resource utilisation of RISPs based on field programmable gate array (FPGA) chips. The method provides a graph multi‐colouring allocating algorithm based on a simplified reconfigurable resource model. In the model, they construct three kinds of FPGA‐based coarse‐grained reconfigurable resources and abstract their attributes mainly through their location and area. The authors' algorithm employs a graph multi‐colouring method to allocate the coarse‐grained reconfigurable resources to custom instructions for RISPs. Experimental results demonstrate that their model and algorithm reinforce a state‐of‐the‐art resource utilisation method for RISPs, and reveals some interesting regularities which provide creditable new insights of reconfigurable resource utilisation.