VEERBENCH - an intelligent computing framework for workload characterisation in multi-core heterogeneous architectures. (2018)

• Record Type: Journal Article
• Authors: Sundaresan, Yokesh Babu; Durai, M.A. Saleem
• Is Part Of: World review of science, technology and sustainable development. Volume 14:Number 1(2018)
• Page Start: 34
• Page End: 51
• Publication Date: 2018
• Subjects: VEERBENCH -- extreme learning machines -- ELM -- formulation of adaptive and cognitive energy -- FACE -- MacPACT -- ESEC -- internet of things; Sustainable development -- Periodicals; Appropriate technology -- Periodicals; Technology transfer -- Periodicals; Globalization -- Periodicals; 338.92705
• Languages: English
• ISSNs: 1741-2242

Abstract

Multi-core heterogeneous architectures are playing the important role in server, mobile and all commercial devices. With the advent of internet of things in today's applications and increase in the workloads inputs, predictive simulating and computing tools are mandatory for the effective implementation of the multi-core heterogeneous architectures for the different applications. Many tools such as MacPACT, ESEC has been into existence but an intelligent computing framework tool for the predictive selection of the cores depending on the workloads remains in the darker side of the research. Hence the new computing framework called VEERBENCH has been proposed which works on the learning and training mechanisms for the usage of the cores in the heterogeneous architectures depending on the workloads. The framework uses the fuzzy clustering with the extreme learning machines and formulation of adaptive and cognitive energy (FACE) rule sets which are used for the energy and performance-based allocation of the cores. The proposed knowledge-based test bench has been compared with the other tools such as MACPACT, ESEC and with the other energy-based scheduler benchmarks and the obtained results are shown.