A Design Space Exploration of VLSI Extreme Machine Learning

• Published in: Journal of physics. Conference series 2021-07, Vol.1964 (6), p.62009
• Main Authors: Gopi, B, Arulkumaran, G, Parameshwaran, R, Khan, Vijitha
• Format: Article
• Language: English
• Subjects: Data processing; Durability; Electromagnetic induction; Electronic equipment; Field programmable gate arrays; FPGA; Integrated circuits; Machine learning; Modules; Radiation; Radiation measurement; Radioactive contaminants; Radioactivity; Redundancy; Solar radiation; Space; System reliability; VLSI
• ISSN: 1742-6588; 1742-6596
• DOI: 10.1088/1742-6596/1964/6/062009

Radiation acceptance in FPGAs is an increasing priority, especially for dependable data processing in electronic equipment used during engineering and geostationary operations. A loss in FPGA board’s durability based on a single impact made by radioactive contaminants is the impetus behind the whole study. Inefficiency is a widely used method to increase the potential of radioactivity systems for data integrity. In excess usage of the zone, delay, and transmission range, durability brings with an overload. Also, with replication induction methods or even the quantity of backup stages, the defective system designs differ in configuration and resources use. The electromagnetic background varies based on the atmosphere and space climatic conditions and during the project’s operational time cycle. About the particular radiation level can also minimize the overhead costs attributable to maintenance at a run. In this article, we design a Dynamic Reliability Management scheme that uses, perceives, determines any appropriate durability degree, and implements operated reorganization using the radiation details, thereby varying the system reliability of the target computing modules. DRM specially developed flow produces a catalogue of similar compatible circuit applications with different output factor amplitudes. Spin software flow picks a necessary redundancy level using the radioactivity data and reconstructs the computational module with the relevant compatible module. The consequences of defects, errors and deficiencies caused by planetary protons emanating from solar radiation on ambient electronic structures in many sectors will begin with a simple but detailed analysis. Prevention steps against particle adverse effects, looking at different levels of both the architecture system from material to the device level, are applied. Challenges are also addressed to maintain effectiveness in the future advancement of technology. For random - access records, switch, GPU and operating networks, such problems in control measures are raised.