Using neuro-fuzzy methodology in cache memory design

Article type: Research Article

Authors: Diab, H.^; | Abdul-Samad, R. | Saade, J.J. | Mrad, F.

Affiliations: Department of Electrical and Computer Engineering, Faculty of Engineering and Architecture, American University of Beirut, Beirut, Lebanon

Note: [] Corresponding author. Department of Electrical and Computer Engineering, P.O. Box 11-0236, Beirut, Lebanon. Tel.: +961 1 350000, ext. 3520; Fax: +961 1 744462; E-mail: diab@aub.edu.lb.

Abstract: Today cache memory is integrated into the design of all state-of-the-art computer systems. Their performance evaluation is gaining a lot of attention, because it significantly affects the performance of the entire computer system. Fuzzy logic, which has proven to be a very powerful and efficient way of tackling problems in many engineering areas is used in this paper to provide a novel simulation methodology for the evaluation of cache memory performance. This methodology can be used during the preliminary stages of cache memory design, to overcome the disadvantages of conventional methods. Advantages of fuzzy logic are used to provide a more simple, reliable and cost effective approach for cache memory performance evaluation, compared to conventional methods like trace driven simulation and hardware monitoring. Many complex approximations and time consuming testing of existing systems are thus avoided. The designer can study the effect of changing different design parameters of cache memory and investigate their effect on the miss ratio.

Journal: Journal of Intelligent and Fuzzy Systems, vol. 8, no. 4, pp. 289-300, 2000