Exportar Publicação

Villon, L. A. Q., Susskind, Z., Bacellar, A. T. L., Miranda, I. D. S., Araújo, L. S. de., Lima, P. M. V....Dutra, D. L. C. (2023). A conditional branch predictor based on weightless neural networks. Neurocomputing. 555

L. A. Villon et al.,  "A conditional branch predictor based on weightless neural networks", in Neurocomputing, vol. 555, 2023

@article{villon2023_1715509880941,
	author = "Villon, L. A. Q. and Susskind, Z. and Bacellar, A. T. L. and Miranda, I. D. S. and Araújo, L. S. de. and Lima, P. M. V. and Breternitz Jr., M. and John, L. K. and França, F. M. G. and Dutra, D. L. C.",
	title = "A conditional branch predictor based on weightless neural networks",
	journal = "Neurocomputing",
	year = "2023",
	volume = "555",
	number = "",
	doi = "10.1016/j.neucom.2023.126637",
	url = "https://www.sciencedirect.com/journal/neurocomputing"
}

TY  - JOUR
TI  - A conditional branch predictor based on weightless neural networks
T2  - Neurocomputing
VL  - 555
AU  - Villon, L. A. Q.
AU  - Susskind, Z.
AU  - Bacellar, A. T. L.
AU  - Miranda, I. D. S.
AU  - Araújo, L. S. de.
AU  - Lima, P. M. V.
AU  - Breternitz Jr., M.
AU  - John, L. K.
AU  - França, F. M. G.
AU  - Dutra, D. L. C.
PY  - 2023
SN  - 0925-2312
DO  - 10.1016/j.neucom.2023.126637
UR  - https://www.sciencedirect.com/journal/neurocomputing
AB  - Conditional branch prediction allows the speculative fetching and execution of instructions before knowing the
direction of conditional statements. As in other areas, machine learning techniques are a promising approach
to building branch predictors, e.g., the Perceptron predictor. However, those traditional solutions demand large
input sizes, which impose a considerable area overhead. We propose a conditional branch predictor based on
the WiSARD (Wilkie, Stoneham, and Aleksander’s Recognition Device) weightless neural network model. The
WiSARD-based predictor implements one-shot online training designed to address branch prediction as a binary
classification problem. We compare the WiSARD-based predictor with two state-of-the-art predictors: TAGESC-
L (TAgged GEometric-Statistical Corrector-Loop) and the Multiperspective Perceptron. Our experimental
evaluation shows that our proposed predictor, with a smaller input size, outperforms the perceptron-based
predictor by about 0.09% and achieves similar accuracy to that of TAGE-SC-L. In addition, we perform an
experimental sensitivity analysis to find the best predictor for each dataset, and based on these results, we
designed new specialized predictors using a particular parameter composition for each dataset. The results show
that the specialized WiSARD-based predictor outperforms the state-of-the-art by more than 2.3% in the best
case. Furthermore, through the implementation of specialized predictor classifiers, we discovered that utilizing
90% of the specialized predictor for a specific dataset yielded comparable performance to the corresponding
specialized predictor.
ER  -