BDBComp
Parceria:
SBC
Analyzing Branch Prediction Under Speculative Execution Using Perceptron

Vinicius RibasMauricio FigueiredoRonaldo Gonçalves

Nowadays, most of commercial processors use aggressive techniques to explore high Instruction Level Parallelism (ILP). In this context, branch prediction is used to improve the fetch bandwidth, expanding the possibilities of finding parallel instructions. Most of branch predictors are based on Branch Target Buffer (BTB) tables, ensuring high accuracy in spite of spending hardware space and cycle time. Other alternatives have been investigated such as intelligent predictors. They present simplicity and functionality, reducing hardware cost implementation. The present work analyzes the effect of speculative execution on the prediction accuracy when using one perceptron neuron in the prediction engine. Clique no link abaixo para buscar o texto completo deste trabalho na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato: bdbcomp@lbd.dcc.ufmg.br
     Mantida por:
LBD