A Methodology for Optimization of Interpreters

Autor(es)

do Rosario, Vanderson Martins ; Hato, Mario Mikio ; Azevedo, Rodolfo ; Borin, Edson ; Bianchini, CD ; DeSouza, PSL ; DeBarros, COF ; Ferreira, RAC

• Tipo de documento: Artigo Científico
• Fonte: 2018 SYMPOSIUM ON HIGH PERFORMANCE COMPUTING SYSTEMS (WSCAD 2018); v. N/A, p. 8-pg., 2018-01-01.

Resumo

Interpretation is a simple and portable technique that enables emulation of instruction set architectures (ISAs). Even though other techniques, such as binary translation, still provide superior emulation performance, interpreters are easier to implement and debug. Because of that, they are often used to complement binary translation techniques or to design reference instruction set simulators (ISS). In this work, we characterize the main causes of overhead in an interpreter and present a methodology to analyze and optimize interpreters code. The methodology was designed having in mind the most common practices in the development of interpreters, the operation of modern processors architectures and memory hierarchy. Finally, to evidence the importance of the methodology we present a case study in which we used it to optimize an interpreter-based ISS, improving its performance by 48%. (AU)

• Processo FAPESP: 13/08293-7 - CECC - Centro de Engenharia e Ciências Computacionais
• Beneficiário: Munir Salomao Skaf
• Modalidade de apoio: Auxílio à Pesquisa - Centros de Pesquisa, Inovação e Difusão - CEPIDs