Evaluación de parámetros de optimización GCC

Compile-time optimization of code can result in significant performance gains. The amount of these gains varies widely depending upon the code being optimized, the hardware being compiled for, the specific performance increase attempted (e.g. speed, throughput, memory utilization, etc.) and the used compiler. We used the latest version of the SPEC CPU 2006 benchmark suite to help gain an understanding of possible performance improvements using GCC (GNU Compiler Collection) options focusing mainly on speed gains made possible by tuning the compiler with the standard compiler optimization levels as well as a specific compiler option for the hardware processor. We compared the best standardized tuning options obtained for a core i7 processor, to the same relative options used on a Pentium4 to determine whether the GNU project has improved its performance tuning capabilities for specific hardware over time.

• Autores: Escobar, Rodrigo D.; Angula, Alekya R; Corsi, Mark
• Fecha de publicación: 2012
• Institución: Universidad de San Buenaventura
• Repositorio: Repositorio USB
• Idioma: spa
• OAI Identifier: oai:bibliotecadigital.usb.edu.co:10819/6404
• Acceso en línea: http://hdl.handle.net/10819/6404
• Palabra clave: Compiler optimization; Machine Learning; Compiler Heuristics; Programming Languages; Processors; Aprendizaje automático; Heurística de compiladores; Procesadores; Optimización de compilador; Heurística; Lenguajes de programación
• License: Atribución-NoComercial-SinDerivadas 2.5 Colombia