LLVM-ACT: uma ferramenta baseada em profiling para seleção de técnica de computação aproximada

• Main Author: Miranda, Lavinia Medeiros
• Publication Date: 2022
• Format: Master thesis
• Language: por
• Source: Repositório Institucional da UFRN
• dARK ID: ark:/41046/001300000t5x2
• Download full: https://repositorio.ufrn.br/handle/123456789/49704
• topic: Computação aproximada; Profiling; Características da aplicação; LLVM; Transformação de código; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::SISTEMAS DE COMPUTACAO

Approximate Computing is currently an emerging paradigm that seeks to replace some data accuracy with aspects such as performance and energy efficiency. There are tools within this scope that apply some approximate computation techniques at software computational level. However, these tools are limited in a way that they only cover some specific scope, apply only one of the known techniques and/or need manual applications annotations to work out. The current state of the art still has questions such as the existence or not of a relation between the characteristics of the studied application and the chosen technique for its exploration; also remains open what would be the most appropriate technique for each particular context. Thus, this dissertation proposes the implementation of a tool that, according to the application profiling, chooses the most appropriate approximate computing technique to be applied. The tool uses the LLVM compilation infrastructure, where each step is implemented as a code analysis or transformation LLVM Pass. In addition to the Profiler, three approximate computation techniques were implemented, and the results obtained show that the technique chosen by the tool is cost-effective if the parameters of error rates and speedup are considered.