Um estudo sobre a eficiência dos compiladores da linguagem GO com o auxilio de algoritmos genéticos

Abstract

The Go language is one of the newer current languages with an increasing popularity in the software industry. It is defined by a specification and implemented by two compilers with different proposals to ensure that the specification is correct and complete. The gc compiler focuses on quick compilation and few optimizations while gccgo focuses on using GCC optimizations without worrying about the compiling time that is spent. This work proposes a study with the objective of constructing a genetic algorithm that assists in the identification of situations that gccgo can be more efficient than the Go Compiler. Results show that the Go Compiler is on average 24.7 times more performance at runtime and with an average executable file size of 33.86 times greater than the generated gccgo executable using a subset of optimization options.