Early errors detection in parallel programs

Abstract

The goal of this paper is to tell about the methodology and tools for errors detection in parallel programs at the code writing stage. Applying static code analysis methodology allows developers to significantly reduce the error correction costs at the testing and support stages. The error diagnostics in multithread applications will be demonstrated with the examples of PC-Lint, VivaMP, and Intel C++ “Parallel Lint” analyzers. The paper will be useful for developers who create parallel Windows applications in C/C++ languages.