Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Computational Science, Engineering & Technology Series
PATTERNS FOR PARALLEL PROGRAMMING ON GPUS
Edited by: F. Magoulès
Using MELT to Improve or Explore your GCC-Compiled Source Code
CEA, LIST, Software Reliability Lab, Gif-sur-Yvette, France
B. Starynkevitch, "Using MELT to Improve or Explore your GCC-Compiled Source Code", in F. Magoulès, (Editor), "Patterns for Parallel Programming on GPUs", Saxe-Coburg Publications, Stirlingshire, UK, Chapter 7, pp 171-186, 2014. doi:10.4203/csets.34.7
Keywords: compilation, GCC, domain-specific language, high-performance computing, GPGPU, OpenCL, MELT, Gimple.
This paper introduces the MELT framework and domain-specific language to extend the GCC compiler. It explains the major internal representations (Gimple, Tree-s, etc.) and the overall organization of GCC. It shows the major features of MELT and illustrates why extending and customizing the GCC compiler using MELT is useful (for instance, to use GPGPUs through OPENCL). It gives some concrete advice and guidelines for development of such extensions with MELT.
purchase the full-text of this chapter (price £20)