Metalanguage for high performance computing on hybrid architectures

Abstract

This research project proposes the specification of a metalanguage and a pre-compiler software to develop applications that require high performance computing in hybrid architectures. The metalanguage proposed consists of annotations which can be embedded in the source code of programs written in a particular high level programming language. Translated by the pre-compiler, the annotations embedded in the source code generate a new code that embeds calls for application programming interfaces (APIs) which uses MPI, OpenMP or Pthreads. Those APIs are frequently used in programs developed for high performance computing (HPC) architectures. The metalanguage proposed in this project will facilitate the development of algorithms and applications for high performance architectures using hybrid architectures, i. e., systems with shared memory and distributed memory, common in most current HPC systems. (AU)