Data-parallel Computing: The Language DimensionUnlike process concurrency, parallelism of data is a language issue. In this important area of parallel computing, using an appropriate programming paradigm with clear semantics and a deliberate structure may significantly alter the software design methodology. In this book, the authors address the language aspect of data-parallel computing by critically evaluating several well-known languages: APL, Fortran-90 and HPF. They also introduce new languages: EVAL and F-code. The many design highlights, examples, and especially the treatment of all languages from a single point of view makes this book invaluable to software developers. It should also appeal to researchers, academics and students in the field of parallel computing. The book comes with some free software, in particular a full implementation of EVAL. Readers will find this in the authors' archive at ftp: //www.ee.surrey.ac.uk/pub/DPar on the Internet. |
Common terms and phrases
access type alignment APL2 arbitrary argument array assignment axes axis character coercion compiler complex composing computed contains contents context position coordinate coordinate vector corresponding created data-parallel defined delivering the result denote descriptor DESTINATION.r determined dictionary dimension dot product dyadic E-code elementwise EVAL evaluation example execution EXPR EXPR.s expression extents F-code Fortran 90 function hardware hieroglyph High Performance Fortran identifier is associated implementation indices input integer scalar value integer type juxtaposed language layer Line logical logical matrix loop mask matrix memory module monadic multi-index node nonscalar objects operand overloading Parallel Computation parameters pointer posteriori integer posteriori type primitive objects priori processor programming language progressor rank reduction reference replication respect selected semantics sequence shape slice sort specified stack static stratification subrange subroutine subtyping syntax target transform translational symmetry transput trimmer type signature type system variable vector yields