Practical Parallel Programming provides scientists and engineers with a detailed, informative, and often critical introduction to parallel programming techniques. Following a review of the fundamentals of parallel computer theory and architecture, it describes four of the most popular parallel programming models in use today - data parallelism, shared variables, message passing, and Linda - and shows how each can be used to solve various scientific and numerical problems. Examples, coded in various dialects of Fortran, are drawn from such domains as the solution of partial differential equations, the solution of linear equations, the simulation of cellular automata, studies of rock fracturing, and image processing. Practical Parallel Programming will be particularly helpful for scientists and engineers who use high-performance computers to solve numerical problems and do physical simulations but who have little experience with networking or concurrency. The book can also be used by advanced undergraduate and graduate students in computer science in conjunction with sources covering parallel architectures and algorithms in more detail. Computer science students will gain a critical appraisal of the current state of the art in parallel programming. The language used in the book, Fortran-K, is a subset of Fortran-90. It is described in an appendix and a compiler, for it is freely available from the MIT Press ftp site.