Scalable Programming Environment

Abstract : This report describes the Scalable Programming Environment (SPE), which provides programmers with a transparent way of creating scalable parallel applications for large-grained parallel computer architectures. The SPE, which has been designed primarily to support data-flow processing applications, allows programs to be scaled to execute on any number of processing nodes while requiring no changes to the compiled binary code. The user is provided with a set of high-level message-passing routines which can be used to connect multi- instanced heterogeneous programs in a system. The SPE library routines hide the intricacies of how the parallel programs communicate. The details of the connections are specified in text files. The SPE allows individual programs to be coded without knowledge of other parts of the system and thus allows systems to be quickly built, modified, or scaled without program recompilation. Parallel dataflow, parallel processing, Parallel message passing, Scalable, Parallel programming tools, Scalable programming environment.