Language Extensions and Compilation Techniques for Data Intensive Computations

Processing and analyzing large volumes of data plays an increasingly important role in many domains of scienti c research. Typical examples of very large scienti c datasets include long running simulations of time-dependent phenomena that periodically generate snapshots of their state, archives of raw and processed remote sensing data, and archives of medical images. High-level language and compiler support for developing applications that analyze and process such datasets has, however, been lacking so far. We are developing language extensions and a compilation framework for expressing the applications that process large multidimensional datasets in a high-level data-parallel fashion. We have chosen a dialect of Java for expressing these applications. Our dialect of Java includes data-parallel extensions for specifying collection of objects, a parallel for loop, and reduction variables. Our compiler will analyze parallel loops and optimize the processing of datasets through the use of an existing runtime system, called Active Data Repository (ADR), developed at University of Maryland. We present design of a compiler/runtime interface which allows the compiler to e ectively utilize the existing runtime system. We show how interprocedural static program slicing can be used by the compiler to extract relevant information for the runtime system. Implementation of these compiler techniques is currently underway using the Titanium infrastructure.