Parallel Paradigm Integration Primary Contact: Mani Chandy (mani@cs.caltech.edu)

Mani Chandy (Project Director), Ian Foster, Jonathon Geisler, Diane Goodfellow, Carl Kesselman, Joe Kiniry, David Kohr, Rakesh Krishnaiyer, Berna Massingill, John Reynders, Matt Richardson, Adam Rifkin, Eve Schooler, Paul Sivilotti, Mei Su, Wesley Tanaka, John Thornley, Steve Tuecke, and Dan Zimmerman.

Abstract. The Parallel Paradigm Integration project focuses on three topics: distributed heterogeneous computing, paradigm integration, and problem-solving environments (PSEs). Distributed computing research supports the reservation and acquisition of resources on networks of computing resources from desktop to supercomputers. The Globus/Nexus tools are examples of the software that we are developing. We are also developing an infrastructure for an object network that allows persistent active objects to communicate. One focus is asynchronous collaboration that allows scientists to participate in archived scientific collaborations.

Paradigm integration research integrates different paradigms for parallel programming, such as data parallelism, task (or functional) parallelism, and object-based parallelism, within a uniform and practical framework and uses the technology in metacomputers. Focus areas inlcude the use of object technology and languages such as C++ and Java for parallel applications. HPC++, CC++, and verified Java libraries are examples of our software development. Research into problem-solving environments (PSEs) develops an integrated environment for problem solving that makes computers in general, and parallel computers in particular, truly usable. Focus areas include domain-specific parallel PSEs such as an airshed model for Southern California, the integration of symbolic and numeric computing, and the use of PSEs in education from K-12 to post-university.  

Corporate Sponsors for this research area are currently not available.

• Globus - A metacomputing infrastructure for wide area computation. (Publications)

• Nexus - A portable library providing the multithreading, communication, and resource management facilities required to implement advanced languages, libraries, and applications in heterogeneous parallel and distributed computing environments. (Publications)

• Infospheres - The theory and implementation of compositional systems that support peer-to-peer communication among persistent multithreaded distributed objects. (Publications)

Paradigm Integration

• CC++ - A parallel programming language based on C++ that integrates multithreaded and message-passing programming paradigms. (Publications)

• High Performance C++ (HPC++) - A widely accepted approach to data-parallel C++. (Publications)

• Fortran M - A set of extensions to Fortran that supports a modular approach to the design of task parallel, message-passing programs. (Publications)

• Parallel Object-Oriented Methods and Applications (POOMA) - A framework for scientific computing applications on parallel computers. (Publications)

• Structured multithreaded programming for shared-memory multiprocessors - A threads library that executes on top of Windows NT, Posix, and Solaris thread libraries that supports large-scale applications; and a translator for C that handles pragmas to indicate parallel execution of loops with dependencies across iterations.

Problem Solving Environments (PSEs)

• Virtual collaboratorium and MetaPSE - Systems that support collaboration between scientists at remote locations.

• A parallel problem-solving environment for air quality modeling. (Related Slides)