A Parallel-Object Programming Model for Petaflops Machines and Blue Gene/Cyclops
Workshop on Next Generation Systems at IPDPS (NSFNGS) 2002
Publication Type: Paper
One approach for building the next generation of parallel computers is based on large aggregates of multiprocessor chips with support for hardware multithreading. An initial design for IBM's Blue Gene/C project exemplifies this approach. Such a machine might consist of a million processors, and is characterized by a low memory-to-processor ratio. To study alternate programming models for such a machine before it is built, we have developed an emulator that allows million-processor programs to be run on conventional parallel machines with hundreds of processors. Here we present the implementation of a parallel object model based on Charm++ as a candidate programming model. Although the ``ideal'' programming model for such machines is a matter of continuing research, we believe that parallel objects represent a good starting point. This paper reviews the target architecture, presents the programming model, and describes the emulator implementation. Case studies of simple applications written using the emulator are also discussed.
Gengbin Zheng and Arun Kumar Singla and Joshua Mostkoff Unger and Laxmikant V. Kale, "A Parallel-Object Programming Model for Petaflops Machines and Blue Gene/Cyclops", NSF Next Generation Systems Program Workshop, 16th International Parallel and Distributed Processing Symposium(IPDPS), Fort Lauderdale, FL, April 2002.