Transitive closure on the imagine stream processor [electronic resource].
- Washington, D.C. : United States. Department of Energy. Office of Advanced Scientific Computing Research, 2003.
Oak Ridge, Tenn. : Distributed by the Office of Scientific and Technical Information, U.S. Dept. of Energy.
- Physical Description:
- 8 pages : digital, PDF file
- Additional Creators:
- Lawrence Berkeley National Laboratory
United States. Department of Energy. Office of Advanced Scientific Computing Research
United States. Department of Energy. Office of Scientific and Technical Information
- The increasing gap between processor and memory speeds is a well-known problem in modern computer architecture. The Imagine system is designed to address the processor-memory gap through streaming technology. Stream processors are best-suited for computationally intensive applications characterized by high data parallelism and producer-consumer locality with minimal data dependencies. This work examines an efficient streaming implementation of the computationally intensive Transitive Closure (TC) algorithm on the Imagine platform. We develop a tiled TC algorithm specifically for the Imagine environment, which efficiently reuses streams to minimize expensive off-chip data transfers. The implementation requires complex stream programming since the memory hierarchy and cluster organization of the underlying architecture are exposed to the Imagine programmer. Results demonstrate that limited performance of TC is achieved primarily due to the complicated data-dependencies of the blocked algorithm. This work is an ongoing effort to identify classes of scientific problems well-suited for streaming processors.
- Published through SciTech Connect.
Fifth Workshop on Media and Stream Processors (MSP5), San Diego, CA (US), 12/01/2003.
Oliker, Leonid; Griem, Gorden.
- Funding Information:
View MARC record | catkey: 14104781