April 25, 2006
Conference Paper

An Approach to Locality-Conscious Load Balancing and Transparent Memory Hierarchy Management with a Global-Address-Space Parallel Programming Model

Abstract

This paper describes a global-addressspace framework for the convenient specification and effi- cient execution of parallel out-of-core applications operating on block-sparse data. The programming model provides a global view of block-sparse matrices and a mechanism for the expression of parallel tasks that operate on blocksparse data. The tasks are automatically partitioned into phases that operate on memory-resident data, and mapped onto processors to optimize load balance and data locality. Experimental results are presented that demonstrate the utility of the approach.

Revised: August 15, 2007 | Published: April 25, 2006

Citation

Krishnamoorthy S., U. Catalyurek, J. Nieplocha, and P. Sadayappan. 2006. An Approach to Locality-Conscious Load Balancing and Transparent Memory Hierarchy Management with a Global-Address-Space Parallel Programming Model. In 20th International Parallel and Distributed Symposium (IPDPS'06), 25-29 April 2006,, 8. Piscataway, New Jersey:Institute of Electrical and Electronics Engineers. PNNL-SA-48518. doi:10.1109/IPDPS.2006.1639719