OpenMP issues arising in the development of parallel BLAS and LAPACK libraries

Issue title: OpenMP

Article type: Research Article

Authors: Addison, C. | Ren, Y. | van Waveren, M.

Affiliations: Fujitsu European Centre for Information Technology, Hayes, UK | Department of Computer Science, University of Manchester, Manchester | Fujitsu Systems Europe, Toulouse, France

Note: [] Corresponding author: 66 Queens Avenue, Meols, Wirral, CH47 0NA, UK. Tel: +44 151 632 6615; E-mail: caddison@addis0.fsnet.co.uk

Abstract: Dense linear algebra libraries need to cope efficiently with a range of input problem sizes and shapes. Inherently this means that parallel implementations have to exploit parallelism wherever it is present. While OpenMP allows relatively fine grain parallelism to be exploited in a shared memory environment it currently lacks features to make it easy to partition computation over multiple array indices or to overlap sequential and parallel computations. The inherent flexible nature of shared memory paradigms such as OpenMP poses other difficulties when it becomes necessary to optimise performance across successive parallel library calls. Notions borrowed from distributed memory paradigms, such as explicit data distributions help address some of these problems, but the focus on data rather than work distribution appears misplaced in an SMP context.

Journal: Scientific Programming, vol. 11, no. 2, pp. 95-104, 2003