Title
Scalable parallelization of FLAME code via the workqueuing model
Abstract
We discuss the OpenMP parallelization of linear algebra algorithms that are coded using the Formal Linear Algebra Methods Environment (FLAME) API. This API expresses algorithms at a higher level of abstraction, avoids the use loop and array indices, and represents these algorithms as they are formally derived and presented. We report on two implementations of the workqueuing model, neither of which requires the use of explicit indices to specify parallelism. The first implementation uses the experimental taskq pragma, which may influence the adoption of a similar construct into OpenMP 3.0. The second workqueuing implementation is domain-specific to FLAME but allows us to illustrate the benefits of sorting tasks according to their computational cost prior to parallel execution. In addition, we discuss how scalable parallelization of dense linear algebra algorithms via OpenMP will require a two-dimensional partitioning of operands much like a 2D data distribution is needed on distributed memory architectures. We illustrate the issues and solutions by discussing the parallelization of the symmetric rank-k update and report impressive performance on an SGI system with 14 Itanium2 processors.
Year
DOI
Venue
2008
10.1145/1326548.1326552
ACM Trans. Math. Softw.
Keywords
Field
DocType
parallel,dense linear algebra algorithm,additional key words and phrases: flame,scalability,smp,formal linear algebra methods,workqueuing implementation,openmp parallelization,openmp,flame code,itanium2 processor,workqueuing,sgi system,use loop,workqueuing model,scalable parallelization,linear algebra algorithm,flame,linear algebra
Linear algebra,Shared memory,Computer science,Parallel algorithm,Parallel computing,Distributed memory,Theoretical computer science,Sorting,Multiprocessing,Memory architecture,Automatic parallelization
Journal
Volume
Issue
ISSN
34
2
0098-3500
Citations 
PageRank 
References 
10
1.97
9
Authors
4
Name
Order
Citations
PageRank
Field G. Van Zee131223.19
Paolo Bientinesi244853.91
Tze Meng Low314619.62
Robert A. van de Geijn42047203.08