Management of Deep Memory Hierarchies – Recursive Blocked Algorithms and Hybrid Data Structures for Dense Matrix Computations

  • Bo Kågström
Conference paper

DOI: 10.1007/11558958_3

Part of the Lecture Notes in Computer Science book series (LNCS, volume 3732)
Cite this paper as:
Kågström B. (2006) Management of Deep Memory Hierarchies – Recursive Blocked Algorithms and Hybrid Data Structures for Dense Matrix Computations. In: Dongarra J., Madsen K., Waśniewski J. (eds) Applied Parallel Computing. State of the Art in Scientific Computing. PARA 2004. Lecture Notes in Computer Science, vol 3732. Springer, Berlin, Heidelberg

Abstract

Recent progress in using recursion as a general technique for producing dense linear algebra library software for today’s memory tiered computer systems is presented. To allow for efficient utilization of a memory hierarchy, our approach is to apply the technique of hierarchical blocking. The success of our approach includes novel recursive blocked algorithms, hybrid data formats and superscalar kernels.

Keywords

recursion automatic variable blocking superscalar GEMM-based level 3 BLAS hybrid data structures superscalar kernels SMP parallelization library software LAPACK SLICOT ESSL RECSY dense linear algebra factorizations matrix equations periodic systems 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • Bo Kågström
    • 1
  1. 1.Department of Computing Science and HPC2NUmeå UniversityUmeåSweden

Personalised recommendations