Treffer: Management of Deep Memory Hierarchies - Recursive Blocked Algorithms and Hybrid Data Structures for Dense Matrix Computations.
Title:
Management of Deep Memory Hierarchies - Recursive Blocked Algorithms and Hybrid Data Structures for Dense Matrix Computations.
Authors:
Source:
Applied Parallel Computing. State of the Art in Scientific Computing. 2006, p21-32. 12p.
Database:
Supplemental Index
Weitere Informationen
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. [ABSTRACT FROM AUTHOR]