Acta Polytechnica
Memory Hierarchy Behavior Study during the Execution of Recursive
Linear Algebra Library
Author: Ivan Šimeček
Keywords
numerical linear algebra, code restructuring, loop unrolling,
recursive implementation, memory hierarchy utilization.
Abstract
For good performance of every computer program, good cache and TLB utilization is crucial. In numerical linear algebra libraries (such as
BLAS or LAPACK), good cache utilization is achieved by explicit loop restructuring (mainly loop blocking), but this requires diącult memory
pattern behavior analysis. In this paper, we represent the recursive implementation ("divide and conquer" approach) of some routines from
numerical algebra libraries. This implementation leads to good cache and TLB utilization with no need to analyze the memory pattern behavior due to "natural" partition of data.
Download:
final version (in .PDF format)
BibTex entry:
@ARTICLE{JA_ACTA4,
author = {Ivan \v{S}ime\v{c}ek},
title = {Memory Hierarchy Behavior Study during the Execution of Recursive
Linear Algebra Library},
journal = {Acta Polytechnica},
year = {2009},
volume = {49},
pages = {29--36},
number = {5/2008},
issn = {1210-2709}
}