Recursive Implementation of High Performance Numerical Algebra Library

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F06%3A00114523" target="_blank" >RIV/68407700:21230/06:00114523 - isvavai.cz</a>

Výsledek na webu

<a href="http://shimi.webzdarma.cz/vyzkum/workshop06/Recursive_workshop.doc" target="_blank" >http://shimi.webzdarma.cz/vyzkum/workshop06/Recursive_workshop.doc</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Recursive Implementation of High Performance Numerical Algebra Library

Popis výsledku v původním jazyce

For a good performance of every computer program, the efficient cache utilization is crucial. In numerical algebra libraries (such as BLAS or LAPACK) is the good cache utilization achieved by the explicit loop restructuring. It includes loop unrolling-and-jam which increase the FPU pipeline utilization in the innermost loop, loop blocking (that is why we called these codes shortly blocked) and loop interchange to maximize the a cache hit ratio. After application of these transformations, these codes aredivided into two parts. Outer loops are "out-cache", inner loops are "in-cache". Codes have almost the same performance independently on the amount of data, but all these code transformations require the difficult cache behavior analysis. In this paper,we represent the recursive implementation of some routines from the numerical algebra library. This implementation leads to cache-sensitive codes due to the "natural" partition of data without need to analyze the cache behavior.

Název v anglickém jazyce

Recursive Implementation of High Performance Numerical Algebra Library

Popis výsledku anglicky

For a good performance of every computer program, the efficient cache utilization is crucial. In numerical algebra libraries (such as BLAS or LAPACK) is the good cache utilization achieved by the explicit loop restructuring. It includes loop unrolling-and-jam which increase the FPU pipeline utilization in the innermost loop, loop blocking (that is why we called these codes shortly blocked) and loop interchange to maximize the a cache hit ratio. After application of these transformations, these codes aredivided into two parts. Outer loops are "out-cache", inner loops are "in-cache". Codes have almost the same performance independently on the amount of data, but all these code transformations require the difficult cache behavior analysis. In this paper,we represent the recursive implementation of some routines from the numerical algebra library. This implementation leads to cache-sensitive codes due to the "natural" partition of data without need to analyze the cache behavior.

Druh

D - Stať ve sborníku

CEP obor

IN - Informatika

OECD FORD obor

—

Projekt

<a href="/cs/project/IBS3086102" target="_blank" >IBS3086102: Paralelní algoritmy pro rozsáhlé simulace na svazcích (klastrech) PC</a><br>

Návaznosti

Z - Vyzkumny zamer (s odkazem do CEZ)

Rok uplatnění

2006

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Proceedings of Workshop 2006

ISBN

80-01-03439-9

ISSN

—

e-ISSN

—

Počet stran výsledku

2

Strana od-do

118-119

Název nakladatele

ČVUT

Místo vydání

Praha

Místo konání akce

Praha

Datum konání akce

20. 2. 2006

Typ akce podle státní příslušnosti

EUR - Evropská akce

Kód UT WoS článku

—