Umeå University's logo

umu.sePublikasjoner
Endre søk
RefereraExporteraLink to record
Permanent link

Direct link
Referera
Referensformat
  • apa
  • ieee
  • vancouver
  • Annet format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annet språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf
A test for FLOPs as a discriminant for linear algebra algorithms
IRTG-MIP, Rwth Aachen University, Aachen, Germany.
Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.ORCID-id: 0000-0002-4972-7097
2022 (engelsk)Inngår i: 2022 IEEE 34th International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD), IEEE, 2022, s. 221-230Konferansepaper, Publicerat paper (Fagfellevurdert)
Abstract [en]

Linear algebra expressions, which play a central role in countless scientific computations, are often computed via a sequence of calls to existing libraries of building blocks (such as those provided by BLAS and LAPACK). A sequence identifies a computing strategy, i.e., an algorithm, and normally for one linear algebra expression many alternative algorithms exist. Although mathematically equivalent, those algorithms might exhibit significant differences in terms of performance. Several high-level languages and tools for matrix computations such as Julia, Armadillo, Linnea, etc., make algorithmic choices by minimizing the number of Floating Point Operations (FLOPs). However, there can be several algorithms that share the same (or have nearly identical) number of FLOPs; in many cases, these algorithms exhibit execution times which are statistically equivalent and one could arbitrarily select one of them as the best algorithm. It is however not unlikely to find cases where the execution times are significantly different from one another (despite the FLOP count being almost the same). It is also possible that the algorithm that minimizes FLOPs is not the one that minimizes execution time. In this work, we develop a methodology to test the reliability of FLOPs as discriminant for linear algebra algorithms. Given a set of algorithms (for an instance of a linear algebra expression) as input, the methodology ranks them into performance classes; i.e., multiple algorithms are allowed to share the same rank. To this end, we measure the algorithms iteratively until the changes in the ranks converge to a value close to zero. FLOPs are a valid discriminant for an instance if all the algorithms with minimum FLOPs are assigned the best rank; otherwise, the instance is regarded as an anomaly, which can then be used in the investigation of the root cause of performance differences.

sted, utgiver, år, opplag, sider
IEEE, 2022. s. 221-230
Serie
Proceedings (Symposium on Computer Architecture and High Performance Computing), ISSN 1550-6533
Emneord [en]
Algorithm ranking, Linear algebra algorithms, Mathematical software performance, Performance Analysis
HSV kategori
Identifikatorer
URN: urn:nbn:se:umu:diva-203573DOI: 10.1109/SBAC-PAD55451.2022.00033ISI: 000905612800023Scopus ID: 2-s2.0-85145881711ISBN: 9781665451550 (digital)OAI: oai:DiVA.org:umu-203573DiVA, id: diva2:1728716
Konferanse
34th IEEE International Symposium on Computer Architecture and High Performance Computing, SBAC-PAD 2022, November 2-5, 2022
Tilgjengelig fra: 2023-01-19 Laget: 2023-01-19 Sist oppdatert: 2023-11-10bibliografisk kontrollert

Open Access i DiVA

Fulltekst mangler i DiVA

Andre lenker

Forlagets fulltekstScopus

Person

Bientinesi, Paolo

Søk i DiVA

Av forfatter/redaktør
Bientinesi, Paolo
Av organisasjonen

Søk utenfor DiVA

GoogleGoogle Scholar

doi
isbn
urn-nbn

Altmetric

doi
isbn
urn-nbn
Totalt: 195 treff
RefereraExporteraLink to record
Permanent link

Direct link
Referera
Referensformat
  • apa
  • ieee
  • vancouver
  • Annet format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annet språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf