kth.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
A SIMD Approach to the Convex Hull Problem: A Vectorized Implementation of Jarvis March
KTH, School of Electrical Engineering and Computer Science (EECS).
KTH, School of Electrical Engineering and Computer Science (EECS).
2024 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesisAlternative title
En SIMD-ansats till konvexa höljen : En vektoriserad implementering av Jarvis March (Swedish)
Abstract [en]

This thesis introduces a SIMD approach to the convex hull algorithm Jarvis March, also known as the Gift Wrapping algorithm. A vectorized version of the algorithm is presented, and the study compares its performance against scalar versions to assess its effectiveness. Experimental results show a speedup factor of approximately 3.5 in the vectorized portion of the algorithm using AVX-512 instructions, contributing to an overall speedup factor of 1.18 compared to its scalar execution. However, contrary to expectations, the experimental results demonstrated that the vectorized implementation exhibited slower performance compared to the commonly implemented scalar version, achieving only half the speed. A roofline model analysis was employed to identify potential bottlenecks, pinpointing memory bandwidth as the primary constraint in the vectorized implementation. Consequently, the vectorized version of the Jarvis March algorithm presented is not recommended unless sufficient memory bandwidth or further improvements to the algorithm can be attained.

Abstract [sv]

Denna avhandling introducerar en SIMD-ansats till den konvexa höljalgoritmen Jarvis March. En vektoriserad version av algoritmen presenteras, och studien jämför algoritmens prestanda mot skalära versioner för att bedöma dess effektivitet. Resultaten visar en hastighetsökning med en faktor 3,5 i den vektoriserbara delen av algoritmen vid anvädning av AVX-512-instruktioner, vilket bidrog till en total hastighetsökning med en faktor 1,18 jämfört med den skalära exekveringen. Emellertid, i strid med förväntningarna, visar de experimentella resultaten att den vektoriserade implementationen uppvisade långsammare prestanda jämfört med den vanligtvis implementerade skalära versionen och uppnådde endast hälften av hastigheten. Roofline-modellen användes för att identifiera potentiella flaskhalsar och pekar ut minnesbandbredden som den primära begränsningen i den vektoriserade implementationen. Följaktligen rekommenderas inte vektorisering av Jarvis March-algoritmen om inte tillräcklig minnesbandbredd eller vidare förbättringar av algoritmen kan erhållas.

Place, publisher, year, edition, pages
2024. , p. 26
Series
TRITA-EECS-EX ; 2024:335
National Category
Computer and Information Sciences
Identifiers
URN: urn:nbn:se:kth:diva-351044OAI: oai:DiVA.org:kth-351044DiVA, id: diva2:1885908
Supervisors
Examiners
Available from: 2024-08-22 Created: 2024-07-26 Last updated: 2024-08-22Bibliographically approved

Open Access in DiVA

fulltext(676 kB)141 downloads
File information
File name FULLTEXT01.pdfFile size 676 kBChecksum SHA-512
a7847c103f62dad4bf23c3236b8213bfc31f8f750ee22cc8b9443a107648214bc09cea69f8f3703739fc85d695de1df873076b982b424e1cf8ada6df1c6e999d
Type fulltextMimetype application/pdf

By organisation
School of Electrical Engineering and Computer Science (EECS)
Computer and Information Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 141 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 265 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf