izpis_h1_title_alt

Indeksna struktura za učinkovito vzporedno iskanje in vstavljanje točk v večdimenzionalnem prostoru : magistrsko delo
ID Rojc, Blaž (Author), ID Cabello, Sergio (Mentor) More about this mentor... This link opens in a new window, ID Depolli, Matjaž (Co-mentor)

.pdfPDF - Presentation file, Download (10,55 MB)
MD5: 0A9907010DEA4C28B4B29466FA107F43

Abstract
Za učinkovito delo s točkami v prostoru potrebujemo primerno podatkovno in indeksno strukturo, na katero se lahko zanašamo pri razvoju algoritmov. Na voljo imamo obsežen nabor takšnih struktur, a v večini nam ne omogočajo vstavljati novih točk v kontekstu več niti izvajanja brez dodatnih varovalnih mehanizmov. Takšne strukture omejijo izvajanje kritičnih delov programa eni niti naenkrat, kar nam prepreči učinkovito izrabo vseh računskih sredstev, ki so nam na voljo. Potrebujemo torej indeksno strukturo, ki nam omogoča tako vzporedno iskati kot tudi vzporedno vstavljati točke brez zunanjih varovalnih mehanizmov, ki omejujejo hitrost izvajanja. Razvili smo indeksno strukturo na podlagi štiriškega drevesa, ki kritične dele programa omeji na liste drevesa, kar zmanjša verjetnost trka dveh niti in posledično omogoča veliko hitrejše izvajanje algoritma. Drevo smo testirali na vzporednem algoritmu za diskretizacijo domene in pokazali, da pri uporabi velikega števila niti izvajanja uporaba naše indeksne strukture vodi do veliko hitrejšega izvajanja algoritma kot prejšnja rešitev.

Language:Slovenian
Keywords:iskalno drevo, vzporedno računanje, diskretizacija domene
Work type:Master's thesis/paper
Organization:FMF - Faculty of Mathematics and Physics
Year:2022
PID:20.500.12556/RUL-140056 This link opens in a new window
COBISS.SI-ID:120760323 This link opens in a new window
Publication date in RUL:10.09.2022
Views:605
Downloads:37
Metadata:XML RDF-CHPDL DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:English
Title:An indexing structure for efficient parallel point lookups and insertions in a multidimensional space
Abstract:
For efficient work with points in space we require a suitable data and indexing structure, upon which we are able to rely on during the development of algorithms. We have a wide selection of such structures to choose from, but the majority of them do not allow us to insert new points in a multithreaded context without the use of auxiliary protection mechanisms. Such structures limit the execution of the critical parts of programs to one thread at a time, which prevents us from efficiently utilizing all the available computing power at our disposal. We therefore require an indexing structure that allows us to lookup points as well as insert new points in parallel without auxiliary protection mechanisms that limit execution speed. We developed a quadtree-based indexing structure which limits the critical parts of the program to the leaves of the tree, which in turn lowers the probability of a thread collision and subsequently allows for significantly faster execution of the algorithm. We tested the tree using the parallel domain discretization algorithm and showed that when using many threads of execution the use of our indexing structure leads to significantly faster execution of the algorithm compared to the previous solution.

Keywords:indexing tree, parallel computing, domain discretization

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back