izpis_h1_title_alt

Primerjava univerzalnih knjižnic za grafične pospeševalnike
ID Purgar, Vid (Author), ID Lotrič, Uroš (Mentor) More about this mentor... This link opens in a new window

.pdfPDF - Presentation file, Download (11,70 MB)
MD5: 63BBFF68F33699C96C8EC7310F7C2682

Abstract
V diplomskem delu obravnavamo univerzalne knjižnice za vzporedno programiranje v jezikih C in C++ s poudarkom na ogrodjih OpenCL in SYCL. Pristop OpenCL na nizki ravni ponuja natančen nadzor nad heterogenimi računskimi napravami, medtem ko abstrakcija SYCL omogoča pisanje programov za vzporedno računanje na višji ravni v jeziku C++. Ogrodji smo testirali na treh algoritmih, ki pokrijejo področja, za katera se najpogosteje uporabljajo grafične procesne enote, to so računanje zgoščenih vrednosti, obdelava slik ter učenje nevronskih mrež. Rezultati so pokazali, da je bila rešitev, napisana s pomočjo ogrodja OpenCL v vseh treh primerih hitrejša kot rešitev, napisana s pomočjo ogrodja SYCL, pri kateri pa je bila koda precej krajša in bolj pregledna.

Language:Slovenian
Keywords:grafični pospeševalniki, OpenCL, SYCL, vzporedno računanje
Work type:Bachelor thesis/paper
Typology:2.11 - Undergraduate Thesis
Organization:FRI - Faculty of Computer and Information Science
Year:2023
PID:20.500.12556/RUL-149231 This link opens in a new window
COBISS.SI-ID:164767491 This link opens in a new window
Publication date in RUL:05.09.2023
Views:850
Downloads:80
Metadata:XML DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:English
Title:Comparison of universal libraries for graphics accelerators
Abstract:
This paper is about universal libraries for parallel programming in C and C++, focusing on OpenCL and SYCL. The low-level approach of OpenCL offers precise control over heterogeneous devices, while the abstraction of SYCL allows for writing of higher-level parallel computing programs in C++. We tested the two frameworks on three algorithms covering the areas most commonly used by GPUs, which are computing hash values, image processing and training neural networks. The results showed that the algorithm written using the OpenCL framework outperformed the algorithm written using the SYCL framework for all three algorithms in terms of speed, but the SYCL framework required less code and had better readability.

Keywords:graphics accelerators, OpenCL, SYCL, parallel computing

Similar documents

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

Back