prikaz prve stranice dokumenta Ubrzavanje rješavanja 2D Poissonove jednadžbe u hibridnom CPU-GPU računskom okruženju
No public access
master's thesis
Ubrzavanje rješavanja 2D Poissonove jednadžbe u hibridnom CPU-GPU računskom okruženju
Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. urn:nbn:hr:168:150712
Glavan, Marko
University of Zagreb
Faculty of Electrical Engineering and Computing

Institutional repository: FER Repository

Cite this document

Glavan, M. (2019). Ubrzavanje rješavanja 2D Poissonove jednadžbe u hibridnom CPU-GPU računskom okruženju (Master's thesis). Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing. Retrieved from https://urn.nsk.hr/urn:nbn:hr:168:150712

Glavan, Marko. "Ubrzavanje rješavanja 2D Poissonove jednadžbe u hibridnom CPU-GPU računskom okruženju." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. https://urn.nsk.hr/urn:nbn:hr:168:150712

Glavan, Marko. "Ubrzavanje rješavanja 2D Poissonove jednadžbe u hibridnom CPU-GPU računskom okruženju." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. https://urn.nsk.hr/urn:nbn:hr:168:150712

Glavan, M. (2019). 'Ubrzavanje rješavanja 2D Poissonove jednadžbe u hibridnom CPU-GPU računskom okruženju', Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, accessed 25 July 2024, https://urn.nsk.hr/urn:nbn:hr:168:150712

Glavan M. Ubrzavanje rješavanja 2D Poissonove jednadžbe u hibridnom CPU-GPU računskom okruženju [Master's thesis]. Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing; 2019 [cited 2024 July 25] Available at: https://urn.nsk.hr/urn:nbn:hr:168:150712

M. Glavan, "Ubrzavanje rješavanja 2D Poissonove jednadžbe u hibridnom CPU-GPU računskom okruženju", Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, 2019. Available at: https://urn.nsk.hr/urn:nbn:hr:168:150712

Metadata
TitleUbrzavanje rješavanja 2D Poissonove jednadžbe u hibridnom CPU-GPU računskom okruženju
Title (english)Accelerating the 2D Poisson equation solver in a hybrid CPU-GPU computational environment
AuthorMarko Glavan
MentorMirko Poljak (mentor)
Committee memberMirko Poljak (predsjednik povjerenstva)
Committee memberMarko Koričić (član povjerenstva)
Committee memberVladimir Čeperić (član povjerenstva)
GranterUniversity of Zagreb
Faculty of Electrical Engineering and Computing
Zagreb
Defense date and country2019-07-19, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Electrical Engineering
Abstract
U ovom radu je napravljen pregled metoda za ubrzavanje računanja 2D Poissonove jednadžbe u CPU, GPU te hibridnom CPU-GPU okruženju, te su napravljene implementacije i uspoređena trajanja izvođenja u MATLAB-u, korištenjem numeričkih biblioteka BLAS i LAPACK u C-u, te na GPU-u korištenjem C CUDA-e. Analiza je provedena na silicijskom MOSFET-u s dvije upravljačke elektrode. Poissonova jednadžba je riješena metodom konačnih razlika sa koracima diskretizacije veličine svega 0,1 nm, što je u slučaju najvećeg tranzistora dalo mrežu od 27300 točaka. Utvrđeno je da MATLAB nije pogodan za rješavanje problema ove veličine. CUDA sparse implementacija postiže ubrzanje od 27 puta u usporedbi sa MATLAB dense implementacijom, odnosno 21 puta u usporedbi sa MATLAB sparse implementacijom.
Abstract (english)
This thesis presents a review of the methods of accelerating the computation of the 2D Poisson equation in the CPU, GPU and the hybrid CPU-GPU environment. Implementation and comparison were performed in MATLAB, in C using the BLAS and LAPACK numeric libraries and at the GPU using C CUDA. The analysis was done on double-gate silicon MOSFET. Poisson solution was obtained using a 2D finite difference method with discretized steps of only 0.1 nm, which in the case of the largest transistor formed a grid of 27300 points. It was found that MATLAB is not suitable for solving the problem of this size. CUDA sparse implementation achieved acceleration of 27 times compared to MATLAB dense implementation, respectively to 21 times compared to MATLAB sparse implementation.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:168:150712
Study programmeTitle: Electrical Engineering and Information Technology
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije (magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije)
Type of resourceText
File originBorn digital
Access conditionsClosed access
Terms of useLicence In copyright icon
Public note
Created on2020-01-07 18:06:23