Ubrzavanje rješavanja kvantno-transportnih jednadžbi u heterogenom računskom okruženju

master's thesis

• Title (english): Accelerating the quantum transport solver in a heterogenous computational environment
• Author: Sandra Kuzmić
• Mentor: Mirko Poljak (mentor)
• Committee member: Mirko Poljak (predsjednik povjerenstva)
• Committee member: Marko Koričić (član povjerenstva)
• Committee member: Vladimir Čeperić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-02-18, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom radu dan je pregled platformi, numeričkih biblioteka i jezika za implementaciju numeričkih simulatora u paralelnim računskim okruženjima. Implementirano je rješavanje retardirane Greenove funkcije i transmisijske funkcije u jeziku C korištenjem numeričkih biblioteka otvorenog koda BLAS i LAPACK u okruženju s dijeljenom memorijom gdje je paralelizacija implementirana OpenMP sučeljem. Također su analizirane performanse kvantno-transportnog simulatora za nanovrpce 2D materijala implementiranog na platformi CUDA korištenjem biblioteka cuBLAS i cuSOLVER. Uspoređena su vremena izvođenja za različit broj jezgri, različite formate matrica te za različite dimenzije nanostrukture. Pokazano je da izbor optimalne implementacije simulatora nije trivijalan i da ovisi o dimenzijama nanostrukture.

Abstract (english)

This thesis gives an overview of platforms, numerical libraries and programming languages for implementing numerical simulators in parallel computational environments. The calculation of the retarded Green's function and transmission function is implemented in C language using open source libraries BLAS and LAPACK in a shared memory environment in which the parallelization is implemented using the OpenMP interface. The performance of a quantum transport simulator for 2D material nanoribbons implemented in the CUDA platform using cuBLAS and cuSOLVER libraries is also analyzed. Execution times for different numbers of cores, different matrix formats and for different dimensions of the nanostructure are compared. It is shown that the choice of the optimal simulator implementation is not trivial and that it depends on the dimensions of the nanostructure.

• Keywords: kvantni transport; nanostruktura; računska znanost; paralelizacija; LAPACK; BLAS; OpenMP; CUDA; cuBLAS; cuSOLVER
• Keywords (english): quantum transport; nanostructure; computational science; parallelization; LAPACK; BLAS; OpenMP; CUDA; cuBLAS; cuSOLVER
• Language: croatian
• URN:NBN: urn:nbn:hr:168:243574
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 15:22:28