Dissertation/Thesis Abstract

Automatic Transcompilation of Affine C Programs to CUDA
by Krasniqi, Leart, M.E., University of Connecticut, 2020, 103; 28150976
Abstract (Summary)

Parallel computing is a rapidly growing field due to its extreme performance boosts when dealing with large amounts of data. General purpose computing on graphics processing units (GPGPU) allows programmers to utilize GPUs to exploit parallelism in CPU code. In this thesis, we present a system that automatically transcompiles source C code into CUDA code, which can be executed on a GPU. Unlike other similar systems, our transcompiler is a complete end-to-end system capable of handling certain while loops and imperfectly nested for loops. We tested our system on a variety of computationally expensive applications and achieved immense computational speedups and decent overall speedups.

Indexing (document details)
Advisor: Sable, Carl
Commitee:
School: University of Connecticut
Department: Electrical Engineering
School Location: United States -- Connecticut
Source: MAI 82/4(E), Masters Abstracts International
Source Type: DISSERTATION
Subjects: Electrical engineering, Computer Engineering, Computer science
Keywords: Automatic Translation, Compilers, GPGPU, Parallel Computing
Publication Number: 28150976
ISBN: 9798678184573
Copyright © 2020 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy
ProQuest