Performance Analysis of Modern Parallel Programming Models for GPUs: OpenMP, OpenACC, CUDA

Student: Khalilov Mikhail

Faculty: HSE Tikhonov Moscow Institute of Electronics and Mathematics (MIEM HSE)

Educational Programme: Mathematical Methods of Modelling and Computer Technologies (Master)

Final Grade: 10

Year of Graduation: 2020

Modern supercomputers use graphics processors as accelerators in compute nodes. GPUs enable scientific applications to significantly increase performance using fine-grained parallelism. The CUDA programming model focuses on optimized use of the SIMT architecture by writing low-level C/C++ code. At the same time, OpenACC and OpenMP 4.5 are part of the family of declarative parallel programming models. This study compares the effectiveness of the CUDA, OpenACC, and OpenMP programming models for various tests. The analysis of the GPU memory performance, the performance of physical tasks and various matrix multiplication implementations on Nvidia Tesla V100 and MX940 GPUs and modern processors is presented.

Full text (added May 17, 2020)

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses