Highly parallel GPU accelerator for HEVC transform and quantization

Znanstveni rad - Izvorni znanstveni rad

• Autor: Mate Čobrnić https://orcid.org/0000-0002-1422-0392
• Autor: Alen Duspara
• Autor: Leon Dragić
• Autor: Igor Piljić https://orcid.org/0000-0003-2345-0322
• Autor: Mario Kovač
• Urednik: Su Ruidan
• Autorova ustanova: Sveučilište u Zagrebu; Fakultet elektrotehnike i računarstva; (Zavod za automatiku i računalno inženjerstvo)
• Znanstveno / umjetničko područje, polje i grana: TEHNIČKE ZNANOSTI; Računarstvo; Arhitektura računalnih sustava

Sažetak (engleski)

When analysing Internet traffic today it can be found that digital video content prevails. Its domination will continue to
grow in the upcoming years and reach 82% of all traffic by 2021. If converted to Internet video minutes per second, this
equals about one million video minutes per second. Providing and supporting improved compression capability is therefore
expected from video processing devices. This will relieve the pressure on storage systems and communication networks
while creating preconditions for further development of video services. Transform and quantization is one of the most
compute-intensive parts of modern hybrid video coding systems where coding algorithm itself is commonly standardized.
High Efficiency Video Coding (HEVC) is state-of-the-art video coding standard which achieves high compression
efficiency at the cost of high computational complexity. In this paper we present highly parallel GPU accelerator for HEVC
transform and quantization which targets most common heterogeneous computing CPU+GPU system. The accelerator is
implemented using CUDA programming model. All the relevant state-of-the-art techniques related to kernel vectorization,
shared memory optimization and overlapping data transfers with computation were investigated, customized and carefully
combined to obtain a performance efficient solution across all applicable transform sizes. The proposed solution is
compared against reference implementation which uses NVIDIA cuBLAS library to perform the same work. Obtained
speedup factors for DCI 4K frame are 2.46 times for largest transform size and 130.17 times for smallest transform size
what revealed substantial performance gap of this library when targeting GPU of the Kepler architecture. Achieved
processing time of frame transform and quantization are up to 4.82 ms.

• Ključne riječi (engleski): Integer discrete cosine transform (DCT), high efficiency video coding (HEVC), graphics processor unit (GPU), matrix multiplication, compute unified device architecture (CUDA), high performance computing (HPC)
• Jezik: engleski
• Vrsta publikacije: Znanstveni rad - Izvorni znanstveni rad
• Status objave: Objavljen
• Vrsta recenzije: Recenziran - međunarodna recenzija
• Verzija publikacije: Objavljena verzija rada (izdavačev PDF)
• Naslov zbornika: Proc. SPIE 11584, 2020 International Conference on Image, Video Processing and Artificial Intelligence
• Brojčani podaci: vol. 11584, str. 81-86
• ISBN: 9781510639973
• e-ISSN: 1996-756X
• DOI: https://doi.org/10.1117/12.2581228
• URN:NBN: urn:nbn:hr:168:263496
• Datum objave publikacije: 2020-11-10
• Projekt: Šifra: 761349; Naziv (hrvatski): TEchnology TRAnsfer via Multinational Application eXperiments; Kratica: TETRAMAX; Pravna nadležnost: eu; Financijer: EC; Linija financiranja: H2020
• Skup: Naziv: Third International Conference on Image, Video Processing and Artificial Intelligence, 2020, Shanghai, China; Lokacija: Shangai, Kina; Organizator: Shanghai Advanced Research Institute, Chinese Academy of Sciences; Datum početka: 2020-08-21; Datum završetka: 2020-08-23; Vrsta izlaganja: Predavanje
• URL dokumenta: https://www.spiedigitallibrary.org/conference-proceedings-of-spie/11584/2581228/Highly-parallel-GPU-accelerator-for-HEVC-transform-and-quantization/10.1117/12.2581228.short
• Vrsta resursa: Tekst
• Izdavač: SPIE
• Mjesto izdavanja: Washington, USA
• Prava pristupa: Otvoreni pristup
• Javna napomena: The work in this paper was partially financed by the MEEP project that has received funding from the European High- Performance Computing Joint Undertaking (JU) under grant agreement No 946002 and by the TETRAMAX project that has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement number 761349.
• Datum i vrijeme pohrane: 2020-12-08 08:36:28