Leveraging parallel computing in modern video coding standards

Kiho Choi, Euee S. Jang

Research output: Contribution to journalArticle

12 Citations (Scopus)

Abstract

The influence of the parallel computing trend on the design of modern video coding standards is reviewed. A typical video encoder is required to encode each video frame within tens of milliseconds to support real-time applications at a given frame rate. Generally, the number of cores proportionally improves system performance. A simple method for accelerating the decoding process with parallel computing is to remove dependency at the bitstream level by specifying the bitstream syntax so that several independent chunks of bits can be decoded simultaneously. Another possible method for removing dependency is to exploit the video coding structure to better support parallel computing with additional syntax information. The co-exploration between algorithm and architecture (CEAA) approach to video coding will draw attention because it attempts to address conventional algorithm definition requirements as well as implementation requirements by exploring algorithm and architecture together.

Original languageEnglish
Article number6272278
Pages (from-to)7-11
Number of pages5
JournalIEEE Multimedia
Volume19
Issue number3
DOIs
StatePublished - 2012 Sep 19

Fingerprint

Parallel processing systems
Image coding
Decoding

Keywords

  • CEAA
  • co-exploration between algorithm and architecture
  • HEVC
  • High Efficiency Video Coding standard
  • multicore processing
  • multimedia
  • multimedia standards
  • parallel processing
  • video coding standards

Cite this

@article{822cfd8ac8c64a58be8970a1cb48fda9,
title = "Leveraging parallel computing in modern video coding standards",
abstract = "The influence of the parallel computing trend on the design of modern video coding standards is reviewed. A typical video encoder is required to encode each video frame within tens of milliseconds to support real-time applications at a given frame rate. Generally, the number of cores proportionally improves system performance. A simple method for accelerating the decoding process with parallel computing is to remove dependency at the bitstream level by specifying the bitstream syntax so that several independent chunks of bits can be decoded simultaneously. Another possible method for removing dependency is to exploit the video coding structure to better support parallel computing with additional syntax information. The co-exploration between algorithm and architecture (CEAA) approach to video coding will draw attention because it attempts to address conventional algorithm definition requirements as well as implementation requirements by exploring algorithm and architecture together.",
keywords = "CEAA, co-exploration between algorithm and architecture, HEVC, High Efficiency Video Coding standard, multicore processing, multimedia, multimedia standards, parallel processing, video coding standards",
author = "Kiho Choi and Jang, {Euee S.}",
year = "2012",
month = "9",
day = "19",
doi = "10.1109/MMUL.2012.36",
language = "English",
volume = "19",
pages = "7--11",
journal = "IEEE Multimedia",
issn = "1070-986X",
number = "3",

}

Leveraging parallel computing in modern video coding standards. / Choi, Kiho; Jang, Euee S.

In: IEEE Multimedia, Vol. 19, No. 3, 6272278, 19.09.2012, p. 7-11.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Leveraging parallel computing in modern video coding standards

AU - Choi, Kiho

AU - Jang, Euee S.

PY - 2012/9/19

Y1 - 2012/9/19

N2 - The influence of the parallel computing trend on the design of modern video coding standards is reviewed. A typical video encoder is required to encode each video frame within tens of milliseconds to support real-time applications at a given frame rate. Generally, the number of cores proportionally improves system performance. A simple method for accelerating the decoding process with parallel computing is to remove dependency at the bitstream level by specifying the bitstream syntax so that several independent chunks of bits can be decoded simultaneously. Another possible method for removing dependency is to exploit the video coding structure to better support parallel computing with additional syntax information. The co-exploration between algorithm and architecture (CEAA) approach to video coding will draw attention because it attempts to address conventional algorithm definition requirements as well as implementation requirements by exploring algorithm and architecture together.

AB - The influence of the parallel computing trend on the design of modern video coding standards is reviewed. A typical video encoder is required to encode each video frame within tens of milliseconds to support real-time applications at a given frame rate. Generally, the number of cores proportionally improves system performance. A simple method for accelerating the decoding process with parallel computing is to remove dependency at the bitstream level by specifying the bitstream syntax so that several independent chunks of bits can be decoded simultaneously. Another possible method for removing dependency is to exploit the video coding structure to better support parallel computing with additional syntax information. The co-exploration between algorithm and architecture (CEAA) approach to video coding will draw attention because it attempts to address conventional algorithm definition requirements as well as implementation requirements by exploring algorithm and architecture together.

KW - CEAA

KW - co-exploration between algorithm and architecture

KW - HEVC

KW - High Efficiency Video Coding standard

KW - multicore processing

KW - multimedia

KW - multimedia standards

KW - parallel processing

KW - video coding standards

UR - http://www.scopus.com/inward/record.url?scp=84866240724&partnerID=8YFLogxK

U2 - 10.1109/MMUL.2012.36

DO - 10.1109/MMUL.2012.36

M3 - Article

AN - SCOPUS:84866240724

VL - 19

SP - 7

EP - 11

JO - IEEE Multimedia

JF - IEEE Multimedia

SN - 1070-986X

IS - 3

M1 - 6272278

ER -