Prediction complexity-based HEVC parallel processing for asymmetric multicores

Hyun Joon Roh, Sung Won Han, Eun Seok Ryu

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

This paper proposes a novel Tile allocation method considering the computational ability of asymmetric multicores as well as the computational complexity of each Tile. This paper measures the computational ability of asymmetric multicores in advance, and measures the computational complexity of each Tile by using the amount of HEVC prediction unit (PU) partitioning. The implemented system counts and sorts the amount of PU partitions of each Tile, and also allocates Tiles to asymmetric big.LITTLE cores according to their expected computational complexity. When experiments were conducted, the amount of PU partitioning and the computational complexity (decoding time) showed a close correlation, and average performance gains of decoding time with the proposed adaptive allocation were around 36 % with 12 Tiles, 28 % with 18 Tiles, and 31 % with 24 Tiles, respectively.

Original languageEnglish
Pages (from-to)25271-25284
Number of pages14
JournalMultimedia Tools and Applications
Volume76
Issue number23
DOIs
Publication statusPublished - 2017 Dec 1

Fingerprint

Tile
Processing
Computational complexity
Decoding
Partitions (building)

Keywords

  • Asymmetric multicores
  • HEVC
  • Parallel processing
  • Prediction Complexity

ASJC Scopus subject areas

  • Software
  • Media Technology
  • Hardware and Architecture
  • Computer Networks and Communications

Cite this

Prediction complexity-based HEVC parallel processing for asymmetric multicores. / Roh, Hyun Joon; Han, Sung Won; Ryu, Eun Seok.

In: Multimedia Tools and Applications, Vol. 76, No. 23, 01.12.2017, p. 25271-25284.

Research output: Contribution to journalArticle

@article{7fee7d476b8a4511958533b969a40c2a,
title = "Prediction complexity-based HEVC parallel processing for asymmetric multicores",
abstract = "This paper proposes a novel Tile allocation method considering the computational ability of asymmetric multicores as well as the computational complexity of each Tile. This paper measures the computational ability of asymmetric multicores in advance, and measures the computational complexity of each Tile by using the amount of HEVC prediction unit (PU) partitioning. The implemented system counts and sorts the amount of PU partitions of each Tile, and also allocates Tiles to asymmetric big.LITTLE cores according to their expected computational complexity. When experiments were conducted, the amount of PU partitioning and the computational complexity (decoding time) showed a close correlation, and average performance gains of decoding time with the proposed adaptive allocation were around 36 {\%} with 12 Tiles, 28 {\%} with 18 Tiles, and 31 {\%} with 24 Tiles, respectively.",
keywords = "Asymmetric multicores, HEVC, Parallel processing, Prediction Complexity",
author = "Roh, {Hyun Joon} and Han, {Sung Won} and Ryu, {Eun Seok}",
year = "2017",
month = "12",
day = "1",
doi = "10.1007/s11042-017-4413-7",
language = "English",
volume = "76",
pages = "25271--25284",
journal = "Multimedia Tools and Applications",
issn = "1380-7501",
publisher = "Springer Netherlands",
number = "23",

}

TY - JOUR

T1 - Prediction complexity-based HEVC parallel processing for asymmetric multicores

AU - Roh, Hyun Joon

AU - Han, Sung Won

AU - Ryu, Eun Seok

PY - 2017/12/1

Y1 - 2017/12/1

N2 - This paper proposes a novel Tile allocation method considering the computational ability of asymmetric multicores as well as the computational complexity of each Tile. This paper measures the computational ability of asymmetric multicores in advance, and measures the computational complexity of each Tile by using the amount of HEVC prediction unit (PU) partitioning. The implemented system counts and sorts the amount of PU partitions of each Tile, and also allocates Tiles to asymmetric big.LITTLE cores according to their expected computational complexity. When experiments were conducted, the amount of PU partitioning and the computational complexity (decoding time) showed a close correlation, and average performance gains of decoding time with the proposed adaptive allocation were around 36 % with 12 Tiles, 28 % with 18 Tiles, and 31 % with 24 Tiles, respectively.

AB - This paper proposes a novel Tile allocation method considering the computational ability of asymmetric multicores as well as the computational complexity of each Tile. This paper measures the computational ability of asymmetric multicores in advance, and measures the computational complexity of each Tile by using the amount of HEVC prediction unit (PU) partitioning. The implemented system counts and sorts the amount of PU partitions of each Tile, and also allocates Tiles to asymmetric big.LITTLE cores according to their expected computational complexity. When experiments were conducted, the amount of PU partitioning and the computational complexity (decoding time) showed a close correlation, and average performance gains of decoding time with the proposed adaptive allocation were around 36 % with 12 Tiles, 28 % with 18 Tiles, and 31 % with 24 Tiles, respectively.

KW - Asymmetric multicores

KW - HEVC

KW - Parallel processing

KW - Prediction Complexity

UR - http://www.scopus.com/inward/record.url?scp=85011865482&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85011865482&partnerID=8YFLogxK

U2 - 10.1007/s11042-017-4413-7

DO - 10.1007/s11042-017-4413-7

M3 - Article

AN - SCOPUS:85011865482

VL - 76

SP - 25271

EP - 25284

JO - Multimedia Tools and Applications

JF - Multimedia Tools and Applications

SN - 1380-7501

IS - 23

ER -