Performance comparison of SIMD-based HEVC decoders on mobile processor
- Authors
- Van Dien, N.; Ryu, E.-S.
- Issue Date
- 2017
- Publisher
- Institute of Electrical and Electronics Engineers Inc.
- Keywords
- HEVC; Libde265; NEON; OpenHEVC; Parallel processing; Performance comparison; SIMD
- Citation
- Proceedings of KICS-IEEE International Conference on Information and Communications with Samsung LTE and 5G Special Workshop, ICIC 2017, pp.298 - 303
- Journal Title
- Proceedings of KICS-IEEE International Conference on Information and Communications with Samsung LTE and 5G Special Workshop, ICIC 2017
- Start Page
- 298
- End Page
- 303
- URI
- https://scholarworks.bwise.kr/gachon/handle/2020.sw.gachon/6675
- DOI
- 10.1109/INFOC.2017.8001683
- ISSN
- 0000-0000
- Abstract
- This paper implements SIMD mechanism on ARM-based mobile processor for HEVC Decoders. The experimental results evaluating the performance of HEVC Decoders with/without SIMD are presented. In summary, the decoding rate with SIMD is 4∼5 times higher than decoding rate without SIMD. The results are up to 28 frames/s on multi-core ARM64 architecture with full HD video. The experiments also show the decoding performance comparison between the well-known open-source HEVC decodes such as Libde265 and openHEVCDecoder. In general, Libde265 achieved better decoding speedup than openHEVCDecoder in case of tile-based parallel processing. But it was less efficient in case of single threaded decoding because of the thread allocation on asymmetric multi-core processors. All experiments are conducted on Samsung Galaxy S6 with arm 64-bit architecture. © 2017 IEEE.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - ETC > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/gachon/handle/2020.sw.gachon/6675)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.