25 October 1995 MPEG-2 video decoding on programmable processors: computational and architectural requirements
Author Affiliations +
Proceedings Volume 10282, Standards and Common Interfaces for Video Information Systems: A Critical Review; 102820F (1995); doi: 10.1117/12.227954
Event: Photonics East '95, 1995, Philadelphia, PA, United States
Abstract
Design of a cost-effective microprocessor architecture for MPEG video decoding requires not only the analysis of MPEG algorithms in terms of the raw number of additions or multiplications, but more importantly, a careful study of mapping those algorithms to the target processor architecture at the instruction level. Even though smart compilers can generate fast and compact codes that rival the hand-optimized code on sequential algorithms, a more accurate evaluation of the processor performance on parallel algorithms such as the core functions in the MPEG-2 decoding is achieved by hand-optimizing the tight loop.

In this paper, we analyze the computational requirements of the key MPEG-2 decoding functions from the instruction level. We use a generic sequential processor model with a RISC-like instruction set to map the tight loop of the key functions to machine instructions. We then extend our processor model to instruction-level parallel (ILP) architectures and present the speed improvement of each key function on two different ILP features and as the degree of parallel processing is increased.
© (1995) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Woobin Lee, Yongmin Kim, "MPEG-2 video decoding on programmable processors: computational and architectural requirements", Proc. SPIE 10282, Standards and Common Interfaces for Video Information Systems: A Critical Review, 102820F (25 October 1995); doi: 10.1117/12.227954; https://doi.org/10.1117/12.227954
PROCEEDINGS
23 PAGES


SHARE
KEYWORDS
Detection and tracking algorithms

Video

Video processing

Process modeling

Parallel computing

Parallel processing

Back to Top