Share Email Print
cover

Proceedings Paper

Software-based MPEG-2 encoding system with scalable and multithreaded architecture
Author(s): Ishfaq Ahmad; Dick-Kwong Yeung; Weiguo Zheng; Shehzad Mehmood
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

MPEG-2 video encoders are now available in a variety of forms using both hardware and software based approaches. The software-based approach potentially offers a better picture quality but is computationally quite intensive. MPEG-2 video encoding can be fast processed using parallelism. A number of approaches using parallel machines or networks of workstations have been reported. While these approaches promise good concepts they do not offer commercial solutions due to factors such as cost, size, etc. In this paper, we propose a new approach with the aim to build a cost-effective and a completely practical solution that is not only highly efficient but is also scalable from single-processor to multiple-processor PC. The highlights of the proposed work include an algorithm for enhancing the efficiency of motion estimation, speeding up the computation of motion estimation and DCT with Intel's SIMD (Single Instruction, Multiple Data) style MMX and SSE instruction sets within a single processor, and scheduling and allocation of a multithreading scheme on a multiple processor PC for managing I/O, synchronization, audio and video encoding, and multiplexing. The proposed multithreaded encoder exploits temporal parallelism in MPEG video sequences with small overhead. The encoder, providing a complete compression solution, achieves faster than the real-time and half of real-time encoding rates for CIF (352 x 288) and CCIR601 (720 x 576) video sequences, respectively, on multiple processor PC.

Paper Details

Date Published: 27 July 2001
PDF: 6 pages
Proc. SPIE 4528, Commercial Applications for High-Performance Computing, (27 July 2001); doi: 10.1117/12.434875
Show Author Affiliations
Ishfaq Ahmad, Hong Kong Univ. of Science and Technology (Hong Kong)
Dick-Kwong Yeung, Hong Kong Univ. of Science and Technology (Hong Kong)
Weiguo Zheng, Hong Kong Univ. of Science and Technology (Hong Kong)
Shehzad Mehmood, Hong Kong Univ. of Science and Technology (Hong Kong)


Published in SPIE Proceedings Vol. 4528:
Commercial Applications for High-Performance Computing
Howard Jay Siegel, Editor(s)

© SPIE. Terms of Use
Back to Top