• DocumentCode
    2311675
  • Title

    Recent changes and future trends in general purpose processor architectures to support image and video applications

  • Author

    Debes, Eric

  • Author_Institution
    Microprocessor Res., Intel Labs, Santa Clara, CA, USA
  • Volume
    3
  • fYear
    2003
  • fDate
    14-17 Sept. 2003
  • Abstract
    Image and video applications have been driving general- purpose processors (GPPs) and systems for more than a decade and will continue influencing the design of next generation architectures. In this paper we will first summarize the main characteristics of image and video applications compared to other workloads running on GPPs. Then we will give an overview of the recent changes and future trends in general purpose processor architectures to better handle image and video applications. First we will show that new SlMD (single instruction multiple data) features allow better exploitation of data-level parallelism and that media instructions have been added for some very specific algorithms. Another important aspect for image and video applications running on GPPs is the increasing memory latency compared to the processing time. In this context, the hardware prefetcher is slowly replacing software-prefetching instructions. In addition we will describe how additional support for multithreading like the newly introduced Intel Hyperthreading Technology or future multicore processors can benefit image and video applications. Finally we will explain how recently introduced frequency voltage scaling features can help to reduce power consumption for media applications due to their regular execution and fixed dead-line characteristics.
  • Keywords
    computer architecture; image processing; microprocessor chips; multimedia communication; multimedia computing; parallel processing; storage management; Intel Hyperthreading Technology; SIMD; data-level parallelism; frequency voltage scaling; general purpose processor; image application; memory latency; multicore processor; power consumption reduction; single instruction multiple data; video application; Application software; Educational institutions; Energy consumption; Image coding; Microprocessors; Multithreading; Parallel processing; Personal digital assistants; Streaming media; Yarn;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Image Processing, 2003. ICIP 2003. Proceedings. 2003 International Conference on
  • ISSN
    1522-4880
  • Print_ISBN
    0-7803-7750-8
  • Type

    conf

  • DOI
    10.1109/ICIP.2003.1247187
  • Filename
    1247187