• DocumentCode
    2875709
  • Title

    Fast AVS Prediction Residual and Integer DCT Implementations for VLIW DSP

  • Author

    Jia Li-na ; Li Fu-jiang ; Zhang Gang

  • Author_Institution
    Coll. of Inf. Eng., Taiyuan Univ. of Technol., Taiyuan, China
  • fYear
    2009
  • fDate
    19-20 Dec. 2009
  • Firstpage
    1
  • Lastpage
    4
  • Abstract
    The very-long-instruction-word architecture is widely used for DSP. In AVS coding process, the prediction residual and integer DCT are frequently called and the methods of optimizing the loops are significant for reducing coding time. Based on the VLIW character, the paper analyzes the optimization methods to accelerate the speed of prediction residual and integer DCT. The AVS prediction residual and integer DCT modules are optimized and simulated on the platform of TMS320DM6446. The experiments show that the instruction cycles are reduced by about 71% and 60% for prediction residual and 8 × 8 integer DCT separately.
  • Keywords
    digital signal processing chips; discrete cosine transforms; optimisation; parallel architectures; AVS coding process; AVS prediction residual implementations; TMS320DM6446; VLIW DSP; integer DCT implementations; very-long-instruction-word architecture; Acceleration; Clocks; Digital signal processing; Discrete cosine transforms; Educational institutions; Optimization methods; Parallel processing; Predictive models; VLIW; Video coding;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Information Engineering and Computer Science, 2009. ICIECS 2009. International Conference on
  • Conference_Location
    Wuhan
  • Print_ISBN
    978-1-4244-4994-1
  • Type

    conf

  • DOI
    10.1109/ICIECS.2009.5366935
  • Filename
    5366935