臺灣博碩士論文加值系統

本篇論文主要目的是實作一個可攜式低必v音訊播放器，在達到低必v設計的概念上，我們是利用軟體最佳化來降低CPU工作頻率，進而達成低必v的設計目標。本論文之實作的平台有UniRISC與ARM7TDMI，其操作頻率均為200MHz，我們將取得的來源程式碼整合於上述兩個平台，透過初步數據分析來得知程式執行時所佔週期數多的部份，再透過軟體最佳化以針對這些運算比重大的部份來改善其效能。本論文所提出之改善的方法分成幾個種類，包含採用快速演算法、改進原本的程式寫法、使用inline assembly來撰寫程式以及利用UniRISC的硬體資源提出針對音訊處理加速的指令，採用所提出之方法後，我們統計每個音訊解碼器在不同平台上所實測得到的數據，MP3在UniRISC上增進了15.5%且CPU時脈降到37MHz左右，在ARM上面增進了5.6%且CPU時脈能降到48MHz；MPEG 2/4 AAC在UniRISC上增進了24.6%且CPU時脈降至39MHz，在ARM上增進了42.5%且CPU時脈降到50MHz；OGG在UniRISC上增進了58.1%且CPU時脈降至45MHz，在ARM上改善了54.3%且CPU時脈降至48MHz。大約可以看出經過改善後的解碼器效能，在兩個平台上大多可以在50MHz以下的CPU工作時脈上即時解碼。

This thesis presents an low power design and realization of audio decoders for portable multimedia player applications. We take advantage of the software optimization to reduce CPU’s clock rate and consequently reach the low power design. In this thesis, we carry out our design on UniRISC and ARM7TDMI these two platforms. We also take a profile on the reference software to extract the distribution of computation complexity and then utilize software optimization to improve its efficiency to the part of the large proportion. We propose several kinds of methods that are adopting fast algorithm, amending reference source code, using inline assembly and proposed audio acceleration instructions. After adopting the proposed methods, we gather statistics of every audio decoder at each platform. MP3 has 15.5% and 5.6% enhancement in performance and can execute at 37MHz and 48MHz on UniRISC and ARM7TDMI, respectively. AAC has 24.6% and 42.5% enhancement in performance and can execute at 39MHz and 50MHz on UniRISC and ARM7TDMI, respectively. AAC has 58.1% and 54.3% enhancement in performance and can execute at 45MHz and 48MHz on UniRISC and ARM7TDMI, respectively. Finally, we can achieve real-time decoding with a clock rate less than 50MHz on these two platforms based on our improvement.

摘要 I
ABSTRACT II
CHAPTER 1 INTRODUCTION 1
1.1. BACKGROUND AND MOTIVATION 1
1.2. DESIGN CHALLENGES 1
1.3. SCOPE OF THE THESIS 2
CHAPTER 2 PREVIOUS WORK 3
2.1. SOFTWARE OPTIMIZATION IN MODERN AUDIO DECODER 3
2.2. UNIRISC PLATFORM 7
2.2.1. RISC in UniCore 7
2.2.2. RISC Function Unit Hardware 8
2.2.3. Software Development Kit 9
2.3. MP3 DECODER 9
2.4. MPEG 2/4 AAC DECODER 21
2.5. OGG DECODER 26
CHAPTER 3 REFERENCE CODE PORTING ON GENERAL PLATFORM 32
3.1. REFERENCE CODE MODIFICATION 32
3.1.1. I/O Rewrite Method 32
3.1.2. Memory Management 35
3.2. VERIFICATION METHOD 37
3.3. PERFORMANCE EVALUATION 39
3.3.1. Test Environment 39
3.3.2. Performance of MP3 Decoder 39
3.3.3. Performance of MPEG 2/4 AAC Decoder 41
3.3.4. Performance of OGG Decoder 42
CHAPTER 4 SOFTWARE OPTIMIZED STRATEGY 45
4.1. IMDCT OPTIMIZATION 45
4.1.1. DCT-based IMDCT 45
4.1.2. FFT-based IMDCT 47
4.2. IMPROVEMENT OF THE CODING STYLE 48
4.2.1. Suitable for Three kind of Decoders 48
4.2.2. Suitable for the Specific Decoder 51
4.3. INLINE ASSEMBLY CODING 52
4.4. PROPOSED AUDIO ACCELERATION INSTRUCTIONS 55
4.5. PERFORMANCE EVALUATION 56
4.3.1. Test Environment 56
4.3.2. Performance of MP3 Decoder 57
4.3.3. Performance of MPEG 2/4 AAC Decoder 59
4.3.4. Performance of OGG Decoder 61
4.6. PERFORMANCE COMPARISON 64
CHAPTER 5 CONCLUSION AND FUTURE WORK 67
5.1. CONCLUSION 67
5.2. FUTURE WORK 68
REFERENCE 69

[1]ISO/IEC 11172-3, “Coding of moving picture and associated audio for digital storage media at up to about 1.5Mbit/s, Part 3: Audio”, Aug. 1993.
[2]Y. Yao, Q. Yao, P. Liu, and Z. Xiao, ”Embedded software optimization for MP3 decoder implemented on RISC core”, IEEE Trans. on Consumer Electronics, vol. 50, no. 4, pp. 1244-1249, Nov. 2004.
[3]M. Anguita and J. M. Martinez-Lechado, “MP3 optimization exploiting processor architecture and using better algorithms”, IEEE Micro, vol. 25, no. 3, pp. 81-92, May-June 2005.
[4]B. D. Choi, K. S. Choi, S. J. Ko, and A. W. Morales, “Efficient real-time implementation of MPEG-4 audiovisual decoder using DSP and RISC chips”, Proc. ICCE, pp. 246-247, June 2003.
[5]Kosaka, S. Yamaguchi, H. Okuhata, T. Onoye, and I. Shirakawa, “VLSI implementation of Ogg Vorbis decoder for embedded applications”, Proc. 15th Annual IEEE International ASIC/SOC Conference, pp. 20-24, Sept. 2002.
[6]M. A. Watson and P. Buttner, “Design and implementation of AAC Decoders”, Proc. ICCE, pp.408-409, June 2000.
[7]A. Kosaka, S. Yamaguchi, H. Okuhata, T. Onoye and I. Shirakawa, “SoC design of Ogg Vorbis decoder using embedded processor”, Proc. 1st Computing frontiers, pp.481-487, April 2004.
[8]K. H. Bang, J. S. Kim, Y. C. Park and D. H. Youn, “Design optimization of a dual MP3/AAC decoder”, Proc. ICASSP, vol. 4, pp. 3776 – 3779, May 2002.
[9]S. D. You and Y. C. Hou, ” Implementation of IMDCT for MPEG2/4 AAC on 16-bit fixed-point digital signal processors”, Proc. Asia-Pacific Conference on Circuit and Systems, vol. 2, pp. 813 – 816, Dec. 2004.
[10]V. Gurkhe, “Optimization of an MP3 decoder on the ARM processor”, Proc. Conference on Convergent Technologies for Asia-Pacific Region, TENCON 2003, vol. 4, pp. 1475 – 1478, Oct. 2003.
[11]K. H. Bang, J. S. Kim, Y. C. Park and D. H. Youn, “Design optimization of MPEG-2 AAC decoder”, IEEE Trans. on Consumer Electronics, vol. 4, no. 4, pp. 895 – 903, Nov. 2001
[12]K. S. Lee, Y. C. Park and D. H. Youn, “Software optimization of the MPEG-audio decoder using a 32-bit MCU RISC processor”, IEEE Trans. on Consumer Electronics, vol. 48, no. 3, pp. 671 - 676, Aug. 2002.
[13]C. K. Yang and S. G. Chen, “New static and dynamic search algorithms for fast MP3 bit allocations”, Proc. ICME, vol. 1, pp. 77-80, July 2003.
[14]Y. K. Cho, T. H. Song and H. S. Kim, “An optimized algorithm for computing the modified discrete cosine transform and its inverse transform”, Proc. TENCON, vol. A, pp. 626 – 628, Nov. 2004.
[15]S. Shamshiri and S.M. Fakhraie, “Parallel alias reduction for MP3 decoding”, Proc. 16th ICM, pp. 438 – 441, Dec. 2004.
[16]S. W. Lee, “Improved algorithm for efficient computation of the forward and backward MDCT in MPEG audio coder”, IEEE Trans. on Circuits and Systems II: Analog and Digital Signal Processing, vol. 48, no. 10, pp. 990-994, Oct. 2001.
[17]V. Britanak and K. R. Rao, “An efficient implementation of the forward and inverse MDCT in MPEG audio coding”, IEEE Signal Processing Letters, vol. 8, no. 2, pp. 48-51, Feb. 2001.
[18]P. Duhamel, Y. Mahieux, and J. P. Petit, ”A fast algorithm for the implementation of filter banks based on time domain aliasing cancellation”, Proc. ICASSP, vol. 3, pp. 2209-2212, Apr. 1991.
[19]ISO/IEC 13818-7, “Generic coding of moving pictures and associated audio, Part 7: Advanced Audio Coding”, Apr. 1997.
[20]ISO/IEC 14496-3, “Coding of audio-visual objects, Part 3: Audio”, Oct. 1998.
[21]M. K. Tsatsanis and G. B. Giannakis, “Multirate filter banks for code-division multiple access systems”, Proc. ICASSP, vol. 2, pp. 1484-1487, May 1995.