Performance Impact of Unaligned Memory Operations in SIMD Extensions for Video Codec Applications

Although SIMD extensions are a cost effective way to exploit the data level parallelism present in most media applications, we will show that they had have a very limited memory architecture with a weak support for unaligned memory accesses. In video codec, and other applications, the overhead for accessing unaligned positions without an efficient architecture support has a big performance penalty and in some cases makes vectorization counter-productive. In this paper we analyze the performance impact of extending the Altivec SIMD ISA with unaligned memory operations. Results show that for several kernels in the H.264/AVC media codec, unaligned access support provides a speedup up to 3.8times compared to the plain SIMD version, translating into an average of 1.2times in the entire application. In addition to providing a significant performance advantage, the use of unaligned memory instructions makes programming SIMD code much easier both for the manual developer and the auto vectorizing compiler

[1]  Emmett Witchel,et al.  Techniques for Increasing and Detecting Memory Alignment , 2001 .

[2]  Vladimir M. Pentkovski,et al.  Implementing Streaming SIMD Extensions on the Pentium III Processor , 2000, IEEE Micro.

[3]  Alan Jay Smith,et al.  Measuring the Performance of Multimedia Instruction Sets , 2002, IEEE Trans. Computers.

[4]  Philippe Roussel,et al.  The microarchitecture of the intel pentium 4 processor on 90nm technology , 2004 .

[5]  Eric Rotenberg,et al.  Trace cache: a low latency approach to high bandwidth instruction fetching , 1996, Proceedings of the 29th Annual IEEE/ACM International Symposium on Microarchitecture. MICRO 29.

[6]  Balaram Sinharoy,et al.  POWER4 system microarchitecture , 2002, IBM J. Res. Dev..

[7]  Pradeep K. Dubey,et al.  How Multimedia Workloads Will Change Processor Design , 1997, Computer.

[8]  Andreas Krall,et al.  Compilation Techniques for Multimedia Processors , 2004, International Journal of Parallel Programming.

[9]  Faouzi Kossentini,et al.  H.264/AVC baseline profile decoder complexity analysis , 2003, IEEE Trans. Circuits Syst. Video Technol..

[10]  Ruby B. Lee,et al.  Challenges to Combining General-Purpose and Multimedia Processors , 1997, Computer.

[11]  Shreekant S. Thakkar,et al.  Internet Streaming SIMD Extensions , 1999, Computer.

[12]  Yen-Kuang Chen,et al.  Implementation of H.264 decoder on general-purpose processors with media instructions , 2003, IS&T/SPIE Electronic Imaging.

[13]  Burzin A. Patel,et al.  Optimization of instruction fetch mechanisms for high issue rates , 1995, Proceedings 22nd Annual International Symposium on Computer Architecture.

[14]  Stamatis Vassiliadis,et al.  The TM3270 media-processor , 2005, 38th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO'05).

[15]  Iain E. G. Richardson,et al.  Video Codec Design: Developing Image and Video Compression Systems , 2002 .

[16]  Jose Fridman.  Data alignment for sub-word parallelism in DSP , 1999, 1999 IEEE Workshop on Signal Processing Systems. SiPS 99. Design and Implementation (Cat. No.99TH8461).

[17]  Mayan Moudgill,et al.  Environment for PowerPC microarchitecture exploration , 1999, IEEE Micro.

[18]  Richard Henderson,et al.  Multi-platform auto-vectorization , 2006, International Symposium on Code Generation and Optimization (CGO'06).

[19]  Mateo Valero,et al.  Adding a vector unit to a superscalar processor , 1999, ICS '99.

[20]  Peng Wu,et al.  Vectorization for SIMD architectures with alignment constraints , 2004, PLDI '04.

[21]  Hunter Scales,et al.  AltiVec Extension to PowerPC Accelerates Media Processing , 2000, IEEE Micro.

[22]  D. Marpe,et al.  Video coding with H.264/AVC: tools, performance, and complexity , 2004, IEEE Circuits and Systems Magazine.