Title :
Common subexpression elimination involving multiple variables linear DSP synthesis
Author :
Hosangadi, Anup ; Fallah, Farzan ; Kastner, Ryan
Author_Institution :
California Univ., Santa Barbara, CA, USA
Abstract :
Common subexpression elimination is commonly employed to reduce the number of operations in DSP algorithms after decomposing constant multiplications into shifts and additions. Conventional optimization techniques for finding common subexpressions can optimize constant multiplications with only a single variable at a time, and hence cannot fully optimize the computations with multiple variables found in matrix form of linear systems like DCT, DFT etc. We transform these computations such that all common subexpressions involving any number of variables can be detected. We then present heuristic algorithms to select the best set of common subexpressions. Experimental results show the superiority of our technique over conventional techniques for common subexpression elimination.
Keywords :
heuristic programming; high level synthesis; optimisation; signal processing; DSP algorithms; common subexpression elimination; digital signal processing; multiple variables linear DSP synthesis; Digital signal processing; Discrete Fourier transforms; Discrete cosine transforms; Distributed power generation; Embedded system; Energy consumption; Hardware; Heuristic algorithms; Linear systems; Matrix decomposition;
Conference_Titel :
Application-Specific Systems, Architectures and Processors, 2004. Proceedings. 15th IEEE International Conference on
Print_ISBN :
0-7695-2226-2
DOI :
10.1109/ASAP.2004.1342471