DocumentCode :
395276
Title :
Short vector code generation and adaptation for DSP algorithms
Author :
Franchetti, Franz ; Püschel, Markus
Author_Institution :
Vienna Univ. of Technol., Austria
Volume :
2
fYear :
2003
fDate :
6-10 April 2003
Abstract :
Most recent general purpose processors feature short vector SIMD instructions, like SSE on Pentium III/4. We automatically generate platform-adapted short vector code for DSP transform algorithms using SPIRAL. SPIRAL represents and generates fast algorithms as mathematical formulas, and translates them into code. Adaptation is achieved by searching in the space of algorithmic and coding alternatives for the fastest implementation on the given platform. We explain the mathematical foundation that relates formula constructs to vector code, and overview the vector code generator within SPIRAL. Experimental results show excellent speed-ups compared to ordinary C code for a variety of transforms and computing platforms. For the DFT on Pentium 4, our automatically generated code compares favorably with the hand-tuned Intel MKL vendor library.
Keywords :
digital signal processing chips; discrete Fourier transforms; discrete cosine transforms; instruction sets; program compilers; DCT; DFT; DSP algorithms; Intel MKL vendor library; Pentium 4; Pentium III; SIMD instructions; SPIRAL library generator; Walsh-Hadamard transform; computing platforms; discrete cosine transform; formula constructs; formula generator; formula translator; general purpose processors; instruction set; mathematical formulas; short vector code generation; short vector code generator; signal processing language; single instruction multiple data instructions; Automatic programming; Computer aided instruction; Digital signal processing; Discrete Fourier transforms; Discrete cosine transforms; Discrete transforms; Fourier transforms; Libraries; Mathematics; Spirals;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03). 2003 IEEE International Conference on
ISSN :
1520-6149
Print_ISBN :
0-7803-7663-3
Type :
conf
DOI :
10.1109/ICASSP.2003.1202422
Filename :
1202422
Link To Document :
بازگشت