DMCA
SPIRAL: Code Generation for DSP Transforms
Cached
Download Links
| Venue: | PROCEEDINGS OF THE IEEE SPECIAL ISSUE ON PROGRAM GENERATION, OPTIMIZATION, AND ADAPTATION |
| Citations: | 222 - 41 self |
BibTeX
@MISC{Püschel_spiral:code,
author = {Markus Püschel and José M. F. Moura and Jeremy Johnson and David Padua and Manuela Veloso and Bryan W. Singer and Jianxin Xiong and Franz Franchetti and Aca Gačić and Yevgen Voronenko and Kang Chen and Robert W. Johnson and Nicholas Rizzolo},
title = { SPIRAL: Code Generation for DSP Transforms},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Fast changing, increasingly complex, and diverse computing platforms pose central problems in scientific computing: How to achieve, with reasonable effort, portable optimal performance? We present SPIRAL that considers this problem for the performance-critical domain of linear digital signal processing (DSP) transforms. For a specified transform, SPIRAL automatically generates high performance code that is tuned to the given platform. SPIRAL formulates the tuning as an optimization problem, and exploits the domain-specific mathematical structure of transform algorithms to implement a feedback-driven optimizer. Similar to a human expert, for a specified transform, SPIRAL “intelligently ” generates and explores algorithmic and implementation choices to find the best match to the computer’s microarchitecture. The “intelligence” is provided by search and learning techniques that exploit the structure of the algorithm and implementation space to guide the exploration and optimization. SPIRAL generates high performance code for a broad set of DSP transforms including the discrete Fourier transform, other trigonometric transforms, filter transforms, and discrete wavelet transforms. Experimental results show that the code generated by SPIRAL competes with, and sometimes outperforms, the best available human tuned transform library code.
Keyphrases
dsp transforms code generation ieee special issue program generation discrete fourier transform high performance code specified transform implementation choice present spiral index term library generation implementation space broad set discrete cosine transform feedback-driven optimizer trigonometric transforms scientific computing abstract fast changing diverse computing platform evolutionary algorithm domain-specific mathematical structure central problem automatic performance tuning code optimization available human optimization problem discrete wavelet transforms linear digital signal processing transform algorithm filter transforms human expert markov decision process experimental result reasonable effort computer microarchitecture high performance computing spiral competes transform library code performance-critical domain portable optimal performance linear signal transform
