DMCA
High-level control flow transformations for performance improvement of address-dominated multimedia applications (2003)
| Venue: | In proceedings of 11th Workshop on Synthesis and System Integration of Mixed Information technologies (SASIMI |
| Citations: | 2 - 0 self |
BibTeX
@INPROCEEDINGS{Falk03high-levelcontrol,
author = {Heiko Falk and Cédric Ghez and Miguel Miranda and Rainer Leupers},
title = {High-level control flow transformations for performance improvement of address-dominated multimedia applications},
booktitle = {In proceedings of 11th Workshop on Synthesis and System Integration of Mixed Information technologies (SASIMI},
year = {2003},
pages = {135--150},
publisher = {Hall}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Abstract — This paper describes a set of novel highlevel control flow transformations for performance improvement of typical address-dominated multimedia applications. We show that these transformations applied at the source code level can have a very large impact on execution time at the cost of limited overhead in code size for a broad range of instruction set processor families (i. e. CISC, RISC, DSP, VLIW,...). For a profound evaluation, all transformations are applied to the C-codes of two real-life applications selected from the video and image processing domains. A detailed analysis of the effect of the transformations is done by compiling and executing the transformed programs on seven different programmable processors. The measured runtimes indicate quite significant improvements in all processor families when comparing the performance of the transformed codes to their initial version even when these are compiled using their native optimizing compilers with their most aggressive optimization features enabled. The average gains in execution time range from 40.2 % and 87.7 % depending on the driver, with an average overhead in code size between 21.1 % and 100.9%. I.
Keyphrases
performance improvement high-level control flow transformation address-dominated multimedia application code size processor family broad range novel highlevel control flow transformation profound evaluation significant improvement average gain large impact different programmable processor source code level initial version transformed code measured runtimes execution time typical address-dominated multimedia application image processing domain execution time range real-life application detailed analysis transformed program aggressive optimization feature average overhead limited overhead
