DocumentCode :
2180684
Title :
A hierarchical Ant-Colony heuristic for architecture synthesis for on-chip communication
Author :
Wei Tang ; Brewer, F.
Author_Institution :
Dept. of Electr. & Comput. Eng., Univ. of California, Santa Barbara, Santa Barbara, CA, USA
fYear :
2013
fDate :
8-10 Oct. 2013
Firstpage :
166
Lastpage :
173
Abstract :
Architecture synthesis and high level synthesis are the paradigms to efficiently organize computations and communications at the high level. While research has been extensively conducted to solve those two problems, a gap between those two paradigms still exists. This paper presents an algorithm for architectural tradeoff for on-chip communication at operation-level granularity. Applied to practical Digital Signal Processing (DSP) and image processing applications, the algorithm can generate superior results over the state-of-the-art ones, while running in sub-quadratic time. Due to the low run time complexity, the algorithm is able to provide task-level applications with solutions that completely describe when and where operations and operands are executed and transferred, respectively. In practice, this opens an opportunity to substantially narrow the gap between high level and architecture level synthesis.
Keywords :
ant colony optimisation; computer architecture; high level synthesis; signal processing; DSP; architectural tradeoff; architecture level synthesis; architecture synthesis; digital signal processing; hierarchical ant-colony heuristic; high level synthesis; image processing applications; low run time complexity; on-chip communication; operation-level granularity; sub-quadratic time; task-level applications; Algorithm design and analysis; Complexity theory; Computer architecture; Partitioning algorithms; Processor scheduling; Schedules; Scheduling;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Design and Architectures for Signal and Image Processing (DASIP), 2013 Conference on
Conference_Location :
Cagliari
Type :
conf
Filename :
6661536
Link To Document :
بازگشت