DocumentCode :
2245714
Title :
Automatic Identification of Custom Functions for Embedded Processors with MIMO Extensions
Author :
Chen, Xiaoyong ; Maskell, Douglas L. ; Sun, Yang
Author_Institution :
Sch. of Comput. Eng., Nanyang Technol. Univ., Singapore
fYear :
2006
fDate :
4-7 Dec. 2006
Firstpage :
1156
Lastpage :
1159
Abstract :
Commercially available configurable processors nowadays offer opportunities to accelerate embedded applications by implementing customized functions. Previous studies have shown that more speedup can be achieved if these functions are allowed to have more than two inputs and one output. However, automatic identification of multiple-input, multiple-output (MIMO) custom functions is usually slow due to the large exploration space. In this paper, we propose a fast algorithm to enumerate all convex subgraphs that satisfy the input/output constraints from the dataflow graph (DFG) of a basic block. The algorithm can be tuned to determine all subgraphs or only those connected subgraphs. This allows a choice between a better instruction set extension or a faster design space exploration. The algorithm´s run time can be orders of magnitude better than previous algorithms when a custom function is allowed to have two or more outputs
Keywords :
MIMO systems; algorithm theory; data flow graphs; embedded systems; identification; instruction sets; logic design; microprocessor chips; MIMO extensions; automatic identification; configurable processors; convex subgraphs; custom functions; dataflow graph; embedded processors; fast algorithm; faster design space exploration; instruction set extension; Acceleration; Algorithm design and analysis; Application software; Embedded computing; Genetic algorithms; Hardware; MIMO; Space exploration; Sun; Variable speed drives; custom function; instruction set extension;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Circuits and Systems, 2006. APCCAS 2006. IEEE Asia Pacific Conference on
Conference_Location :
Singapore
Print_ISBN :
1-4244-0387-1
Type :
conf
DOI :
10.1109/APCCAS.2006.342346
Filename :
4145603
Link To Document :
بازگشت