TY - GEN
T1 - Dynamic VDD switching technique and mapping optimization in dynamically reconfigurable processor for efficient energy reduction
AU - Yamamoto, Tatsuya
AU - Hironaka, Kazuei
AU - Hayakawa, Yuki
AU - Kimura, Masayuki
AU - Amano, Hideharu
AU - Usami, Kimiyoshi
N1 - Copyright:
Copyright 2011 Elsevier B.V., All rights reserved.
PY - 2011
Y1 - 2011
N2 - This paper describes a dynamic VDD switching technique to reduce energy dissipation of Dynamically Reconfigurable Processors. Either high or low supply is dynamically selected at each PE at the context-by-context basis. We designed a part of a PE array and applied this technique. A test chip fabricated in 65nm technology operated successfully. Detailed simulations revealed that energy reduction is hindered by energy overhead due to supply switching when we use even lower VDD. We propose a mapping optimization algorithm "PFCM" to minimize the overhead. PFCM reduced energy overhead by 90.8% and thereby the dynamic VDD switching technique reduced energy dissipation by up to 12.5% when running sepia filter, alpha blender and Laplacian filter programs.
AB - This paper describes a dynamic VDD switching technique to reduce energy dissipation of Dynamically Reconfigurable Processors. Either high or low supply is dynamically selected at each PE at the context-by-context basis. We designed a part of a PE array and applied this technique. A test chip fabricated in 65nm technology operated successfully. Detailed simulations revealed that energy reduction is hindered by energy overhead due to supply switching when we use even lower VDD. We propose a mapping optimization algorithm "PFCM" to minimize the overhead. PFCM reduced energy overhead by 90.8% and thereby the dynamic VDD switching technique reduced energy dissipation by up to 12.5% when running sepia filter, alpha blender and Laplacian filter programs.
KW - Dynamic V Switching
KW - Dynamically Reconfigurable Processor
KW - Mapping Optimization
KW - Power Reduction
UR - http://www.scopus.com/inward/record.url?scp=79953188709&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=79953188709&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-19475-7_24
DO - 10.1007/978-3-642-19475-7_24
M3 - Conference contribution
AN - SCOPUS:79953188709
SN - 9783642194740
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 230
EP - 241
BT - Reconfigurable Computing
T2 - 7th International Symposium on Applied Reconfigurable Computing, ARC 2011
Y2 - 23 March 2011 through 25 March 2011
ER -