A4 Refereed article in a conference publication
DyMeP: An infrastructure to support Dynamic Memory binding for runtime mapping in CGRAs
Authors: Tajammul MA, Jafri S, Ellerve P, Hemani A, Tenhunen H, Plosila J
Editors: Editorial production by Lisa O’Conner
Conference name: International Conference on VLSI Design
Publication year: 2015
Book title : Proceedings of the 28th International Conference on VLSI Design
Journal name in source: 2015 28TH INTERNATIONAL CONFERENCE ON VLSI DESIGN (VLSID)
Journal acronym: I CONF VLSI DESIGN
Series title: International Conference on VLSI Design
First page : 547
Last page: 552
Number of pages: 6
ISBN: 978-1-4799-6659-2
eISBN: 978-1-4799-6658-5
ISSN: 1063-9667
DOI: https://doi.org/10.1109/VLSID.2015.98
Web address : http://ieeexplore.ieee.org/document/7031792/
Coarse Grained Reconfigurable Architectures (CGRAs) are emerging as enabling platforms to meet the high performance demanded by modern applications. Commonly, CGRAs are composed of a computation layer (that performs computations) and a memory layer (that provides data and configware to the computation layer). Tempted by higher platform utilization and reliability, recently proposed CGRAs offer dynamic application remapping (for the computation layer). Distributed scratch pad (compiler programmed) memories offer high data rates, predictability and low the power consumption (compared to caches). Therefore, the distributed scratchpad memories are emerging as preferred implementation alternative for the memory layer in recent CGRAs. However, the scratchpad memories are programmed at compile time, and do not support dynamic application remapping. The existing solutions that allow dynamic application remapping either rely on fat binaries (that significantly enhance configuration memory requirements) or consider a centralized memory. To extract the benefits of both runtime remapping and distributed scratchpad memories, we present a design framework called DyMeP. DyMeP relies on late binding and provides the architectural support to dynamically remap data in CGRAs. Compared to the state of the art, the proposed technique reduces the configuration memory requirements (needed by fat binary solutions) and supports distributed shared scratchpad memory. Synthesis/Simulation results reveal that DyMeP promises a significant (up to 60%) reduction in configware size at the cost of negligible additional overheads (less then 1%).
