MuGRA: A Scalable Multi-Grained Reconfigurable Accelerator Powered by Elastic Neural Network
- 27 July 2021
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 69 (1), 258-271
- https://doi.org/10.1109/tcsi.2021.3099034
Abstract
A massive core computing architecture is developed for accelerating arbitrary calculations in fully parallel with high speed and low cost. The proposed architecture is reconfigurable in fine-grained (arbitrary functions), mid-grained (flexible function feature, accuracy, and number of operands), and coarse-grained (organization of cores). By implementing a large scale of novel bisection neural network (BNN) on hardware, the re-configuration is conducted by partitioning entire BNN into any specific pieces without redundancy. Each piece of BNN retrieves the arbitrary function approximately. By reconfiguring the BNN topology in software, we can easily adjust dimensions of the computing kernel without rewiring, and achieve a wide range of trade-offs between accuracy and efficiency in hardware. In this manner, the multi-grained reconfigurable accelerator (MuGRA) is achieved. Since MuGRA is flexible in all grained levels, various configurations for each validation are demonstrated with rich options of performance-cost matrix. From the FPGA implementation results, compared with other traditional function approximation methods, our method provides fewer parameter storage requirements. The comparison against related works proves that our accelerator effectively reduces the calculation latency with slight accuracy loss.Keywords
Funding Information
- Japan Science and Technology Agency (JST), Precursory Research for Embryonic Science and Technology (PRESTO), Japan (JPMJPR18M7)
This publication has 45 references indexed in Scilit:
- SNNAP: Approximate computing on programmable SoCs via neural accelerationPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2015
- Neural acceleration for general-purpose approximate programsCommunications of the ACM, 2014
- Run-time accelerator binding for tile-based mixed-grained reconfigurable architecturesPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2014
- Neural network-based accelerators for transcendental function approximationPublished by Association for Computing Machinery (ACM) ,2014
- Square-rich fixed point polynomial evaluation on FPGAsPublished by Association for Computing Machinery (ACM) ,2014
- DianNaoACM SIGPLAN Notices, 2014
- A defect-tolerant accelerator for emerging high-performance applicationsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2012
- Programmable Architectures and Design Methods for Two-Variable Numeric Function GeneratorsIPSJ Transactions on System LSI Design Methodology, 2010
- Low-Power FPGA-Implementation of atan(Y/X) Using Look-Up Table Methods for Communication ApplicationsJournal of Signal Processing Systems, 2008
- Learning Internal Representations by Error PropagationPublished by Elsevier BV ,1988