DART: Design Accelerators by Regulating Transformations
Reference Number
EP/V028251/1
Title
DART: Design Accelerators by Regulating Transformations
Status
Completed
Energy Categories
Energy Efficiency(Other)
Not Energy Related
Not Energy Related
Research Types
Basic and strategic applied research
Science and Technology Fields
PHYSICAL SCIENCES AND MATHEMATICS (Computer Science and Informatics)
UKERC Cross Cutting Characterisation
Not Cross-cutting
Principal Investigator
Professor W Luk
Computing
Imperial College London
Computing
Imperial College London
Award Type
Standard
Funding Source
EPSRC
Start Date
01 October 2021
End Date
30 September 2024
Duration
36 months
Total Grant Value
£613,910
Industrial Sectors
Info. & commun. Technol.
Region
London
Programme
NC : ICT
Industrial Collaborator
Project Contact, Maxeler Technologies Ltd
Project Contact, Tianjin University, China
Project Contact, Intel Corporation (UK) Ltd
Project Contact, Corerain Technologies
Project Contact, Stanford University, USA
Project Contact, Cornell University, USA
Project Contact, University of British Columbia
Project Contact, Dunnhumby
Project Contact, Microsoft Research Ltd
Project Contact, Deloitte LLP
Project Contact, Xilinx Ireland
Project Contact, RIKEN
Project Contact, Tianjin University, China
Project Contact, Intel Corporation (UK) Ltd
Project Contact, Corerain Technologies
Project Contact, Stanford University, USA
Project Contact, Cornell University, USA
Project Contact, University of British Columbia
Project Contact, Dunnhumby
Project Contact, Microsoft Research Ltd
Project Contact, Deloitte LLP
Project Contact, Xilinx Ireland
Project Contact, RIKEN
Web Site
Objectives
Abstract
The DART project aims to pioneer a ground-breaking capability to enhance the performance and energy efficiency of reconfigurable hardware accelerators for next-generation computing systems. This capability will be achieved by a novel foundation for a transformation engine based on heterogeneous graphs for design optimisation and diagnosis. While hardware designers are familiar with transformations by Boolean algebra, the proposed research promotes a design-by-transformation style by providing, for the first time, tools which facilitate experimentation with design transformations and their regulation by meta-programming. These tools will cover design space exploration based on machine learning, and end-to-end tool chains mapping designs captured in multiple source languages to heterogeneous reconfigurable devices targeting cloud computing, Internet-of-Things and supercomputing. The proposed approach will be evaluated through a variety of benchmarks involving hardware acceleration, and through codifying strategies for automating the search of neural architectures for hardware implementation with both high accuracy and high efficiency
Data
No related datasets
Projects
No related projects
Publications
No related publications
Added to Database
15/11/21
