The Community for Technology Leaders
2016 International Conference on Parallel Architecture and Compilation Techniques (PACT) (2016)
Haifa, Israel
Sept. 11, 2016 to Sept. 15, 2016
ISBN: 978-1-5090-5308-7
pp: 57-69
Bruno Bodin , University of Edinburgh, United Kingdom
Luigi Nardi , Imperial College London, United Kingdom
M. Zeeshan Zia , Imperial College London, United Kingdom
Harry Wagstaff , University of Edinburgh, United Kingdom
Govind Sreekar Shenoy , University of Edinburgh, United Kingdom
Murali Emani , Lawrence Livermore National Laboratory, United States of America
John Mawer , University of Manchester, United Kingdom
Christos Kotselidis , University of Manchester, United Kingdom
Andy Nisbet , University of Manchester, United Kingdom
Mikel Lujan , University of Manchester, United Kingdom
Bjorn Franke , University of Edinburgh, United Kingdom
Paul H. J. Kelly , Imperial College London, United Kingdom
Michael O'Boyle , University of Edinburgh, United Kingdom
ABSTRACT
System designers typically use well-studied benchmarks to evaluate and improve new architectures and compilers. We design tomorrow's systems based on yesterday's applications. In this paper we investigate an emerging application, 3D scene understanding, likely to be significant in the mobile space in the near future. Until now, this application could only run in real-time on desktop GPUs. In this work, we examine how it can be mapped to power constrained embedded systems. Key to our approach is the idea of incremental co-design exploration, where optimization choices that concern the domain layer are incrementally explored together with low-level compiler and architecture choices. The goal of this exploration is to reduce execution time while minimizing power and meeting our quality of result objective. As the design space is too large to exhaustively evaluate, we use active learning based on a random forest predictor to find good designs. We show that our approach can, for the first time, achieve dense 3D mapping and tracking in the real-time range within a 1W power budget on a popular embedded device. This is a 4.8× execution time improvement and a 2.8× power reduction compared to the state-of-the-art.
INDEX TERMS
Three-dimensional displays, Simultaneous localization and mapping, Space exploration, Real-time systems, Cameras, Benchmark testing, Mobile communication,embedded systems, design space exploration, DSE, computer vision, SLAM
CITATION
Bruno Bodin, Luigi Nardi, M. Zeeshan Zia, Harry Wagstaff, Govind Sreekar Shenoy, Murali Emani, John Mawer, Christos Kotselidis, Andy Nisbet, Mikel Lujan, Bjorn Franke, Paul H. J. Kelly, Michael O'Boyle, "Integrating algorithmic parameters into benchmarking and design space exploration in 3D scene understanding", 2016 International Conference on Parallel Architecture and Compilation Techniques (PACT), vol. 00, no. , pp. 57-69, 2016, doi:10.1145/2967938.2967963
195 ms
(Ver 3.3 (11022016))