Journal
IEEE TRANSACTIONS ON COMPUTERS
Volume 70, Issue 8, Pages 1160-1174Publisher
IEEE COMPUTER SOC
DOI: 10.1109/TC.2021.3059962
Keywords
Hardware; Computer architecture; Analytical models; Search engines; Neural networks; Space exploration; Search problems; DNN; accelerator; design space exploration; memory hierarchy; mapping; scheduling; dataflow; analytical model
Funding
- Flemish Government (AI Research Program)
- Fund For Scientific Research Flanders (FWO-Vlaanderen) [G006718N]
- EU [ERC-2016STG-715037]
Ask authors/readers for more resources
ZigZag is a rapid design space exploration framework for embedding deep learning systems, introducing uneven mapping opportunities and smart mapping search strategies that lead to more energy-efficient solutions.
Building efficient embedded deep learning systems requires a tight co-design between DNN algorithms, hardware, and algorithm-to-hardware mapping, a.k.a. dataflow. However, owing to the large joint design space, finding an optimal solution through physical implementation becomes infeasible. To tackle this problem, several design space exploration (DSE) frameworks have emerged recently, yet they either suffer from long runtimes or a limited exploration space. This article introduces ZigZag, a rapid DSE framework for DNN accelerator architecture and mapping. ZigZag extends the common DSE with uneven mapping opportunities and smart mapping search strategies. Uneven mapping decouples operands (W/I/O), memory hierarchy, and mappings (temporal/spatial), opening up a whole new space for DSE, and thus better design points are found by ZigZag compared to other SotAs. For this, ZigZag uses an enhanced nested-for-loop format as a uniform representation to integrate algorithm, accelerator, and algorithm-to-accelerator mapping. ZigZag consists of three key components: 1) an analytical energy-performance-area Hardware Cost Estimator, 2) two Mapping Search Engines that support spatial and temporal even/uneven mapping search, and 3) an Architecture Generator that auto-explores the wide memory hierarchy design space. Benchmarking experiments against published works, in-house accelerator, and existing DSE frameworks, together with three case studies, show the reliability and capability of ZigZag. Up to 64 percent more energy-efficient solutions are found compared to other SotAs, due to ZigZag's uneven mapping capabilities.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available