DNNVM : End-to-End Compiler Leveraging Heterogeneous Optimizations on FPGA-based CNN Accelerators.
Yu XingShuang LiangLingzhi SuiXijie JiaJiantao QiuXin LiuYushun WangYu WangYi ShanPublished in: CoRR (2019)
Keyphrases
- end to end
- field programmable gate array
- hardware implementation
- admission control
- cellular neural networks
- computing systems
- multipath
- transactional memory
- high bandwidth
- internet protocol
- general purpose
- congestion control
- wireless ad hoc networks
- application specific
- ad hoc networks
- content delivery
- real world
- rate adaptation
- text localization and recognition
- rate allocation
- application layer
- parallel computing
- video sequences