Opara: Exploiting Operator Parallelism for Expediting DNN Inference on GPUs.
Aodong ChenFei XuLi HanYuan DongLi ChenZhi ZhouFangming LiuPublished in: CoRR (2023)
Keyphrases
- parallel processing
- computational power
- inference process
- parallel architectures
- multi core systems
- general purpose
- commodity hardware
- bayesian inference
- bayesian networks
- knowledge representation
- belief networks
- training process
- single instruction multiple data
- neural network
- inference mechanism
- bayesian model
- probabilistic inference
- graphical models
- database systems