14.5 Envision: A 0.26-to-10TOPS/W subword-parallel dynamic-voltage-accuracy-frequency-scalable Convolutional Neural Network processor in 28nm FDSOI.
Bert MoonsRoel UytterhoevenWim DehaeneMarian VerhelstPublished in: ISSCC (2017)
Keyphrases
- convolutional neural network
- parallel processing
- high accuracy
- distributed memory
- multi processor
- single processor
- computer architecture
- multi core processors
- multiprocessor systems
- face detection
- neural network
- parallel programming
- high end
- parallel processors
- parallel architectures
- metal oxide
- power supply
- data parallelism
- graphics processing units
- dynamic environments
- feature space