A Simple Method to Reduce Off-chip Memory Accesses on Convolutional Neural Networks.
Doyun KimKyoung-Young KimSangsoo KoSanghyuck HaPublished in: CoRR (2019)
Keyphrases
- convolutional neural networks
- memory access
- low cost
- high speed
- memory requirements
- real time
- random access memory
- convolutional network
- memory usage
- memory subsystem
- memory size
- digital signal processors
- processor core
- level parallelism
- physical design
- single chip
- limited memory
- random access
- access patterns
- high density