When Massive GPU Parallelism Ain't Enough: A Novel Hardware Architecture of 2D-LSTM Neural Network.
Vladimir RybalkinJonas NeyMenbere Kina TekleyohannesNorbert WehnPublished in: ACM Trans. Reconfigurable Technol. Syst. (2022)
Keyphrases
- hardware architecture
- neural network
- processing elements
- recurrent neural networks
- parallel computing
- associative memory
- field programmable gate array
- parallel computation
- parallel processing
- hardware implementation
- hardware architectures
- artificial neural networks
- real time
- single instruction multiple data
- neural network model
- back propagation
- parallel implementation
- massively parallel
- parallel algorithm
- pattern recognition
- shared memory
- computational power
- graphics processing units
- parallel architectures
- data analysis
- feed forward
- parallel computers
- feature space