OC-DNN: Exploiting Advanced Unified Memory Capabilities in CUDA 9 and Volta GPUs for Out-of-Core DNN Training.
Ammar Ahmad AwanChing-Hsiang ChuHari SubramoniXiaoyi LuDhabaleswar K. PandaPublished in: HiPC (2018)
Keyphrases
- training process
- computational power
- training data
- general purpose
- training algorithm
- gpu implementation
- graphics processors
- parallel programming
- computing power
- parallel processing
- training set
- memory requirements
- graphics hardware
- neural network
- training examples
- random access
- parallel computation
- compute intensive
- memory usage
- parallel algorithm
- information processing
- database systems