Balanced Column-Wise Block Pruning for Maximizing GPU Parallelism.
Cheonjun ParkMincheol ParkHyun Jae OhMinkyu KimMyung Kuk YoonSuhyun KimWon Woo RoPublished in: AAAI (2023)
Keyphrases
- parallel computation
- parallel processing
- parallel computing
- parallel implementation
- real time
- pairwise
- search space
- memory bandwidth
- gpu implementation
- pruning method
- parallel programming
- graphics hardware
- massively parallel
- graphics processing units
- parallel architectures
- shared memory
- data parallelism
- gpu accelerated
- commodity hardware
- fine grain
- pruning methods
- general purpose
- single instruction multiple data
- fixed size
- block size
- pruning algorithm
- multithreading
- heterogeneous computing