TLPGNN: A Lightweight Two-Level Parallelism Paradigm for Graph Neural Network Computation on GPU.
Qiang FuYuede JiH. Howie HuangPublished in: HPDC (2022)
Keyphrases
- lightweight
- parallel computation
- neural network
- parallel processing
- parallel computing
- parallel computers
- graph theory
- real time
- parallel implementation
- parallel algorithm
- random walk
- shared memory
- graphics processing units
- back propagation
- general purpose
- weighted graph
- network architecture
- artificial neural networks
- communication infrastructure
- subgraph isomorphism
- development environments
- graph structure
- directed graph
- massively parallel
- parallel programming
- cost effective
- wireless sensor networks
- social networks