MSHGN: Multi-scenario adaptive hierarchical spatial graph convolution network for GPU utilization prediction in heterogeneous GPU clusters.
Sheng WangShiping ChenFei MengYumei ShiPublished in: J. Parallel Distributed Comput. (2024)
Keyphrases
- real time
- graphics hardware
- prediction accuracy
- parallel implementation
- graphics processors
- spatio temporal
- data transfer
- parallel processing
- real world
- graph clustering
- parallel computing
- hierarchical structure
- heterogeneous computing
- gpu implementation
- graphics processing units
- graphical representation
- social networks
- spatial networks
- hierarchical clustering
- link prediction
- wireless sensor networks
- network structure
- agglomerative clustering
- heterogeneous networks
- spanning tree
- graph theory
- random walk
- spatial information
- clustering algorithm
- peer to peer
- self organizing maps
- fully connected
- proximity graph
- densely connected
- complex networks
- graph structure
- community structure
- strongly connected
- resource utilization
- bipartite graph
- path length
- multiple types