Crossbar based Processing in Memory Accelerator Architecture for Graph Convolutional Networks.
Nagadastagiri ChallapalleKarthik SwaminathanNandhini ChandramoorthyVijaykrishnan NarayananPublished in: ICCAD (2021)
Keyphrases
- processing elements
- memory management
- compute intensive
- network architecture
- real time
- parallel architecture
- computational power
- distributed processing
- parallel computers
- parallel implementation
- random access
- small world
- data processing
- information processing
- directed graph
- memory requirements
- random walk
- average degree
- fully connected
- hardware implementation
- network analysis
- graph structure
- social networks
- management system
- network structure
- neural network
- betweenness centrality
- level parallelism
- memory access
- structured data
- graph structures
- processing units
- connected components
- associative memory
- bipartite graph
- graph theory
- field programmable gate array
- operating system
- parallel processors
- edge weights
- hardware architecture
- query processing
- bayesian networks
- massively parallel
- graph representation
- community detection