AWB-GCN uses three hardware runtime rebalancing techniques (distribution smoothing, remote switching, and evil row remapping) to raise processing-element utilization for irregular GCN workloads.
Understanding reuse, performance, and hardware cost of dnn dataflow: A data-centric approach,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
AWB-GCN: A Graph Convolutional Network Accelerator with Runtime Workload Rebalancing
AWB-GCN uses three hardware runtime rebalancing techniques (distribution smoothing, remote switching, and evil row remapping) to raise processing-element utilization for irregular GCN workloads.