Accel-GCN: High-Performance GPU Accelerator Design for Graph Convolution Networks

Keyphrases

• High Performance 100%
• GPU Accelerators 100%
• Graph Convolutional Network 100%
• Accelerator Design 100%
• Block Level 20%
• Memory Coalescing 20%
• Memory Access 10%
• Memory Bandwidth 10%
• Computational Efficiency 10%
• Dynamically Adjust 10%
• Partitioning Strategy 10%
• Load Balancing 10%
• Effective Solutions 10%
• Working Memory Load 10%
• Shared Memory 10%
• Graph Structure 10%
• Dense Matrix 10%
• Memory Locality 10%
• Multilevel Memory 10%
• Computational Parallelism 10%
• Latent Information 10%
• GitHub 10%
• Column Dimension 10%
• Accelerator Architectures 10%
• Workload Imbalance 10%
• SpMM 10%
• CUSPARSE 10%
• Memory Alignment 10%
• Benchmark Graphs 10%

Computer Science

• Graph Convolutional Network 100%
• Graphics Processing Unit 100%
• Parallelism 10%
• Shared Memories 10%
• Memory Access 10%
• Multilevel Memory 10%
• Memory Bandwidth 10%
• Computational Efficiency 10%
• Effective Solution 10%
• Level Partition 10%