An In-Network Parameter Aggregation using DPDK for Multi-GPU Deep Learning

Mathematics

• Aggregation 58%
• Learning 52%
• Switch 35%
• Communication 30%
• Ethernet 29%
• Gradient 19%
• Training 17%
• Express 14%
• Distributed Networks 12%
• Latency 10%
• Workload 10%
• Vertex of a graph 9%
• Standards 9%
• Neural Networks 7%
• Resources 7%
• Evaluate 6%
• Evaluation 6%

Engineering & Materials Science

• Graphics processing unit 54%
• Deep learning 51%
• Agglomeration 50%
• Switches 23%
• Ethernet 16%
• Communication 16%
• Deep neural networks 8%