18761

Optimizing Network Performance for Distributed DNN Training on GPU Clusters: ImageNet/AlexNet Training in 1.5 Minutes

Peng Sun, Wansen Feng, Ruobing Han, Shengen Yan, Yonggang Wen
SenseTime Research
arXiv:1902.06855 [cs.DC], (19 Feb 2019)

@article{sun2019optimizing,

   title={Optimizing Network Performance for Distributed DNN Training on GPU Clusters: ImageNet/AlexNet Training in 1.5 Minutes},

   author={Sun, Peng and Feng, Wansen and Han, Ruobing and Yan, Shengen and Wen, Yonggang},

   year={2019},

   month={feb},

   archivePrefix={"arXiv"},

   primaryClass={cs.DC}

}

Download Download (PDF)   View View   Source Source   

316

views

It is important to scale out deep neural network (DNN) training for reducing model training time. The high communication overhead is one of the major performance bottlenecks for distributed DNN training across multiple GPUs. Our investigations have shown that popular open-source DNN systems could only achieve 2.5 speedup ratio on 64 GPUs connected by 56 Gbps network. To address this problem, we propose a communication backend named GradientFlow for distributed DNN training, and employ a set of network optimization techniques. First, we integrate ring-based allreduce, mixed-precision training, and computation/communication overlap into GradientFlow. Second, we propose lazy allreduce to improve network throughput by fusing multiple communication operations into a single one, and design coarse-grained sparse communication to reduce network traffic by only transmitting important gradient chunks. When training ImageNet/AlexNet on 512 GPUs, our approach achieves 410.2 speedup ratio and completes 95-epoch training in 1.5 minutes, which outperforms existing approaches.
Rating: 2.0/5. From 1 vote.
Please wait...

* * *

* * *

HGPU group © 2010-2019 hgpu.org

All rights belong to the respective authors

Contact us: