Improving the Performance of Distributed MXNet with RDMA

As one of the most influential deep learning frameworks, MXNet has achieved excellent performance and many breakthroughs in academic and industrial fields for various machine learning situations. The initial implementation of MXNet uses proxy-socket interface, which delivers suboptimal performance i...

Ausführliche Beschreibung

Bibliographische Detailangaben

Veröffentlicht in:	International journal of parallel programming 2019-06, Vol.47 (3), p.467-480
Hauptverfasser:	Li, Mingfan, Wen, Ke, Lin, Han, Jin, Xu, Wu, Zheng, An, Hong, Chi, Mengxian
Format:	Artikel
Sprache:	eng
Schlagworte:	Computer Science Domains Ethernet Machine learning Parameters Processor Architectures Software Engineering/Programming and Operating Systems Source code Subsystems Theory of Computation Training
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!