GTC ON-DEMAND

 
SEARCH SESSIONS
SEARCH SESSIONS

Search All
 
Refine Results:
 
Year(s)

SOCIAL MEDIA

EMAIL SUBSCRIPTION

 
 

GTC ON-DEMAND

HPC and Supercomputing
Presentation
Media
Optimized Inter-GPU Collective Operations with NCCL
Abstract:

We'll present the functionalities of NCCL (pronounced "Nickel"), a standalone library of standard collective communication routines, such as all-gather, reduce, broadcast, etc., that have been optimized to achieve high bandwidth over GPU topologies. NCCL can be used in either single- or multi-process (for example, MPI) applications.

 
Topics:
HPC and Supercomputing, Artificial Intelligence and Deep Learning
Type:
Talk
Event:
GTC Silicon Valley
Year:
2017
Session ID:
S7155
Download:
Share: