A survey on graph processing on GPUs

PhD Qualifying Examination


Title: "A survey on graph processing on GPUs"

by

Mr. Da YAN


Abstract:

Many practical computing problems concern large graphs. Examples include 
web graph, various social networks. The growing scale of graphs poses 
challenges to their efficient processing. Equipped with thousands of ALUs 
and high bandwidth memory, GPUs may be the ideal device to tackle this 
problem.

However, it is non-trivial to map graph processing algorithms on GPU 
hardware. The main reason is that GPUs are designed for dense and regular 
data processing, but real-world graphs are sparse and irregular. To bridge 
the gap between real-world graphs and GPU hardware, researchers have 
proposed many methods to make the memory access regular and to balance the 
load between different GPU cores.

In this survey, we first introduce GPU hardware and software stack, then 
some hardwired graph algorithm implementations on GPU. Finally, we 
introduce some popular high-level GPU graph processing frameworks.


Date:			Tuesday, 7 May 2019

Time:                  	4:00pm - 6:00pm

Venue:                  Room 4472
                         Lifts 25/26

Committee Members:	Dr. Wei Wang (Supervisor)
 			Prof. Bo Li (Chairperson)
 			Dr. Qiong Luo
 			Dr. Ke Yi


**** ALL are Welcome ****