17388

Fine-Grain Acceleration of Graph Algorithms on a Heterogeneous Chip

Cagri Eryilmaz
The University of Texas at Austin
The University of Texas at Austin, 2017

@phdthesis{eryilmaz2017fine,

   title={Fine-grain acceleration of graph algorithms on a heterogeneous chip},

   author={Eryilmaz, Cagri},

   year={2017}

}

Download Download (PDF)   View View   Source Source   

221

views

With the rise of heterogeneous chips available in the market, where integrated GPU cores and CPU cores reside in the same chip and share a unified memory, it is possible to have better execution schemes for many graph algorithms. Graph algorithms can exhibit producer-consumer behavior, a varying amount of parallelism during execution, and irregularity which results in inefficiency. The inefficiency problem could be solved by exploiting heterogeneity between cores. In this work, I provide an understanding of the executions of some graph algorithms in heterogeneous chips and accelerate their executions by using fine-grain software optimization techniques. To achieve this, I introduce two different fine-grain execution techniques to accelerate the Maximal Independent Set and Preflow-push graph algorithms, and present an evaluation of the techniques on a heterogeneous chip. My techniques, namely Overlapping Threads with Hot-Vertices and Task Switcher, provide 1.3x to 16x speedup over CPU-only execution depending on the input and the algorithm.
Rating: 0.5. From 1 vote.
Please wait...

* * *

* * *

HGPU group © 2010-2017 hgpu.org

All rights belong to the respective authors

Contact us: