SYSTAP Chief Scientist Bryan Thompson, Director of Mission Analytics, Brad Bebee, and Cuda Researcher Zhisong Fu will be presenting “Parallel Breadth First Search on GPU Clusters” in the distributed systems track of this year’s IEEE Bigdata conference. This research, in conjunction with the University of Utah SCI institute, presents results for MapGraph(tm) running on NVIDIA’s Research Cluster with 64 K40 GPUs. It achieves performance of 32,000,000,000 traversed edges per second of a graph with 4,300,000,000 directed edges (32 GTEPS)!
We are at a point with our MapGraph(tm) GPU technology where it has been demonstrated and validated for very high performance graph analytics. We are entering phase in which we are actively seeking Beta customers; missions or customers that have a existing graph analytics that could be dramatically accelerated with our GPU Graph analytics software. This could be for a single GPU if the graph size is in the 100s of millions of edges or for multi-GPUs for larger data sizes.
If you would like to become a Beta customer for MapGraph, we’d love to hear from you. Please contact us at firstname.lastname@example.org.
“Parallel Breadth First Search on GPU Clusters”, Zhisong Fu, Harish Dasari, Martin Berzins, Brad Bebee, and Bryan Thompson