Supervisors

Suitable for

Abstract

The size of real-life graphs has increased significantly where the need to process them efficiently, especially for graph pattern queries, has become very important. Previously designed algorithms and graph storage/indexing structures with a system named "BitMat" are very conducive for modern GPUs, and can provide leaps and bound improvement in the query performance. The project as a scope for lot of novel query processing algorithm ideas specifically targeted for massive parallelism provided by GPUs. For more information on the BitMat system and previous papers, please see https://www.cs.ox.ac.uk/people/medha.atre/projects.html

Prerequisites: Knowledge of graphs, knowledge of programming (GPU knowledge not required)