The International Conference for High Performance Computing, Networking, Storage, and Analysis
The International Conference for High Performance Computing, Networking, Storage, and Analysis
Workshop:IA^3 2021: 11th Workshops on Irregular Applications: Architectures and Algorithms
Authors: Umit Catalyurek (Georgia Institute of Technology, Amazon)
Abstract: Designing flexible graph kernels that can run well on various platforms is a crucial research problem due to the frequent usage of graphs for modeling data and recent architectural advances and variety. In this talk, I will present our recent graph processing model and framework, PGAbB, for modern shared-memory heterogeneous platforms. PGAbB implements a block-based programming model. This allows a user to express a graph algorithm using functors that operate on an ordered list of blocks (subgraphs). Our framework deploys these computations to all available resources in a heterogeneous architecture. We will demonstrate that one can implement a diverse set of graph algorithms in our framework, and task-based execution enables graph computations even on large graphs that do not fit in GPU device memory. Our experimental results show that PGAbB achieves competitive or superior performance compared to hand-optimized implementations or existing state-of-the-art graph computing frameworks.