Abstract: Evaluating Topology Mapping via Graph Partitioning

Intelligently mapping applications to machine network topologies has been shown to improve performance, but considerable developer effort is required to find good mappings. Techniques from graph partitioning have the potential to automate topology mapping and relieve the developer burden. Graph partitioning is already used for load balancing parallel applications, but can be applied to topology mapping as well. We show performance gains by using a topology-targeting graph partitioner to map sparse matrix-vector and volumetric 3-D FFT kernels onto a 3-D torus network.