Modeling and optimizing NUMA effects and prefetching with machine learning
Isaac Sánchez Barrera, David Black-Schaffer, Marc Casas, Miquel Moretó, Anastasiia Stupnikova, Mihail Popov
Abstract
Both NUMA thread/data placement and hardware prefetcher configuration have significant impacts on HPC performance. Optimizing both together leads to a large and complex design space that has previously been impractical to explore at runtime.
Topics & Concepts
Computer scienceThread (computing)Parallel computingComputer architectureDistributed computingOperating systemParallel Computing and Optimization TechniquesAdvanced Data Storage TechnologiesInterconnection Networks and Systems