AITuning: Machine Learning-based Tuning Tool for Run-Time Communication Libraries

Abstract

we address the problem of tuning communication libraries by using a deep reinforcement learning approach. Reinforcement learning is a machine learning technique incredibly e ective in solving game-like situations. In fact, tuning a set of parameters in a communication library in order to get be er performance in a parallel application can be expressed as a game: Find the right combination/path that provides the best reward. Even though AITuning has been designed to be utilized with di erent run-time libraries, we focused this work on applying it to the OpenCoarrays run-time communication library, built on top of MPI-3. is work not only shows the potential of using a reinforcement learning algorithm for tuning communication libraries, but also demonstrates how the MPI Tool Information Interface, introduced by the MPI-3 standard, can be used e ectively by run-time libraries to improve the performance without human intervention.