Performance tuning is critical for applications in a world where computing cycles are a precious resource. The process of tuning and optimizing a code, however, remains enshrouded in mystery for many HPC practitioners. This is in part due to the many tools available to profile codes, and also to the sheer amount of data produced by the profiles, which can feel overwhelming for those without previous profiling experience.
This tutorial is designed to provide an introduction to the profiling and tuning process with Intel® Advisor and VTune™ Amplifier. A step-by-step approach for the analysis will be followed using two different applications in order to showcase HPC specific capabilities, including the analysis of Python code and the ease of use of these tools. Submission scripts customized for Theta, and easily modified for other systems, will be provided as part of the presentation.