Understanding the Performance of HPC Applications

Date and time: 
Wed, Mar 13 2019 - 11:00am to Thu, Mar 14 2019 - 10:45am
260 Deschutes
Brian Gravelle
University of Oregon
  • Boyana Norris (Chair)
  • Hank Childs
  • Allen Malony

High performance computing is an important asset to scientific research, enabling the study of phenomena,  such  as  nuclear  physics  or  climate  change,  that  are  difficult  or  impossible  to  be  studied  in traditional experiments or allowing researchers to utilize large amounts of data from experiments such as the Large Hadron Collider.  No matter the use of HPC, the need for performance is always present; however, the fast-changing nature of computer systems means that software must be continually updated to run efficiently on the newest machines.  In this paper, we discuss methods and tools used to understand the performance of an application running on HPC systems and how this understanding can translate into improved performance.  We primarily focus on node-level issues, but also mention some of the basic issues involved with multi-node analysis as well.