What you will learn

• How to instrument HPC application codes with Score-P
• How to make profile and trace measurements with Scalasca

Prerequisites

• Access to HPC cluster or Linux workstation/laptop
• Development software
  • Compiler suite (C, C++, Fortran): GCC, Intel, IBM XL, PGI, …
  • MPI library: OpenMPI, MPICH, Intel, …
• Access to the source code of the application to analyze
• POP tools Score-P and Scalasca installed

Training module

Material

• Presentation slides
• Jacobi example used in demo

Extended and more detailed example

• BT code from the NAS Parallel Benchmark (NPB)
• A full workflow example from the Scalasca documentation
• Performance Analysis Workflow Using Score-P from Score-P documenation
• Slides from 33rd VI-HPS Tuning Workshop
  • Score-P instrumentation & measurement toolset
  • Score-P analysis scoring & measurement filtering
  • Score-P specialized instrumentation and measurement (Advanced)
  • Scalasca automated trace analysis

Score-P

• Support: support@score-p.org

Scalasca

• Support: scalasca@fz-juelich.de