Performance with layout and stencil

Pierre Aubert




Aim of the lecture



This lecture will show how to evaluate performance and how to optimise a simulation program based on a computing stencil.

Prerequisites

  • Compiler : G++11 or a newer version.
  • Compilation tools : CMake et Make
  • Plot results : Gnuplot
  • Versionning : Git


Following lecture with containers

This lecture is available within containers compatible with Apptainer, Docker, Podman, Kubernetes or Docker-Swarm
  • Light image 143 MB OK : docker://gitlab-registry.in2p3.fr/cta-lapp/cours/performancewithlayoutandstencil/performancewithlayoutandstencil_alpine_light:latest
  • Code server 325 MB OK : docker://gitlab-registry.in2p3.fr/cta-lapp/cours/performancewithlayoutandstencil/performancewithlayoutandstencil_alpine_micromamba_code_server:latest
  • Jupyter-hub OK : docker://gitlab-registry.in2p3.fr/cta-lapp/cours/performancewithlayoutandstencil/performancewithlayoutandstencil_alpine_micromamba_vscode:latest


Their use is explained in the lecture demo.

Correction

Full correction is avaiable here.