Workshop curriculum (three half-days)

Assume participants have a cluster login or a strong laptop with MPI + HeFFTe available (or use a prepared container/site module stack). Adjust times for your audience.

Half-day 1 — Build, environment, and first MPI run

Outcomes: configure Release build, run examples/05_simulator, explain world/decomposition at a high level.

Segment

Activity

Primary doc

0:00–0:20

Motivation: PFC, FFT scaling, where OpenPFC sits

README.md (intro), spectral_stack.md

0:20–1:10

Modules, GCC/Open MPI, HeFFTe prefix, cmake

INSTALL.md, dependency_matrix.md

1:10–1:40

Build, mpirun 05_simulator, exit codes

start_here_15_minutes.md, quickstart.md §2A

1:40–2:30

Examples ladder: 02 → 03 → 05 (instructor picks depth)

examples_catalog.md, getting_started/01-basics/README.md

2:30–3:00

Q&A; homework: read architecture.md layers

—

Half-day 2 — JSON app and artifacts

Outcomes: run tungsten with a repo JSON; know where binary fields go; optional VTK path.

Segment

Activity

Primary doc

0:00–0:15

Recap: spectral stack, App vs bare examples/

spectral_stack.md, applications.md

0:15–1:15

mpirun tungsten sample JSON, directory layout

recipes/recipe_spectral_app_json.md, apps/tungsten/inputs_json/README.md

1:15–2:00

Config sections: domain, fields, plan_options skim

spectral_app_config_reference.md, configuration.md

2:00–2:45

Binary I/O contract; optional VTK tutorial path

binary_field_io_spec.md, tutorials/vtk_paraview_workflow.md

2:45–3:00

Q&A

troubleshooting.md

Half-day 3 — HPC and profiling

Outcomes: submit a batch job skeleton; avoid common MPI-IO mistakes; read a profiling export.

Segment

Activity

Primary doc

0:00–0:30

Slurm vs interactive; srun/mpirun

hpc_operator_guide.md, tutorials/hpc_slurm_day_one.md

0:30–1:30

MPI-IO checklist; restarts

mpi_io_layout_checklist.md, operator_playbooks.md

1:30–2:30

Enable profiling in JSON; interpret export

performance_profiling.md, profiling_export_schema.md

2:30–3:00

GPU path decision (if site has GPUs)

gpu_path_decision.md, tutorials/gpu_app_quickstart.md

See also