FARGO3D User Guide

Contents:

• Download
  • Clone
• Contributing to the code
• Developers
• Citing FARGO3D
• License
• Introduction
  • A foreword about the terminology used in this manual
• First Steps
  • Getting FARGO3D
  • Installing
  • First run
  • First parallel run
  • First GPU run
  • First Parallel GPU run
• Directory tree
  • Directories
  • setups/SETUP directory
• Make Process
  • Makefile (Stage1)
  • scripts/make.py (Stage2)
  • src/makefile (Stage3)
  • FARGO_ARCH environment variable
• Boundaries
  • How are boundary conditions applied?
  • boundparser.py
  • Boundaries files format
  • centering.txt
  • boundary_template.c
  • boundaries.txt
  • SETUP.bound
  • Common errors
  • Shearing-periodic boundaries
  • Defining boundaries by hand
• Mesh and Fields
  • Mesh
  • Fields
  • Working with fields
  • Fields on the GPU
  • Useful variables
• Default SETUPS
  • fargo
  • Orszag-Tang Vortex
  • Sod shock tube 1D
  • MRI
• .opt files
• Non-uniform meshes
  • Mesh-density functions implemented
  • Implementing a new mesh-density function
• N-Body solver
  • Introduction
  • Removing the default central star
  • Planets which are stars
  • Initialization of the position and velocities of the stellar objects
  • Indirect term
  • Initial eccentricity and inclination
  • Convenience parameters: PLANETMASS, SEMIMAJORAXIS and ORBITALRADIUS
  • Correcting the shift of resonances
• Units
  • Introduction
  • Specifying the unit system
  • Rescaling the input parameters
  • Specifying the scaling rules
• Multifluid
  • Setups
  • New rules for the .opt file
  • Creation of new fluids
  • Drag force between fluids
  • Dust diffusion module
  • Boundaries
  • Update setups compatible with versions < 2.0
• Induction equation
  • Diffusivities
• Defining a new SETUP
  • Blob test
  • Initial state:
  • Making the executable:
  • Plotting your new setup:
• Run-time visualization.
  • How does it work?
  • Backends
• Outputs
  • Scalar fields
  • Summary files
  • Domain files
  • Variables
  • Grid files
  • Planet files
  • Datacubes
  • MPI INPUT/OUTPUT
  • Monitoring
• Communications
  • GPU/CPU communications
  • MPI
  • MPI-CUDA
  • Spawning a job on a cluster of GPUs: a primer
• Cuda translator (C2CUDA.py)
  • FARGO3D Mesh Functions
  • How the script works
  • Common errors:
• Execution flags
• VTK Compatibility
• Improving CUDA Performance
  • MPICUDA
• GPU vs CPU Benchmarking
  • Some acceleration factors
  • The FARGO_SPEEDUP macrocommand
• Developing a complex setup
  • Setup folder
  • Initial Condition
  • Time dependent explosions
  • Incorporating our kernel
  • Using FARGO_DEBUG
  • Summary
• Tips, Tricks, Todos and Troubleshooting
  • How do I add a new parameter to a setup ?
  • How do I create a new Field variable?
  • How do I write a Field?
  • What is the way to merge the outputs of a run spawned with the -k flag ?
  • I see lots of “!” or “:” instead of “.” at execution. What does that mean?
  • Where my last run data has been output ?
  • My build produces unexpected results. Some files should have been remade and they have not. How do I fix this ?
  • My code does not run much faster on the GPU than on the CPU. Why is this ?
  • What is this @ sign at the beginning of the outputs’ path ?
  • I see that there are .par files in each setup directory, and the same .par files are found in the in/ sub directory. What is this for ?
  • I have noticed a directory named .fargo3drc in my home directory. What is it here for ?
  • How can I see the output of python scripts (in particular the CUDA files) ?
  • Tweaking the CUDA executable for memory optimization
  • Speeding up reduction operations on platform >= 3.5
  • A very incomplete TODO list