FLUME — FNL (OpenACC) Backend

The FNL backend (flume-fnl) is the GPU implementation of FLUME on the FNL library (OpenACC directives, each with its OpenMP offload twin), extending the backend-independent infrastructure of common/. The fields live on the device; the host copies are refreshed for output, restart and the coarse-fine seam fluxes of the reflux register.

Status: development (milestones M1 to M13 complete, the list in docs/applications/flume/index.md). The seam face fluxes of a rank go to the host register in one copy per stage (issue #90).

Source Files

File	Description
adam_flume_fnl.F90	Entry point (program)
adam_flume_fnl_object.F90	flume_fnl_object: device data, ghost update, residual, Runge-Kutta integration, reflux, output
adam_flume_fnl_dissipation_kernels.F90	Device twins of the CPU viscous, heat and Ohmic face-flux kernels (issue #65)
adam_flume_fnl_kernels.F90	Device kernels: face fluxes and flux difference (with the immersed-boundary variant), boundary conditions, time-step and conservation reductions, auxiliary fields, seam skin packing and reflux application, radiative cooling and the cooling time (issue #93)
adam_flume_fnl_<model>[_<solver>]_kernels.F90	Device twins of the CPU per-model and per-Riemann-solver kernel modules (same model and solver names)
adam_flume_fnl_*_kernels_agnostic.INC	The model-agnostic device kernel bodies the model modules include
adam_flume_fnl_library.F90	Barrel re-export of all FNL modules

Build modes

fobis build --mode flume-fnl-nvf --varset local_nvf          # NVIDIA, OpenACC (verified)
fobis build --mode flume-fnl-nvf-debug --varset local_nvf    # NVIDIA, OpenACC, runtime checks
fobis build --mode flume-fnl-omp-amd                          # AMD, OpenMP offload (not verified)