# Simulation backends (2D FDTD)

**Simulation backends** run the same numerical model using different hardware and execution strategies. The driver makes the choice explicit:

```bash
# CPU reference implementation.
$ build/out/debug/fdtd --nx 256 --ny 256 --steps 400 --backend serial

# Parallel CPU updates.
$ build/out/debug/fdtd --nx 256 --ny 256 --steps 400 --backend openmp

# GPU updates.
$ build/out/debug/fdtd --nx 256 --ny 256 --steps 400 --backend cuda
```

These commands hold grid dimensions and step count constant. Backend availability depends on the build and hardware.

The [[fdtd-cuda-serial-backend|serial backend]] is the reference. The [[fdtd-cuda-openmp-backend|OpenMP]] and [[fdtd-cuda-cuda-backend|CUDA]] backends provide `v1` through `v4` optimization levels. Select a level with `--variant v2`, for example. Comparisons should also hold precision, boundary conditions, and output settings constant so that timings reflect execution strategy rather than different work.