C++ example¶

devices.cpp lists the compute devices (CPU, CUDA, OpenCL), self-tests them (--selftest) and runs a system on the fastest suitable one (uf_devices_json, uf_set_default_backend("auto"); docs/plugins/gpu.md).

run_system.cpp uses the C API (src/UltrafastEngine.Native/include/ultrafast.h): load a system and run spec, register a C++ time-response callback and a progress callback, run, read the summary JSON and an array, write the results. The same API is usable from C, Fortran (ISO_C_BINDING), MATLAB (loadlibrary), LabVIEW, Julia...

Build the engine as a native library first (needs the .NET 10 SDK and, on Windows, the VS "Desktop development with C++" workload):

dotnet publish src/UltrafastEngine.Native -c Release -r win-x64 -o artifacts/native/win-x64
cmake -S docs/examples/Interop/cpp -B build/cpp -DULTRAFAST_DIR=artifacts/native/win-x64
cmake --build build/cpp --config Release
build/cpp/Release/run_system docs/examples/GettingStarted/spm-plate/plate.ufs docs/examples/GettingStarted/spm-plate/run.json out