Crate constraint_theory_core_cuda 

constraint-theory-core-cuda

CUDA FFI bridge connecting the constraint-theory-core Rust CSP solver to GPU-accelerated parallel constraint solving via flux-cuda kernels.

Target hardware: JetsonClaw1 (Jetson Xavier NX, 512 CUDA cores, 8GB unified)

Architecture

Rust CSP formulation (constraint-theory-core)
        │
    FFI boundary (this crate)
        │
CUDA kernels (flux-cuda)
    ├── flux_vm_kernel — parallel FLUX VM execution
    ├── csp_solver_kernel — parallel backtracking + AC-3
    └── sonar_physics_kernel — batch Mackenzie 1981 / Francois-Garrison 1982

Re-exports

pub use cuda_ffi::CspProblem;

pub use cuda_ffi::CspSolution;

pub use cuda_ffi::CudaError;

pub use cuda_ffi::GpuCspSolver;

pub use cuda_ffi::SonarResult;

pub use cuda_ffi::LinearConstraint;

Modules

cuda_ffi

CUDA FFI Bridge for constraint-theory-core