Compatibility
Current platform, toolchain, and dependency support policy for Hybrid Compute.
Language Versions
- Python: 3.9+ (tested: 3.9, 3.10, 3.11, 3.12, 3.13, 3.14)
- C++: 20 required
- CUDA: 12.0+ optional (tested: 12.6.1, 13.0.1, 13.0.2, 13.1.0, 13.1.1)
- Objective-C++: 17 for macOS Metal support
- Metal: 3.0+ for macOS Metal acceleration
Build Tools
- CMake: 3.10+ required, 3.21+ recommended for CUDA and Metal support
- CUDA Toolkit: 12.0+ optional
- Xcode: 14.0+ for macOS Metal development
- Ninja: 1.10+ recommended
- Make: 4.3+ supported
Core Dependencies
- OpenCV: 4.0+ required, 4.5.0+ recommended, CUDA support optional
- Eigen: 3.4.0+ required
- GTest: 1.12.0+ for unit testing
- pybind11: 2.10.0+ for Python bindings
- NumPy: 1.21.0+ required for the Python interface
Supported Platforms
| Platform | Architecture | GPU | Status |
|---|---|---|---|
| macOS 12.0+ | x86_64, ARM64 | Metal | Active |
| Ubuntu 22.04+ | x86_64 | CUDA 12.0+ | Active |
| Windows 10+ | x86_64 | CUDA 12.0+ | Active |
Linux Details
Ubuntu
- Ubuntu 22.04+ LTS tested on 22.04 and 24.04
- GCC: 11.0+ or Clang: 14.0+
libtbb12(TBB 2021.0+)- CUDA: 12.0+ for GPU support
- OpenCV: 4.0+ with 4.5.0+ recommended
Other Linux Distributions
- GCC: 11.0+ or Clang: 14.0+
libtbb12(TBB 2021.0+)
Optional Dependencies
- TensorRT: 8.5+ for TensorRT acceleration
- ONNX Runtime: 1.13.0+ for ONNX model support
- TensorFlow: 2.12.0+ optional
- PyTorch: 2.0.0+ optional