| Age | Commit message (Collapse) | Author | |
|---|---|---|---|
| 2025-12-05 | Use python-ninja (ninja package doesn't provide Python bindings) | Prof. Andy Couturier | |
| 2025-12-05 | Revert to ninja (Arch's ninja package includes Python bindings) | Prof. Andy Couturier | |
| 2025-12-05 | Fix ninja dependency: use python-ninja instead of ninja | Prof. Andy Couturier | |
| 2025-12-05 | Bump pkgrel to 2 (add ninja makedepends) | Prof. Andy Couturier | |
| 2025-12-05 | Add ninja to makedepends | Prof. Andy Couturier | |
| 2025-12-05 | Update to 2.9.0 with AUR guideline compliance fixes | Prof. Andy Couturier | |
| - Remove aarch64 support, x86_64 only - Add fix-glog-0.7.patch for glog 0.7.x compatibility - Add missing dependencies: python-tensorrt, python-rich, python-nvidia-modelopt, python-torchvision, pybind11 - Include Bazel 8.1.1 in source arrays with proper checksums (no network downloads during build) - Use modern PEP 517 build system (python -m build, python -m installer) - Relax pybind11 version requirement to work with Arch's pybind11 3.0.1 - Use system TensorRT libraries instead of downloading - Add proper SHA256 checksums for all sources | |||
| 2025-12-04 | Initial commit: python-torch-tensorrt 2.9.0 | Prof. Andy Couturier | |
| PyTorch/TorchScript/FX compiler for NVIDIA GPUs using TensorRT. Requires CUDA and TensorRT to build and run. | |||