Install ONNX Runtime
See the installation matrix for recommended instructions for desired combinations of target operating system, hardware, accelerator, and language.
Details on OS versions, compilers, language versions, dependent libraries , etc can be found under Compatibility.
Contents
Inference
The following build variants are available as officially supported packages. Others can be built from source from each release branch.
- Default CPU Provider
- GPU Provider - NVIDIA CUDA
- GPU Provider - DirectML (Windows) - recommended for optimized performance and compatibility with a broad set of GPUs on Windows devices
| Repository | Official build | Nightly build |
|---|---|---|
| Python | If using pip, run pip install --upgrade pip prior to downloading. | |
| CPU: onnxruntime | ort-nightly (dev) | |
| GPU: onnxruntime-gpu | ort-gpu-nightly (dev) | |
| C#/C/C++ | CPU: Microsoft.ML.OnnxRuntime | ort-nightly (dev) |
| GPU: Microsoft.ML.OnnxRuntime.Gpu | ort-nightly (dev) | |
| Java | CPU: com.microsoft.onnxruntime/onnxruntime | |
| GPU: com.microsoft.onnxruntime/onnxruntime_gpu | ||
| nodejs | CPU: onnxruntime |
Note: Dev builds created from the master branch are available for testing newer changes between official releases. Please use these at your own risk. We strongly advise against deploying these to production workloads as support is limited for dev builds.
Requirements
-
All builds require the English language package with
en_US.UTF-8locale. On Linux, install language-pack-en package by runninglocale-gen en_US.UTF-8andupdate-locale LANG=en_US.UTF-8 -
The GPU CUDA build requires installation of compatible CUDA and cuDNN libraries: see CUDA Execution Provider requirements.
-
Windows builds require Visual C++ 2019 runtime.
Training
COMING SOON