No onnxruntime_gpu in sbsa/cu130

chao.w · September 17, 2025, 1:56am

no onnxruntime_gpu find in https://pypi.jetson-ai-lab.io/sbsa/cu130, can I use https://pypi.jetson-ai-lab.io/jp6/cu129/+f/2e3/a07114007df15/onnxruntime_gpu-1.23.0-cp312-cp312-linux_aarch64.whl?

DavidDDD · September 17, 2025, 8:47am

Hi,

We will check with our developer team.

https://pypi.jetson-ai-lab.io/jp6/cu129/+f/2e3/a07114007df15/onnxruntime_gpu-1.23.0-cp312-cp312-linux_aarch64.whl is not compatible for jetson thor.

Thanks

johnny_nv · September 17, 2025, 11:00am

Hello,
onnxruntime is not compatible with cuda 13 yet. Microsoft members are working on it

github.com/microsoft/onnxruntime

[Build] CUDA 13 Failed

opened 03:08PM - 03 Sep 25 UTC

johnnynunez

build ep:CUDA

### Describe the issue It has a lot of errors related to flash attention must …be upstream and apply this fix https://github.com/Dao-AILab/flash-attention/pull/1860 ### Urgency _No response_ ### Target platform Blackwell CUDA 13 ### Build script ```bash #!/usr/bin/env bash set -ex echo "Building onnxruntime ${ONNXRUNTIME_VERSION} (branch=${ONNXRUNTIME_BRANCH}, flags=${ONNXRUNTIME_FLAGS})" # Check TensorRT installation if [ ! -d "/usr/lib/$(uname -m)-linux-gnu" ]; then echo "TensorRT libraries not found in /usr/lib/$(uname -m)-linux-gnu" exit 1 fi # Ensure TensorRT libraries are in LD_LIBRARY_PATH export LD_LIBRARY_PATH=/usr/lib/$(uname -m)-linux-gnu:$LD_LIBRARY_PATH # Verify TensorRT components if [ ! -f "/usr/lib/$(uname -m)-linux-gnu/libnvinfer.so" ]; then echo "TensorRT core library not found" exit 1 fi if [ ! -f "/usr/lib/$(uname -m)-linux-gnu/libnvdla_compiler.so" ]; then echo "TensorRT NVDLA compiler library not found" fi pip3 uninstall -y onnxruntime || echo "onnxruntime was not previously installed" git clone https://github.com/microsoft/onnxruntime /opt/onnxruntime cd /opt/onnxruntime git checkout ${ONNXRUNTIME_BRANCH} git submodule update --init --recursive install_dir="/opt/onnxruntime/install" pip3 install "cmake<4" ./build.sh --config Release --update --parallel --build --build_wheel --build_shared_lib \ --skip_tests --skip_submodule_sync ${ONNXRUNTIME_FLAGS} \ --cmake_extra_defines CMAKE_CXX_FLAGS="-Wno-unused-variable -I/usr/local/cuda/include" \ --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES="${CUDA_ARCHITECTURES}" \ --cmake_extra_defines CMAKE_INSTALL_PREFIX=${install_dir} \ --cmake_extra_defines onnxruntime_BUILD_UNIT_TESTS=OFF \ --cuda_home /usr/local/cuda --cudnn_home /usr/lib/$(uname -m)-linux-gnu \ --use_tensorrt --tensorrt_home /usr/lib/$(uname -m)-linux-gnu cd build/Linux/Release make install ls -ll dist cp dist/onnxruntime*.whl /opt cd / pip3 install /opt/onnxruntime*.whl python3 -c 'import onnxruntime; print(onnxruntime.__version__);' twine upload --verbose /opt/onnxruntime*.whl || echo "failed to upload wheel to ${TWINE_REPOSITORY_URL}" tarpack upload onnxruntime-gpu-${ONNXRUNTIME_VERSION} ${install_dir} || echo "failed to upload tarball" cd ${install_dir} cp -r * /usr/local/ ls #rm -rf /tmp/onnxruntime ``` ### Error / output [10o10_onnxruntime_r38.2.aarch64-cu130-24.04-onnxruntime.txt](https://github.com/user-attachments/files/22122227/10o10_onnxruntime_r38.2.aarch64-cu130-24.04-onnxruntime.txt) ### Visual Studio Version _No response_ ### GCC / Compiler Version _No response_

johnny_nv · October 13, 2025, 9:37pm

Hello, you can now build onnxruntime

system · October 27, 2025, 9:38pm

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.

Topic		Replies	Views
Still no onnxruntime_gpu in sbsa/cu130 Jetson Thor onnx	1	23	January 21, 2026
Method to compile onnxruntime-gpu on Thor Jetson Thor onnx	5	405	December 1, 2025
Onnx runtime GPU Jetson Orin Nano onnx	6	3695	March 24, 2025
Can you add GPU SM110 (Jetson Thor) support to the tensorrt:25.06-py3 container? NGC Catalog Feedback tensorrt , cuda , onnx	0	30	January 22, 2026
Issue running ONNXRUNTIME in Jetson Jetson TX2 onnx	5	1670	February 9, 2022
I want to use onnxruntime-gpu for my NVIDIA Jetson Nano 2Gb Kit GPU-Accelerated Libraries jetson	8	523	October 31, 2024
ONNXRUNTIME_GPU install or build fails Jetson Xavier NX onnx	4	2120	October 13, 2022
Build ONNXInference-gpu wheel for Jetpack5 with Cuda and TRT Jetson AGX Orin tensorrt , cuda , onnx	6	2888	August 10, 2022
Onnxruntime link for 1.20 Jetson Orin NX cuda , onnx	3	167	October 9, 2025
Onnxruntime-gpu for jetson Jetson AGX Orin onnx	9	450	November 4, 2025

No onnxruntime_gpu in sbsa/cu130

Related topics