-
Notifications
You must be signed in to change notification settings - Fork 3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Build] Fail building OnnxRuntime with Cuda 11.2 #17961
Comments
|
I will close it as we do not support CUDA 11.2 anymore. |
@snnn I archives installing the last version of onnxruntime downloaded from github with cuda 11.2 in a docker container. It is the same version that failed me in my post. When using docker, it worked. I think because it is what you said, a conda environment may have extra dependencies. What I have done next is to use the python wheel generated inside the docker container and install it in my conda environment. All works. However, when importing onnxruntime I get this error:
I have glibc 2.27, but I wonder if this error is directly because my version is old or it has a relationship with other dependencies, because when I used old onnxruntime versions like 1.4 or 1.12, this error didn't appear. |
Describe the issue
Hi everyone, I'm trying to build onnxruntime for On-Device Training. To do so, I'm following the guidelines written in the tutorial: https://onnxruntime.ai/docs/build/training.html . I also cloned the main branch of the GitHub repo yesterday, so it is recent and the build script it is the one currently saved in the GitHub Repo.
However, my environment is not equal. I'm building onnxruntime with ubuntu 20.04 together with cuda 11.2 and cudnn 8. Cmake version is 3.27 and python version is 3.8 in a conda environment. I have some troubles with the installation
The shell command I use is: ./build.sh --config RelWithDebInfo --build_shared_lib --parallel --enable_training --allow_running_as_root --build_wheel --use_cuda --cuda_home /usr/local/cuda-11.2/ --cudnn_home /usr/local/cuda-11.2/ --cuda_version=11.2 --skip_tests
And the following commands for imports:
export CUDA_HOME=/usr/local/cuda-11.2/
export CUDNN_HOME=/usr/local/cuda-11.2/
export CUDACXX=/usr/local/cuda-11.2/bin/nvcc
And the error is the following:
I would like to know where is the error.
Thank you.
Urgency
No response
Target platform
Ubuntu 20.04
Build script
#!/bin/bash
Error / output
Visual Studio Version
No response
GCC / Compiler Version
9.5.0
The text was updated successfully, but these errors were encountered: