Illegal instruction when importing onnxruntime #23957
Assignees
Labels
quantization
issues related to quantization
Comments
|
I ran into this too. Seems like a regression with If it helps anyone, this is the strace: |
Thank you. The previous version works well ! |
|
I uploaded a debug version to https://aiinfra.visualstudio.com/PublicPackages/_artifacts/feed/ORT-Nightly/PyPI/onnxruntime/overview/1.21.0 , could you please help me test it? python3 -m pip install coloredlogs flatbuffers numpy packaging protobuf sympy
python3 -m pip install -i https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/ORT-Nightly/pypi/simple/ onnxruntime==1.21.0Then when please run your program with gdb, and send me the stacktrace. |
|
The new package is about 440MB large. |
|
I have a Raspberry PI 4, but it needs an HDMI adapter. I search all over my house but couldn't find one. |
|
No problem. Looks like the version you uploaded works well |
|
I found a device that can reproduce the issue. |
|
Here is the stacktrace: |
|
It's a fp16 instruction I believe |
|
Most likely it was because of PR #23597 |
|
here is the fix #23978 |
I'm sorry for the late reply. I missed these emails from github somehow. |
|
I updated a new version to the nightly feed: The version number is 1.22.0.dev20250310006. You may install it by using the following command: The package was built from @fajin-corp 's dev branch with the bug fix. @feiticeir0 / @mikeesto , would you please help verify it? |
I tried to install it, but it asks for a user for a repository when trying to install numpy (I know because I've canceled it and it's the error):
|
|
Please run before running that installation command |
It works. I can import it without any errors. |
|
Thanks |
NeonDaniel
added a commit
to NeonGeckoCom/NeonCore
that referenced
this issue
Mar 12, 2025
# Description Update core_modules to validate latest alpha version compat. # Issues Blacklists an onnxruntime version to mitigate microsoft/onnxruntime#23957 # Other Notes Based on [this run from an old tag failing](https://github.com/NeonGeckoCom/NeonCore/actions/runs/13816049514/job/38649389777), it appears that the failures are related to some change in GitHub Actions runners or some low-level dependency change. Cause for Pi image failures is unknown, but updating to a newer CPU arch in the shared action appears to resolve the issue. Note that the previously spec'd A53 was used in the RPi3 and the now spec'd A76 is used in the RPi5 Issue was traced back to a new onnxruntime release
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Describe the issue
In a fresh Raspberry PI OS Bookworm 64 bits install on a Raspberry PI 4, after installing onnxruntime using pip in a VirtualEnvironment, when importing it, i get:
Here's the system specs:
Raspberry PI OS:
Python Version:
ONNX Runtime version installed by pip:
onnxruntime-1.21.0-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whlTo reproduce
Urgency
I'm not saying it's urgent, but it's definitely an issue. Raspberry PIs are used at the edge all the time.
Platform
Linux
OS Version
Debian GNU/Linux 12 (bookworm)
ONNX Runtime Installation
Released Package
ONNX Runtime Version or Commit ID
onnxruntime-1.21.0-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl
ONNX Runtime API
Python
Architecture
ARM64
Execution Provider
Default CPU
Execution Provider Library Version
No response
The text was updated successfully, but these errors were encountered: