intel
OSError: libmkl_intel_lp64.so.2 not found after pip install on ChromeOS
Describe the bug
My Goal: I am trying to install and run ComfyUI on my Chromebook, which has an integrated Intel GPU.
My System:
- Device: Acer Chromebook Spin 713 (codename: kohaku)
- ChromeOS Version: 140.16371.0
- Linux Kernel: (In your Linux terminal, run the command
uname -aand paste the output here)
The Problem: When I try to run my application after a clean install, I get the following error:
OSError: libmkl_intel_lp64.so.2: cannot open shared object file: No such file or directory
Steps Taken:
I installed torch-intel and intel-extension-for-pytorch using the recommended pip command. The installation appears to complete successfully, but the required libmkl_intel_lp64.so.2 library is missing. I have confirmed with the find command that this file is not present in the virtual environment after installation.
3. Submit the Report
After you've pasted that template and filled in your Linux Kernel information, just click the green "Submit new issue" button at the bottom.
The developers will then see your report and can help figure out why the installer isn't working on your system.
What is the typical response time for GitHub issue reports?
What is the output of 'uname -a'?
Are there alternative AI tools that work on Intel GPUs?
Versions
My Goal: I am trying to install and run ComfyUI on my Chromebook, which has an integrated Intel GPU.
My System:
- Device: Acer Chromebook Spin 713 (codename: kohaku)
- ChromeOS Version: 140.16371.0
- Linux Kernel: (In your Linux terminal, run the command
uname -aand paste the output here)
The Problem: When I try to run my application after a clean install, I get the following error:
OSError: libmkl_intel_lp64.so.2: cannot open shared object file: No such file or directory
Steps Taken:
I installed torch-intel and intel-extension-for-pytorch using the recommended pip command. The installation appears to complete successfully, but the required libmkl_intel_lp64.so.2 library is missing. I have confirmed with the find command that this file is not present in the virtual environment after installation.
3. Submit the Report
After you've pasted that template and filled in your Linux Kernel information, just click the green "Submit new issue" button at the bottom.
The developers will then see your report and can help figure out why the installer isn't working on your system.
What is the typical response time for GitHub issue reports?
What is the output of 'uname -a'?
Are there alternative AI tools that work on Intel GPUs?
@Gdub662 Which version of PyTorch* and Intel® Extension for PyTorch* are you using in your env? As we have discontinued active development on Intel® Extension for PyTorch* after 2.8, it is recommended to use PyTorch* 2.8 directly without Intel® Extension for PyTorch*. Please refer to the doc here: https://www.intel.com/content/www/us/en/developer/articles/tool/pytorch-prerequisites-for-intel-gpu/2-8.html.
And if you still need Intel® Extension for PyTorch* in your environment, please follow the installation guide here: https://intel.github.io/intel-extension-for-pytorch/xpu/latest/. You may also check if the issue you're encountering now can be resolved by preloading oneMKL. Please refer to https://intel.github.io/intel-extension-for-pytorch/xpu/latest/tutorials/known_issues.html for more details.
