I encountered a problem while installing xpu2.6.The error is [WinError 127]

[LIwii1]

Describe the issue

My GPU is A770 16G, and my system is running Windows 11. When testing commands, I encountered an error named .[WinError 127]'. I tried to find solutions, including installing conda install libuv, creating a new conda, and trying other similar problem-solving methods, but the issue was not resolved. I need some help.
Below are the logs:

(ipex_fresh) C:\Users\LWZ>python -c "import torch; import intel_extension_for_pytorch as ipex; print(torch.version); print(ipex.version); [print(f'[{i}]: {torch.xpu.get_device_properties(i)}') for i in range(torch.xpu.device_count())];"
Traceback (most recent call last):
File "C:\Users\LWZ\AppData\Roaming\Python\Python310\site-packages\torch_init_.py", line 2756, in import_device_backends
entrypoint = backend_extension.load()
File "G:\ttl\anaconda\envs\ipex_fresh\lib\importlib\metadata_init.py", line 171, in load
module = import_module(match.group('module'))
File "G:\ttl\anaconda\envs\ipex_fresh\lib\importlib_init_.py", line 126, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
File "", line 1050, in _gcd_import
File "", line 1027, in _find_and_load
File "", line 1006, in _find_and_load_unlocked
File "", line 688, in _load_unlocked
File "", line 883, in exec_module
File "", line 241, in call_with_frames_removed
File "C:\Users\LWZ\AppData\Roaming\Python\Python310\site-packages\intel_extension_for_pytorch_init.py", line 113, in
raise err
OSError: [WinError 127] 找不到指定的程序（Cannot find the specified program.）。 Error loading "C:\Users\LWZ\AppData\Roaming\Python\Python310\site-packages\intel_extension_for_pytorch\bin\intel-ext-pt-gpu.dll" or one of its dependencies.

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "", line 1, in
File "C:\Users\LWZ\AppData\Roaming\Python\Python310\site-packages\torch_init_.py", line 2784, in
import_device_backends()
File "C:\Users\LWZ\AppData\Roaming\Python\Python310\site-packages\torch_init.py", line 2760, in _import_device_backends
raise RuntimeError(
RuntimeError: Failed to load the backend extension: intel_extension_for_pytorch. You can disable extension auto-loading with TORCH_DEVICE_BACKEND_AUTOLOAD=0.

[ZailiWang]

Hi, would you elaborate on the steps you used? Please double check with the installation guide. No conda install libuv needed for latest 2.6.10+xpu version.

[LIwii1]

Indeed, I followed the installation guide step by step to complete the installation，i installed the latest version of C++ and the latest driver, Many people have encountered this problem.
Below is my screenshot

Hi, would you elaborate on the steps you used? Please double check with the installation guide. No conda install libuv needed for latest 2.6.10+xpu version.您好，您能详细说明您使用的步骤吗？请仔细检查安装指南 。最新的 2.6.10+xpu 版本不需要 conda install libuv。

[shangerxin]

Hi @LIwii1 ,

Could you please try these steps?

1. Disable the integrated graphics card when the machine has it.
2. Create clean conda env
   > conda create --no-default-packages python=3.10 -y
3. Install libuv > conda install pkg-config libuv -y
4. Install torch* 2.6 wheels
   > pip install torch==2.6.0 --index-url https://download.pytorch.org/whl/xpu
   > pip install torchvision==0.21.0 --index-url https://download.pytorch.org/whl/xpu
   > pip install torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/xpu
5. Install IPEX
6. Run the sanity test code.

If you still have the issue could you please share your conda packages list and your GPU device list from your device manager?

Thank you.

[LIwii1]

Thank you, I tried to install xpu according to your method, but I'm not sure what to do in the fifth step, do you still copy the previous two lines of code? Such as this：
python -m pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/xpu
python -m pip install intel-extension-for-pytorch==2.6.10+xpu --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/

Attached below is my log.

Anaconda Prompt - conda install pkg-config libuv.txt

[shangerxin]

Yes you can use the previous documented code to install IPEX. The flow is used to make sure the dependencies are installed correctly.

[tye1]

Have you ever tried 'TORCH_DEVICE_BACKEND_AUTOLOAD=0'? It might be a circle import issue if you import any third party library such as Transformers before import torch, such as this known issue:

https://intel.github.io/intel-extension-for-pytorch/xpu/latest/tutorials/known_issues.html

Problem: RuntimeError: Failed to load the backend extension: intel_extension_for_pytorch. You can disable extension auto-loading with TORCH_DEVICE_BACKEND_AUTOLOAD=0.

Cause: If you import any third party library such as Transformers before import torch, and the third party library has dependency to torch and then implicitly autoloads intel_extension_for_pytorch, which introduces circle import.

Solution: Disable extension auto-loading with TORCH_DEVICE_BACKEND_AUTOLOAD=0.

[LIwii1]

Yes you can use the previous documented code to install IPEX. The flow is used to make sure the dependencies are installed correctly.是的，您可以使用之前记录的代码来安装 IPEX。该流程用于确保正确安装依赖项。

I tried your method and there was no error, but I don't know if it was successful. Here is the log:

(ipex) C:\Users\LWZ>python -c "import torch; import intel_extension_for_pytorch as ipex; print(torch.version); print(ipex.version); [print(f'[{i}]: {torch.xpu.get_device_properties(i)}') for i in range(torch.xpu.device_count())];"
[W324 14:42:32.000000000 OperatorEntry.cpp:161] Warning: Warning only once for all operators, other operators may also be overridden.
Overriding a previously registered kernel for the same operator and the same dispatch key
operator: aten::_validate_compressed_sparse_indices(bool is_crow, Tensor compressed_idx, Tensor plain_idx, int cdim, int dim, int nnz) -> ()
registered at C:\actions-runner_work\pytorch\pytorch\pytorch\build\aten\src\ATen\RegisterSchema.cpp:6
dispatch key: XPU
previous kernel: registered at C:\actions-runner_work\pytorch\pytorch\pytorch\build\aten\src\ATen\RegisterCPU.cpp:30477
new kernel: registered at E:\frameworks.ai.pytorch.ipex-gpu\build\Release\csrc\gpu\csrc\aten\generated\ATen\RegisterXPU.cpp:468 (function operator ())
2.6.0+xpu
2.6.10+xpu
[0]: _XpuDeviceProperties(name='Intel(R) Arc(TM) A770 Graphics', platform_name='Intel(R) oneAPI Unified Runtime over Level-Zero', type='gpu', driver_version='1.6.31896', total_memory=15930MB, max_compute_units=512, gpu_eu_count=512, gpu_subslice_count=64, max_work_group_size=1024, max_num_sub_groups=128, sub_group_sizes=[8 16 32], has_fp16=1, has_fp64=0, has_atomic64=1)

(ipex) C:\Users\LWZ>

[shangerxin]

I think it works. The IPEX is loaded and the device Arc A770 is found.

[LIwii1]

Thank you, your method is effective

[LIwii1]

Anaconda Prompt - conda install pkg-config libuv.txt