Compatibility issues with NumPy 2.0.0

[GuangChen2333]

When I run the code

import torch_directml

device = torch_directml.device()

print(device)

I got the error msg:

 A module that was compiled using NumPy 1.x cannot be run in
NumPy 2.0.0 as it may crash. To support both 1.x and 2.x
versions of NumPy, modules must be compiled with NumPy 2.0.
Some module may need to rebuild instead e.g. with 'pybind11>=2.12'.

If you are a user of the module, the easiest solution will be to
downgrade to 'numpy<2' or try to upgrade the affected module.
We expect that some modules will need time to support NumPy 2.

Traceback (most recent call last):  File "D:\JetBrains\PyCharm Professional\plugins\python\helpers\pydev\pydevd.py", line 2252, in <module>
    main()
  File "D:\JetBrains\PyCharm Professional\plugins\python\helpers\pydev\pydevd.py", line 2234, in main
    globals = debugger.run(setup['file'], None, None, is_module)
  File "D:\JetBrains\PyCharm Professional\plugins\python\helpers\pydev\pydevd.py", line 1544, in run
    return self._exec(is_module, entry_point_fn, module_name, file, globals, locals)
  File "D:\JetBrains\PyCharm Professional\plugins\python\helpers\pydev\pydevd.py", line 1551, in _exec
    pydev_imports.execfile(file, globals, locals)  # execute the script
  File "D:\JetBrains\PyCharm Professional\plugins\python\helpers\pydev\_pydev_imps\_pydev_execfile.py", line 18, in execfile
    exec(compile(contents+"\n", file, 'exec'), glob, loc)
  File "E:\Projects\---\main.py", line 1, in <module>
    import torch_directml
  File "E:\Projects\---\.venv\Lib\site-packages\torch_directml\__init__.py", line 94, in <module>
    torch._register_device_module('privateuseone', PrivateUse1Module)
  File "E:\Projects\---\.venv\Lib\site-packages\torch\__init__.py", line 1908, in _register_device_module
    device_type = torch.device(device_type).type
E:\Projects\---\.venv\Lib\site-packages\torch\__init__.py:1908: UserWarning: Failed to initialize NumPy: _ARRAY_API not found (Triggered internally at ..\torch\csrc\utils\tensor_numpy.cpp:84.)
  device_type = torch.device(device_type).type

Enviroments:

• Windows 11 23H2 22631.3880
• Python 3.12
• torch-directml 0.2.2.dev240614
• torch 2.3.1

[GongHuilin-DL]

I got the same error too. My system environment is：

1. system: Windows 11 专业工作站版
2. cpu：AMD Ryzen 9 7940HS
3. python 3.10.13
4. torch 2.3.1
5. numpy 2.0.1

[VadimShabashov]

As a temporary workaround, it helped me to downgrade numpy:

pip install numpy==1.26.4

[GuangChen2333]

As a temporary workaround, it helped me to downgrade numpy:

pip install numpy==1.26.4

Yes it works. It needs to limit the version or try to become compatible with numpy 2

[joshjkim]

We're tracking an issue where the latest Windows PyTorch builds are built against numpy<2 https://github.com/pytorch/pytorch/issues/131668. This issue should be fixed in the upcoming PyTorch 2.4.1 release which we plan to add support for