ImportError: No module named 'upfirdn2d_plugin'

[maizer2]

Error message

warnings.warn('Failed to build CUDA kernels for upfirdn2d. Falling back to slow reference implementation. Details:\n\n' + traceback.format_exc())
Setting up PyTorch plugin "upfirdn2d_plugin"... Failed!
/workspace/PASTA-GAN-plusplus/torch_utils/ops/upfirdn2d.py:34: UserWarning: Failed to build CUDA kernels for upfirdn2d. Falling back to slow reference implementation. Details:

Traceback (most recent call last):
  File "/workspace/PASTA-GAN-plusplus/torch_utils/ops/upfirdn2d.py", line 32, in _init
    _plugin = custom_ops.get_plugin('upfirdn2d_plugin', sources=sources, extra_cuda_cflags=['--use_fast_math'])
  File "/workspace/PASTA-GAN-plusplus/torch_utils/custom_ops.py", line 110, in get_plugin
    torch.utils.cpp_extension.load(name=module_name, verbose=verbose_build, sources=sources, **build_kwargs)
  File "/opt/conda/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 986, in load
    return _jit_compile(
  File "/opt/conda/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1213, in _jit_compile
    return _import_module_from_library(name, build_directory, is_python_module)
  File "/opt/conda/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1560, in _import_module_from_library
    file, path, description = imp.find_module(module_name, [path])
  File "/opt/conda/lib/python3.8/imp.py", line 296, in find_module
    raise ImportError(_ERR_MSG.format(name), name=name)
ImportError: No module named 'upfirdn2d_plugin'

torch_utils/ops/upfirdn2d.py

line 31 ~ 35

try:
   _plugin = custom_ops.get_plugin('upfirdn2d_plugin', sources=sources, extra_cuda_cflags=['--use_fast_math'])
except:
   warnings.warn('Failed to build CUDA kernels for upfirdn2d. Falling back to slow reference implementation. Details:\n\n' + traceback.format_exc())

Reference

StyleGAN2 issue

[Cwj1212]

I have the same problem, did you solve it？

[maizer2]

@Cwj1212 I think that problem is caused by the ninja module

i just annotate that module line

[Cwj1212]

@Cwj1212 I think that problem is caused by the ninja module

i just annotate that module line

If it is directly annotated, it will not use custom operators for acceleration, and will not involve cpp extensions. It can be run directly but the efficiency will be reduced.