DLL load failed

[marwanmusa]

I got error with cpp_extension.py

---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
Cell In[1], line 5
      2 sys.path.append(".")
      3 sys.path.append('[./DragGAN](https://file+.vscode-resource.vscode-cdn.net/f%3A/MyFile/DragGAN/DragGAN)')
----> 5 from gradio_app import main

File [f:\MyFile\DragGAN\gradio_app.py:9](file:///F:/MyFile/DragGAN/gradio_app.py:9)
      6 from PIL import Image
      7 import uuid
----> 9 from drag_gan import drag_gan, stylegan2
     11 device = 'cuda'
     14 SIZE_TO_CLICK_SIZE = {
     15     1024: 5,
     16     256: 2
     17 }

File [f:\MyFile\DragGAN\drag_gan.py:12](file:///F:/MyFile/DragGAN/drag_gan.py:12)
      9 from torchvision import utils
     10 from tqdm import tqdm
---> 12 from stylegan2.model import Generator
     15 class DownloadProgressBar(tqdm):
     16     def update_to(self, b=1, bsize=1, tsize=None):

File [f:\MyFile\DragGAN\stylegan2\model.py:8](file:///F:/MyFile/DragGAN/stylegan2/model.py:8)
      5 from torch import nn
      6 from torch.nn import functional as F
----> 8 from .op import FusedLeakyReLU, fused_leaky_relu, upfirdn2d, conv2d_gradfix
     11 class PixelNorm(nn.Module):
     12     def __init__(self):

File [f:\MyFile\DragGAN\stylegan2\op\__init__.py:1](file:///F:/MyFile/DragGAN/stylegan2/op/__init__.py:1)
----> 1 from .fused_act import FusedLeakyReLU, fused_leaky_relu
      2 from .upfirdn2d import upfirdn2d

File [f:\MyFile\DragGAN\stylegan2\op\fused_act.py:11](file:///F:/MyFile/DragGAN/stylegan2/op/fused_act.py:11)
      7 from torch.utils.cpp_extension import load
     10 module_path = os.path.dirname(__file__)
---> 11 fused = load(
     12     "fused",
     13     sources=[
     14         os.path.join(module_path, "fused_bias_act.cpp"),
     15         os.path.join(module_path, "fused_bias_act_kernel.cu"),
     16     ],
     17 )
     20 class FusedLeakyReLUFunctionBackward(Function):
     21     @staticmethod
     22     def forward(ctx, grad_output, out, bias, negative_slope, scale):

File [f:\MyFile\DragGAN\venv\lib\site-packages\torch\utils\cpp_extension.py:1284](file:///F:/MyFile/DragGAN/venv/lib/site-packages/torch/utils/cpp_extension.py:1284), in load(name, sources, extra_cflags, extra_cuda_cflags, extra_ldflags, extra_include_paths, build_directory, verbose, with_cuda, is_python_module, is_standalone, keep_intermediates)
   1192 def load(name,
   1193          sources: Union[str, List[str]],
   1194          extra_cflags=None,
   (...)
   1202          is_standalone=False,
   1203          keep_intermediates=True):
   1204     r'''
   1205     Loads a PyTorch C++ extension just-in-time (JIT).
   1206 
   (...)
   1282         ...     verbose=True)
   1283     '''
-> 1284     return _jit_compile(
   1285         name,
   1286         [sources] if isinstance(sources, str) else sources,
   1287         extra_cflags,
   1288         extra_cuda_cflags,
   1289         extra_ldflags,
   1290         extra_include_paths,
   1291         build_directory or _get_build_directory(name, verbose),
   1292         verbose,
   1293         with_cuda,
   1294         is_python_module,
   1295         is_standalone,
   1296         keep_intermediates=keep_intermediates)

File [f:\MyFile\DragGAN\venv\lib\site-packages\torch\utils\cpp_extension.py:1535](file:///F:/MyFile/DragGAN/venv/lib/site-packages/torch/utils/cpp_extension.py:1535), in _jit_compile(name, sources, extra_cflags, extra_cuda_cflags, extra_ldflags, extra_include_paths, build_directory, verbose, with_cuda, is_python_module, is_standalone, keep_intermediates)
   1532 if is_standalone:
   1533     return _get_exec_path(name, build_directory)
-> 1535 return _import_module_from_library(name, build_directory, is_python_module)

File [f:\MyFile\DragGAN\venv\lib\site-packages\torch\utils\cpp_extension.py:1929](file:///F:/MyFile/DragGAN/venv/lib/site-packages/torch/utils/cpp_extension.py:1929), in _import_module_from_library(module_name, path, is_python_module)
   1927 spec = importlib.util.spec_from_file_location(module_name, filepath)
   1928 assert spec is not None
-> 1929 module = importlib.util.module_from_spec(spec)
   1930 assert isinstance(spec.loader, importlib.abc.Loader)
   1931 spec.loader.exec_module(module)

ImportError: DLL load failed while importing fused: The specified module could not be found.

previously, I got this error

CalledProcessError: Command '['ninja', '-v']' returned non-zero exit status 1.

then, I try to change ['ninja', '-v'] to ['ninja', '--version'] (got this solution while exploring on stackoverflow) .
and then, this Error : *DLL load failed* show up.

Can anybody help with this?

[marwanmusa]

I use virtual environment, and these all lib that I installed

-i https://pypi.org/simple
aiofiles==23.1.0 ; python_version >= '3.7' and python_version < '4.0'
aiohttp==3.8.4 ; python_version >= '3.6'
aiosignal==1.3.1 ; python_version >= '3.7'
altair==5.0.0 ; python_version >= '3.7'
anyio==3.6.2 ; python_full_version >= '3.6.2'
asttokens==2.2.1
async-timeout==4.0.2 ; python_version >= '3.6'
attrs==23.1.0 ; python_version >= '3.7'
backcall==0.2.0
certifi==2023.5.7 ; python_version >= '3.6'
charset-normalizer==3.1.0 ; python_full_version >= '3.7.0'
click==8.1.3 ; python_version >= '3.7'
colorama==0.4.6 ; platform_system == 'Windows'
comm==0.1.3 ; python_version >= '3.6'
contourpy==1.0.7 ; python_version >= '3.8'
cycler==0.11.0 ; python_version >= '3.6'
debugpy==1.6.7 ; python_version >= '3.7'
decorator==5.1.1 ; python_version >= '3.5'
executing==1.2.0
fastapi==0.95.2 ; python_version >= '3.7'
ffmpy==0.3.0
filelock==3.12.0 ; python_version >= '3.7'
fire==0.5.0
fonttools==4.39.4 ; python_version >= '3.8'
frozenlist==1.3.3 ; python_version >= '3.7'
fsspec==2023.5.0 ; python_version >= '3.8'
gradio==3.32.0
gradio-client==0.2.5 ; python_version >= '3.7'
h11==0.14.0 ; python_version >= '3.7'
httpcore==0.17.1 ; python_version >= '3.7'
httpx==0.24.1 ; python_version >= '3.7'
huggingface-hub==0.14.1 ; python_full_version >= '3.7.0'
idna==3.4 ; python_version >= '3.5'
imageio==2.28.1
imageio-ffmpeg==0.4.8
ipykernel==6.23.1
ipython==8.13.2 ; python_version >= '3.9'
ipywidgets==8.0.6
jedi==0.18.2 ; python_version >= '3.6'
jinja2==3.1.2 ; python_version >= '3.7'
jsonschema==4.17.3 ; python_version >= '3.7'
jupyter-client==8.2.0 ; python_version >= '3.8'
jupyter-core==5.3.0 ; python_version >= '3.8'
jupyterlab-widgets==3.0.7 ; python_version >= '3.7'
kiwisolver==1.4.4 ; python_version >= '3.7'
linkify-it-py==2.0.2
markdown-it-py[linkify]==2.2.0 ; python_version >= '3.7'
markupsafe==2.1.2 ; python_version >= '3.7'
matplotlib==3.7.1 ; python_version >= '3.8'
matplotlib-inline==0.1.6 ; python_version >= '3.5'
mdit-py-plugins==0.3.3 ; python_version >= '3.7'
mdurl==0.1.2 ; python_version >= '3.7'
multidict==6.0.4 ; python_version >= '3.7'
nest-asyncio==1.5.6 ; python_version >= '3.5'
ninja==1.11.1
numpy==1.24.3
orjson==3.8.12 ; python_version >= '3.7'
packaging==23.1 ; python_version >= '3.7'
pandas==2.0.1 ; python_version >= '3.8'
parso==0.8.3 ; python_version >= '3.6'
pickleshare==0.7.5
pillow==9.5.0 ; python_version >= '3.7'
platformdirs==3.5.1 ; python_version >= '3.7'
prompt-toolkit==3.0.38 ; python_full_version >= '3.7.0'
psutil==5.9.5 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'
pure-eval==0.2.2
pydantic==1.10.7 ; python_version >= '3.7'
pydub==0.25.1
pygments==2.15.1 ; python_version >= '3.7'
pyparsing==3.0.9 ; python_full_version >= '3.6.8'
pyrsistent==0.19.3 ; python_version >= '3.7'
python-dateutil==2.8.2 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2'
python-multipart==0.0.6 ; python_version >= '3.7'
pytz==2023.3
pywin32==306 ; sys_platform == 'win32' and platform_python_implementation != 'PyPy'
pyyaml==6.0 ; python_version >= '3.6'
pyzmq==25.0.2 ; python_version >= '3.6'
requests==2.30.0 ; python_version >= '3.7'
semantic-version==2.10.0 ; python_version >= '2.7'
six==1.16.0 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2'
sniffio==1.3.0 ; python_version >= '3.7'
stack-data==0.6.2
starlette==0.27.0 ; python_version >= '3.7'
termcolor==2.3.0 ; python_version >= '3.7'
toolz==0.12.0 ; python_version >= '3.5'
tornado==6.3.2 ; python_version >= '3.8'
tqdm==4.65.0
traitlets==5.9.0 ; python_version >= '3.7'
typing-extensions==4.5.0 ; python_version >= '3.7'
tzdata==2023.3 ; python_version >= '2'
uc-micro-py==1.0.2 ; python_version >= '3.7'
urllib3==2.0.2 ; python_version >= '3.7'
uvicorn==0.22.0 ; python_version >= '3.7'
wcwidth==0.2.6
websockets==11.0.3 ; python_version >= '3.7'
widgetsnbextension==4.0.7 ; python_version >= '3.7'
yarl==1.9.2 ; python_version >= '3.7'

torch              2.0.1+cu118
torchaudio         2.0.2+cu118
torchvision        0.15.2+cu118

[Zeqiang-Lai]

are you running the colab.ipynb locally？ you should use for local env

python gradio_app.py

[marwanmusa]

are you running the colab.ipynb locally yes
and executing this python gradio_app.py
in terminal (in the virtual environment), gives the same error

[marwanmusa]

[marwanmusa]

These are steps I did:

py -m venv venv
pip install pipenv
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118 #my nvcc version relate to this torch version
pipenv install #another libs except torch and torchvision

Is there any mistake?

[Zeqiang-Lai]

I have no idea what it causes too. I use conda personally, therefore I would recommend to use conda. For your python venv, could you try to install all requirements via

pip install -r requirements.txt

this is verified that works on remote google colab notebook.

[marwanmusa]

I re-setup the repo locally and then install all libs with pip install -r requirements.txt
but here a new error

it seems like caused by torch version installed trough pip install is not for cuda version (I assume).
But, I'll try to use conda then.

[Zeqiang-Lai]

ok, btw, if you have a workable environment with torch installed correctly, I think you can simply use it directly and install the missing packages.

We have very loose requirements for the package versions, so that most of your existing packages could remain unchanged.

[wanghdUT]

I got the same questions!
I guess it is because of the C++ version.
I download the latest Visual Studio and add the cl into the system pathway. But in other issues, they use VS2019 or sth instead.
So I am still confused with this error

[marwanmusa]

I got the same questions! I guess it is because of the C++ version. I download the latest Visual Studio and add the cl into the system pathway. But in other issues, they use VS2019 or sth instead. So I am still confused with this error
@wanghdUT
Which environment did you use? python env or conda?

I try to re-setup the virtual env using conda and it works,
It seems like conda env handle all visual runtime, dlls, and compiler things

[marwanmusa]

ok, btw, if you have a workable environment with torch installed correctly, I think you can simply use it directly and install the missing packages.

We have very loose requirements for the package versions, so that most of your existing packages could remain unchanged.
@Zeqiang-Lai
when we install torch instantly without specify which version, it'll install the latest version 2.x of torch and it doesn't support conv2d_gradfix. However the program will still running but not using gpu computation I guess.