search

VAST-AI-Research / TripoSR

MIT License
4.6k stars 535 forks source link

Followed the trobleshooting but still "torchmcubes was not compiled with CUDA support, use CPU version instead." #83

Closed luke2023 closed 7 months ago

luke2023 commented 7 months ago

I have already flowed:

setuptools>=49.6.0. If not, upgrade by pip install --upgrade setuptools. Then re-install torchmcubes by:

but there's no changes I used conda system. I downloaded pytorch for 12.1 and my cuda version is 12.4

package installed: aiofiles 23.2.1 pypi_0 pypi altair 5.3.0 pypi_0 pypi annotated-types 0.6.0 pypi_0 pypi antlr4-python3-runtime 4.9.3 pypi_0 pypi anyio 4.3.0 pypi_0 pypi attrs 23.2.0 pypi_0 pypi bzip2 1.0.8 h2bbff1b_5 ca-certificates 2024.3.11 haa95532_0 certifi 2024.2.2 pypi_0 pypi charset-normalizer 3.3.2 pypi_0 pypi click 8.1.7 pypi_0 pypi colorama 0.4.6 pypi_0 pypi coloredlogs 15.0.1 pypi_0 pypi contourpy 1.2.1 pypi_0 pypi cycler 0.12.1 pypi_0 pypi einops 0.7.0 pypi_0 pypi fastapi 0.110.1 pypi_0 pypi ffmpy 0.3.2 pypi_0 pypi filelock 3.9.0 pypi_0 pypi flatbuffers 24.3.25 pypi_0 pypi fonttools 4.51.0 pypi_0 pypi fsspec 2023.4.0 pypi_0 pypi gradio 4.8.0 pypi_0 pypi gradio-client 0.7.1 pypi_0 pypi h11 0.14.0 pypi_0 pypi httpcore 1.0.5 pypi_0 pypi httpx 0.27.0 pypi_0 pypi huggingface-hub 0.17.3 pypi_0 pypi humanfriendly 10.0 pypi_0 pypi idna 3.6 pypi_0 pypi imageio 2.34.0 pypi_0 pypi imageio-ffmpeg 0.4.9 pypi_0 pypi importlib-resources 6.4.0 pypi_0 pypi jinja2 3.1.2 pypi_0 pypi jsonschema 4.21.1 pypi_0 pypi jsonschema-specifications 2023.12.1 pypi_0 pypi kiwisolver 1.4.5 pypi_0 pypi lazy-loader 0.4 pypi_0 pypi libffi 3.4.4 hd77b12b_0 llvmlite 0.42.0 pypi_0 pypi markdown-it-py 3.0.0 pypi_0 pypi markupsafe 2.1.3 pypi_0 pypi matplotlib 3.8.4 pypi_0 pypi mdurl 0.1.2 pypi_0 pypi mpmath 1.3.0 pypi_0 pypi networkx 3.2.1 pypi_0 pypi numba 0.59.1 pypi_0 pypi numpy 1.26.3 pypi_0 pypi omegaconf 2.3.0 pypi_0 pypi onnxruntime 1.17.1 pypi_0 pypi opencv-python-headless 4.9.0.80 pypi_0 pypi openssl 3.0.13 h2bbff1b_0 orjson 3.10.0 pypi_0 pypi packaging 24.0 pypi_0 pypi pandas 2.2.1 pypi_0 pypi pillow 10.1.0 pypi_0 pypi pip 23.3.1 py311haa95532_0 platformdirs 4.2.0 pypi_0 pypi pooch 1.8.1 pypi_0 pypi protobuf 5.26.1 pypi_0 pypi psutil 5.9.8 pypi_0 pypi pydantic 2.6.4 pypi_0 pypi pydantic-core 2.16.3 pypi_0 pypi pydub 0.25.1 pypi_0 pypi pygments 2.17.2 pypi_0 pypi pymatting 1.1.12 pypi_0 pypi pyparsing 3.1.2 pypi_0 pypi pyreadline3 3.4.1 pypi_0 pypi python 3.11.8 he1021f5_0 python-dateutil 2.9.0.post0 pypi_0 pypi python-multipart 0.0.9 pypi_0 pypi pytz 2024.1 pypi_0 pypi pyyaml 6.0.1 pypi_0 pypi referencing 0.34.0 pypi_0 pypi regex 2023.12.25 pypi_0 pypi rembg 2.0.56 pypi_0 pypi requests 2.31.0 pypi_0 pypi rich 13.7.1 pypi_0 pypi rpds-py 0.18.0 pypi_0 pypi safetensors 0.4.2 pypi_0 pypi scikit-image 0.22.0 pypi_0 pypi scipy 1.13.0 pypi_0 pypi semantic-version 2.10.0 pypi_0 pypi setuptools 69.2.0 pypi_0 pypi shellingham 1.5.4 pypi_0 pypi six 1.16.0 pypi_0 pypi sniffio 1.3.1 pypi_0 pypi sqlite 3.41.2 h2bbff1b_0 starlette 0.37.2 pypi_0 pypi sympy 1.12 pypi_0 pypi tifffile 2024.2.12 pypi_0 pypi tk 8.6.12 h2bbff1b_0 tokenizers 0.14.1 pypi_0 pypi tomlkit 0.12.0 pypi_0 pypi toolz 0.12.1 pypi_0 pypi torch 2.2.2+cu121 pypi_0 pypi torchaudio 2.2.2+cu121 pypi_0 pypi torchmcubes 0.1.0 pypi_0 pypi torchvision 0.17.2+cu121 pypi_0 pypi tqdm 4.66.2 pypi_0 pypi transformers 4.35.0 pypi_0 pypi trimesh 4.0.5 pypi_0 pypi typer 0.12.1 pypi_0 pypi typing-extensions 4.8.0 pypi_0 pypi tzdata 2024.1 pypi_0 pypi urllib3 2.2.1 pypi_0 pypi uvicorn 0.29.0 pypi_0 pypi vc 14.2 h21ff451_1 vs2015_runtime 14.27.29016 h5e58377_2 websockets 11.0.3 pypi_0 pypi wheel 0.41.2 py311haa95532_0 xz 5.4.6 h8cc25b3_0 zlib 1.2.13 h8cc25b3_0

luke2023 commented 7 months ago

And my terminal 2024-04-09 01:03:08,493 - INFO - Running model ... C:\Users\Luke\Documents\recons\TripoSR-main\tsr\models\transformer\attention.py:629: UserWarning: 1Torch was not compiled with flash attention. (Triggered internally at ..\aten\src\ATen\native\transformers\cuda\sdp_utils.cpp:263.) hidden_states = F.scaled_dot_product_attention( 2024-04-09 01:03:11,240 - INFO - Running model finished in 2747.05ms. 2024-04-09 01:03:11,240 - INFO - Exporting mesh ... torchmcubes was not compiled with CUDA support, use CPU version instead. 2024-04-09 01:03:15,305 - INFO - Exporting mesh finished in 4065.16ms.

mrbid commented 7 months ago

Did you make sure that you installed all the of the Nvidia packages first? CUDA, CUDNN? You then need to uninstall torchmcubes and re-install it from the git.

luke2023 commented 7 months ago

I did the second one , now trying CUDA toolkit 12.4 exe (3gb installation)

luke2023 commented 7 months ago

@mrbid Thank you, it worked, but there's still a bug(I tried reinstalling everything but didn't worked) Any advices? 2024-04-11 20:38:38,302 - INFO - Initializing model finished in 6055.76ms. 2024-04-11 20:38:38,304 - INFO - Processing images ... 2024-04-11 20:38:39,165 - INFO - Processing images finished in 861.30ms. 2024-04-11 20:38:39,165 - INFO - Running image 1/1 ... 2024-04-11 20:38:39,166 - INFO - Running model ... C:\Users\Luke\Documents\recons\TripoSR-main\tsr\models\transformer\attention.py:629: UserWarning: 1Torch was not compiled with flash attention. (Triggered internally at ..\aten\src\ATen\native\transformers\cuda\sdp_utils.cpp:263.) hidden_states = F.scaled_dot_product_attention( 2024-04-11 20:38:41,497 - INFO - Running model finished in 2330.69ms. 2024-04-11 20:38:41,497 - INFO - Exporting mesh ... 2024-04-11 20:38:44,839 - INFO - Exporting mesh finished in 3342.01ms.

luke2023 commented 7 months ago

i am useing rtx3060 and windows 11

mrbid commented 7 months ago

@mrbid Thank you, it worked, but there's still a bug(I tried reinstalling everything but didn't worked) Any advices? 2024-04-11 20:38:38,302 - INFO - Initializing model finished in 6055.76ms. 2024-04-11 20:38:38,304 - INFO - Processing images ... 2024-04-11 20:38:39,165 - INFO - Processing images finished in 861.30ms. 2024-04-11 20:38:39,165 - INFO - Running image 1/1 ... 2024-04-11 20:38:39,166 - INFO - Running model ... C:\Users\Luke\Documents\recons\TripoSR-main\tsr\models\transformer\attention.py:629: UserWarning: 1Torch was not compiled with flash attention. (Triggered internally at ..\aten\src\ATen\native\transformers\cuda\sdp_utils.cpp:263.) hidden_states = F.scaled_dot_product_attention( 2024-04-11 20:38:41,497 - INFO - Running model finished in 2330.69ms. 2024-04-11 20:38:41,497 - INFO - Exporting mesh ... 2024-04-11 20:38:44,839 - INFO - Exporting mesh finished in 3342.01ms.

https://discuss.pytorch.org/t/flash-attention-compilation-warning/196692/12 This issue is fairly recent but doesn't seem to offer a solution? It suggests that Flash Attention is not supported on windows? you might have more luck trying TripoSR under WSL2 with Ubuntu Subsystem?