tinny-cuda wont install. CUDA not found

[Sinicyn78]

(nerfstudio) root@home:/home/sinicyn/Downloads# pip install git+https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch
Collecting git+https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch
Cloning https://github.com/NVlabs/tiny-cuda-nn/ to /tmp/pip-req-build-b8w1_isc
Running command git clone --filter=blob:none --quiet https://github.com/NVlabs/tiny-cuda-nn/ /tmp/pip-req-build-b8w1_isc
Resolved https://github.com/NVlabs/tiny-cuda-nn/ to commit 14053e9a87ebf449d32bda335c0363dd4f5667a4
Running command git submodule update --init --recursive -q
Preparing metadata (setup.py) ... error
error: subprocess-exited-with-error

× python setup.py egg_info did not run successfully.
│ exit code: 1
╰─> [10 lines of output]
/root/anaconda3/envs/nerfstudio/lib/python3.8/site-packages/torch/cuda/init.py:83: UserWarning: CUDA initialization: Unexpected error from cudaGetDeviceCount(). Did you run some cuda functions before calling NumCudaDevices() that might have already set an error? Error 803: system has unsupported display driver / cuda driver combination (Triggered internally at ../c10/cuda/CUDAFunctions.cpp:109.)
return torch._C._cuda_getDeviceCount() > 0
Traceback (most recent call last):
File "", line 2, in
File "", line 34, in
File "/tmp/pip-req-build-b8w1_isc/bindings/torch/setup.py", line 30, in
raise EnvironmentError("Unknown compute capability. Specify the target compute capabilities in the TCNN_CUDA_ARCHITECTURES environment variable or install PyTorch with the CUDA backend to detect it automatically.")
OSError: Unknown compute capability. Specify the target compute capabilities in the TCNN_CUDA_ARCHITECTURES environment variable or install PyTorch with the CUDA backend to detect it automatically.
No CUDA runtime is found, using CUDA_HOME='/usr'
Building PyTorch extension for tiny-cuda-nn version 1.7
[end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.
hint: See above for details.

[THE-COB]

Where did you install CUDA_HOME and CUDA_PATH?

[Sinicyn78]

I'm trying to install in windows via conda
Installed CUDA 11.3
now these errors

[THE-COB]

I'm trying to install in windows via conda
Installed CUDA 11.3
now these errors

Can I see a screenshot of your system and user environment variables

[Sinicyn78]

I'm trying to install in windows via conda
Installed CUDA 11.3
now these errors

Can I see a screenshot of your system and user environment variables

[THE-COB]

Try making a CUDA_HOME env variable with the same value as CUDA_PATH in both system and user variables (this way you can do it on both admin and user accounts)

[Sinicyn78]

Try making a CUDA_HOME env variable with the same value as CUDA_PATH in both system and user variables (this way you can do it on both admin and user accounts)

now these errors:

[THE-COB]

Try installing ninja like in this issue NVlabs/tiny-cuda-nn#169 (comment)

[dooonnis]

I also ran into this problem and tried many solutions. I found the solution here #1177 and installed tiny-cuda-nn thanks to this comment.

To document what I have done to get nerfstudio via conda to work properly:

1. Installed CUDA 11.7 and VS 2019 and deinstalled all other versions

2. Created a conda envoriment
   conda create --name nerfstudio -y python=3.8
conda activate nerfstudio
python -m pip install --upgrade pip

3. Installed pytorch
   pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 -f https://download.pytorch.org/whl/torch_stable.html

4. Git cloned tiny-cuda-nn via the repo into miniconda3\envs\nerfstudio\Lib\site-packages
   git clone --recursive https://github.com/nvlabs/tiny-cuda-nn
cd tiny-cuda-nn
cmake . -B build
cmake --build build --config RelWithDebInfo -j
   (cmake commands run in VS developer command prompt)

5. Installed nerfstudio
   pip install nerfstudio

6. Reinstalled tiny-cuda-nn via pip
   pip install git+https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch

How tiny-cuda-nn was installed may not makes sense, but that's what worked in the end. Nerfstudio is now fully functional.

[tancik]

Closing, feel free to reopen if it is still and issue.

[Xerthor]

This one worked for me but I had to install CUDA 12.1.1 before runing cmake . -B build