Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

CUDA Error #5

Open
besnardb opened this issue Jul 11, 2024 · 3 comments
Open

CUDA Error #5

besnardb opened this issue Jul 11, 2024 · 3 comments

Comments

@besnardb
Copy link

Hi, I got the following error :

Traceback (most recent call last): File "/render.py" rasterizer = GaussianRasterizer(raster_settings=raster_settings) File "/fast_gaussian_rasterization/fast_gauss/__init__.py", line 39, in __init__ raster_context = GSplatContextManager(init_buffer_size=init_buffer_size, init_texture_size=init_texture_size, dtype=dtype, tex_dtype=tex_dtype) # only created once File "/fast_gaussian_rasterization/fast_gauss/gsplat_utils.py", line 63, in __init__ self.resize_buffers(init_buffer_size) File "/fast_gaussian_rasterization/fast_gauss/gsplat_utils.py", line 247, in resize_buffers self.init_gl_buffers(self.max_verts) File "/fast_gaussian_rasterization/fast_gauss/gsplat_utils.py", line 184, in init_gl_buffers raise e File "/fast_gaussian_rasterization/fast_gauss/gsplat_utils.py", line 177, in init_gl_buffers self.cu_vbo = CHECK_CUDART_ERROR(cudart.cudaGraphicsGLRegisterBuffer(self.vbo, flags)) File "/fast_gaussian_rasterization/fast_gauss/cuda_utils.py", line 93, in CHECK_CUDART_ERROR raise RuntimeError(FORMAT_CUDART_ERROR(err)) RuntimeError: cudaErrorOperatingSystem(304): OS call failed or operation not supported on this OS

Environment Details:

  • Driver Version: 550.54.15
  • Docker Image: nvidia/cuda:11.6.2-devel-ubuntu20.04
  • CUDA Toolkit: 11.6

Could you please provide information about the environment used ?

@dendenxu
Copy link
Owner

Hi, thanks for using our code.
The rasterizer is tested on native Windows and Ubuntu.
However, in my previous experience on other projects, the CUDA-GL interop should work fine on docker-based environments (at least for offline rendering). Thus this might be a usage issue. Could you provide more details, for example is your machine connected to a display, or are you using any Graphical forwarding (X11?) if this is a remote server?

@besnardb
Copy link
Author

I'm currently using offline rendering and the issue seems to occur at the beginning : raster_context = GSplatContextManager(init_buffer_size=init_buffer_size, init_texture_size=init_texture_size, dtype=dtype, tex_dtype=tex_dtype) line 39 of __init__.py file.

Could it be that some required packages are missing in my environment?
Do you have a list of packages used for offline rendering?

@dendenxu
Copy link
Owner

The required packages should have already been installed if you've run pip install -r requirements.txt without problems. In this situation, could you try running this test: https://github.com/zju3dv/EasyVolcap/blob/main/tests/headless_opengl_tests.py and check whether any error occurs? Note that since we're only testing the OpenGL environment, there's no need to install extra dependencies:

git clone https://github.com/zju3dv/EasyVolcap
cd EasyVolcap
pip install -e . --no-build-isolation --no-deps
python tests/headless_opengl_tests.py # maybe install missing packages if any fatal errors are reported

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants