search

graphdeco-inria / hierarchical-3d-gaussians

Official implementation of the SIGGRAPH 2024 paper "A Hierarchical 3D Gaussian Representation for Real-Time Rendering of Very Large Datasets"
Other
935 stars 87 forks source link

MemoryError: std::bad_alloc: cudaErrorMemoryAllocation: out of memory #35

Open yuancaimaiyi opened 3 months ago

yuancaimaiyi commented 3 months ago

HI , I created a docker and had no problems on my computer NVIDIA GeForce RTX 3060,like : image but when I switched to A100, the following error occurred: A100: image A100 error: image gaussian_model.py", line 190, in create_from_pcd dist2 = torch.clamp_min(distCUDA2(xyz), 0.0000001) MemoryError: std::bad_alloc: cudaErrorMemoryAllocation: out of memory Isn't the A100 better? I don't understand why this error occurs?

Gaaaavin commented 3 months ago

Does solution in this issue solve the problem?

yuancaimaiyi commented 3 months ago

@Gaaaavin I followed the setup.py mentioned in the issue, but it didn't solve the problem. The key to this problem is that I am using a docker image. There is no problem on one machine, but this error will occur on A100.