Closed shzcuber closed 4 weeks ago
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 198.00 MiB (GPU 0; 3.81 GiB total capacity; 1.99 GiB already allocated; 196.25 MiB free; 2.25 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
Here's my OutOfMemoryError if that helps.
It looks like the point cloud obtained from the mask you entered is too sparse and the registration is not successful.
Please check #44
Sorry, do you mind elaborating on what you mean by the mask I entered?
Not sure if this helps, but all the data I'm using is the demo_data
as mentioned in the README.md - https://drive.google.com/drive/folders/1pRyFmxYXmAnpku7nGRioZaKrVJtIsroP?usp=sharing.
Because you change the shorter edge value to 10, this will cause your depth image to be scaled out too small, resulting in no depth data. Suggest trying FoundationPose on a GPU with larger memory.
Hello!
I installed FoundationPose with Ubuntu 20.04 and docker. I'm trying to run the
run_demo.py
script. Initially, I ran into the same memory issue as #26, so I changed my shorter side value to 10 (I can't do anything much above that value). Now, I'm getting this issue:Any ideas on what the issue is?
Here's the error:
Please let me know if there's any more information that could be provided.