May I ask if spnn.conv3d is a submanifold sparse convolution or a regular sparse convolution?

mit-han-lab / torchsparse

[MICRO'23, MLSys'22] TorchSparse: Efficient Training and Inference Framework for Sparse Convolution on GPUs.

https://torchsparse.mit.edu

MIT License

1.15k stars 131 forks source link

May I ask if spnn.conv3d is a submanifold sparse convolution or a regular sparse convolution? #265

Closed Shixiaomeng7 closed 6 months ago

Shixiaomeng7 commented 7 months ago

Is there an existing issue for this?

[X] I have searched the existing issues

Current Behavior

May I ask if spnn. conv3d in torch sparse is a submanifold sparse convolution or a regular sparse convolution? How to distinguish like spconv.SubMConv3d and spconv.SparseConv3d in spconv?

Expected Behavior

No response

Environment

- GCC:
- NVCC:
- PyTorch:
- PyTorch CUDA:
- TorchSparse:

Anything else?

No response

ys-2020 commented 7 months ago

We support both submanifold and regular sparse convolution in spnn.conv3d. Just change the stride parameter to run the two types of convolution.