[BUG] AttributeError: module 'torchsparse.backend' has no attribute 'build_kernel_map_subm_hashmap'

[Xiao]

Is there an existing issue for this?

• I have searched the existing issues

Current Behavior

When I run the test.py in this project, I got 2 errors:

FAILED (errors=2)

Error
Traceback (most recent call last):
File "/home/hx/PycharmProjects/torchsparse/tests/test.py", line 20, in test_single_layer
mean_adiff, max_rdiff = test_single_layer_convolution_forward(
File "/home/hx/PycharmProjects/torchsparse/tests/python/test_single_layer_conv.py", line 202, in test_single_layer_convolution_forward
out = model(feats_t, coords_t)
File "/home/hx/anaconda3/envs/torchsparse/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1190, in _call_impl
return forward_call(*input, **kwargs)
File "/home/hx/PycharmProjects/torchsparse/tests/python/test_single_layer_conv.py", line 61, in forward
return self.net(ts_tensor)
File "/home/hx/anaconda3/envs/torchsparse/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1190, in _call_impl
return forward_call(*input, **kwargs)
File "/home/hx/anaconda3/envs/torchsparse/lib/python3.9/site-packages/torch/nn/modules/container.py", line 204, in forward
input = module(input)
File "/home/hx/anaconda3/envs/torchsparse/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1190, in _call_impl
return forward_call(*input, **kwargs)
File "/home/hx/PycharmProjects/torchsparse/torchsparse/nn/modules/conv.py", line 98, in forward
return F.conv3d(
File "/home/hx/PycharmProjects/torchsparse/torchsparse/nn/functional/conv/conv.py", line 92, in conv3d
kmap = F.build_kernel_map(
File "/home/hx/PycharmProjects/torchsparse/torchsparse/nn/functional/conv/kmap/build_kmap.py", line 85, in build_kernel_map
kmap = build_kmap_implicit_GEMM_hashmap_on_the_fly(
File "/home/hx/PycharmProjects/torchsparse/torchsparse/nn/functional/conv/kmap/func/hashmap_on_the_fly.py", line 48, in build_kmap_implicit_GEMM_hashmap_on_the_fly
func = torchsparse.backend.build_kernel_map_subm_hashmap
AttributeError: module 'torchsparse.backend' has no attribute 'build_kernel_map_subm_hashmap'

Error
Traceback (most recent call last):
File "/home/hx/PycharmProjects/torchsparse/tests/test.py", line 46, in test_to_dense
max_adiff = test_to_dense_forward()
File "/home/hx/PycharmProjects/torchsparse/tests/python/test_to_dense.py", line 48, in test_to_dense_forward
output = to_dense(feats_t, coords_t, spatial_range).cpu().numpy()
File "/home/hx/PycharmProjects/torchsparse/torchsparse/utils/to_dense.py", line 64, in to_dense
return ToDenseFunction.apply(feats, coords, spatial_range)
File "/home/hx/PycharmProjects/torchsparse/torchsparse/utils/to_dense.py", line 30, in forward
torchsparse.backend.to_dense_forward_cuda(
AttributeError: module 'torchsparse.backend' has no attribute 'to_dense_forward_cuda'

Expected Behavior

No error

Environment

- GCC:gcc (Ubuntu 11.4.0-2ubuntu1~20.04) 11.4.0
- NVCC:Cuda compilation tools, release 11.7, V11.7.99
- PyTorch:1.13.0+cu117
- PyTorch CUDA:11.7
- TorchSparse:2.1.0+torch113cu117

Anything else?

No response

[Shang Yang]

Hi @huohuohuohuohuohuohuohuo . Seems like you haven't installed TorchSparse++ correctly. Have you solved the problem?

[Xiao]

I indeed installed TorchSprase++ successfully and no error is reported. I copied test_single_layer_conv.py and test_to_dense.py to another path and run them respectively and it worked.

[Shang Yang]

Thank you for pointing that out. Generally, you can call these two *.py files directly by runing test.py in tests. And I have fixed some outdated import operations in the codebase.

I didn't observe the same error as yours. I guess the error you met might be related to your python environment. The torchsparse.backend may not be successfully linked at first.

[Shang Yang]

Close this issue as completed. Feel free to reopen it if you have further questions.