Home

Awesome

spconv (group-wise subconv supported)

Linux

  1. install build-essential, install CUDA
  2. run export SPCONV_DISABLE_JIT="1"
  3. run pip install pccm cumm wheel
  4. run python setup.py bdist_wheel+pip install dists/xxx.whl

QA

If you compile spconv successfully, but found the following error when import spconv.pytorch:

ImportError: arg(): could not convert default argument 'workspace: tv::Tensor' in method '<class 'spconv.core_cc.csrc.sparse.convops.gemmops.GemmTunerSimple'>.run_with_tuned_result' into a Python object (type not registered yet?)

it is because the cumm does not match the cuda version. Please rebuild the cumm by running

git clone https://github.com/FindDefinition/cumm, cd ./cumm, pip install -e .

LICENSE

Apache 2.0