Skip to main content

spatial sparse convolution

Project description

SpConv: Spatially Sparse Convolution Library

Build Status pypi versions

PyPI Install Downloads
CPU (Linux Only) PyPI Version pip install spconv pypi monthly download
CUDA 10.2 PyPI Version pip install spconv-cu102 pypi monthly download
CUDA 11.3 PyPI Version pip install spconv-cu113 pypi monthly download
CUDA 11.4 PyPI Version pip install spconv-cu114 pypi monthly download
CUDA 11.6 PyPI Version pip install spconv-cu116 pypi monthly download
CUDA 11.7 PyPI Version pip install spconv-cu117 pypi monthly download
CUDA 11.8 PyPI Version pip install spconv-cu118 pypi monthly download
CUDA 12.0 PyPI Version pip install spconv-cu120 pypi monthly download

spconv is a project that provide heavily-optimized sparse convolution implementation with tensor core support. check benchmark to see how fast spconv 2.x runs.

Spconv 1.x code. We won't provide any support for spconv 1.x since it's deprecated. use spconv 2.x if possible.

Check spconv 2.x algorithm introduction to understand sparse convolution algorithm in spconv 2.x!


Use spconv >= cu114 if possible. cuda 11.4 can compile greatly faster kernel in some situation.

Update Spconv: you MUST UNINSTALL all spconv/cumm/spconv-cuxxx/cumm-cuxxx first, use pip list | grep spconv and pip list | grep cumm to check all installed package. then use pip to install new spconv.


  • spconv 2.3: int8 quantization support. see docs and examples for more details.

  • spconv 2.2: ampere feature support (by EvernightAurora), pure c++ code generation, nvrtc, drop python 3.6

Spconv 2.2 vs Spconv 2.1

  • faster fp16 conv kernels (~5-30%) in ampere GPUs (tested in RTX 3090)
  • greatly faster int8 conv kernels (~1.2x-2.7x) in ampere GPUs (tested in RTX 3090)
  • drop python 3.6 support
  • nvrtc support: kernel in old GPUs will be compiled in runtime.
  • libspconv: pure c++ build of all spconv ops. see example
  • tf32 kernels, faster fp32 training, disabled by default. set import spconv as spconv_core; spconv_core.constants.SPCONV_ALLOW_TF32 = True to enable them.
  • all weights are KRSC layout, some old model can't be loaded anymore.

Spconv 2.1 vs Spconv 1.x

  • spconv now can be installed by pip. see install section in readme for more details. Users don't need to build manually anymore!
  • Microsoft Windows support (only windows 10 has been tested).
  • fp32 (not tf32) training/inference speed is increased (+50~80%)
  • fp16 training/inference speed is greatly increased when your layer support tensor core (channel size must be multiple of 8).
  • int8 op is ready, but we still need some time to figure out how to run int8 in pytorch.
  • doesn't depend on pytorch binary, but you may need at least pytorch >= 1.5.0 to run spconv 2.x.
  • since spconv 2.x doesn't depend on pytorch binary (never in future), it's impossible to support torch.jit/libtorch inference.


Firstly you need to use import spconv.pytorch as spconv in spconv 2.x.

Then see this.

Don't forget to check performance guide.

Common Solution for Some Bugs

see common problems.


You need to install python >= 3.7 first to use spconv 2.x.

You need to install CUDA toolkit first before using prebuilt binaries or build from source.

You need at least CUDA 11.0 to build and run spconv 2.x. We won't offer any support for CUDA < 11.0.


We offer python 3.7-3.11 and cuda 10.2/11.3/11.4/11.7/12.0 prebuilt binaries for linux (manylinux).

We offer python 3.7-3.11 and cuda 10.2/11.4/11.7/12.0 prebuilt binaries for windows 10/11.

For Linux users, you need to install pip >= 20.3 first to install prebuilt.

WARNING: spconv-cu117 may require CUDA Driver >= 515.

pip install spconv for CPU only (Linux Only). you should only use this for debug usage, the performance isn't optimized due to manylinux limit (no omp support).

pip install spconv-cu102 for CUDA 10.2

pip install spconv-cu113 for CUDA 11.3 (Linux Only)

pip install spconv-cu114 for CUDA 11.4

pip install spconv-cu117 for CUDA 11.7

pip install spconv-cu120 for CUDA 12.0

NOTE It's safe to have different minor cuda version between system and conda (pytorch) in CUDA >= 11.0 because of CUDA Minor Version Compatibility. For example, you can use spconv-cu114 with anaconda version of pytorch cuda 11.1 in a OS with CUDA 11.2 installed.

NOTE In Linux, you can install spconv-cuxxx without install CUDA to system! only suitable NVIDIA driver is required. for CUDA 11, we need driver >= 450.82. You may need newer driver if you use newer CUDA. for cuda 11.8, you need to have driver >= 520 installed.

Prebuilt GPU Support Matrix

See this page to check supported GPU names by arch.

If you use a GPU architecture that isn't compiled in prebuilt, spconv will use NVRTC to compile a slightly slower kernel.

CUDA version GPU Arch List
11.1~11.7 52,60,61,70,75,80,86
11.8+ 60,70,75,80,86,89,90

Build from source for development (JIT, recommend)

The c++ code will be built automatically when you change c++ code in project.

For NVIDIA Embedded Platforms, you need to specify cuda arch before build: export CUMM_CUDA_ARCH_LIST="7.2" for xavier, export CUMM_CUDA_ARCH_LIST="6.2" for TX2, export CUMM_CUDA_ARCH_LIST="8.7" for orin.

You need to remove cumm in requires section in pyproject.toml after install editable cumm and before install spconv due to pyproject limit (can't find editable installed cumm).

You need to ensure pip list | grep spconv and pip list | grep cumm show nothing before install editable spconv/cumm.


  1. uninstall spconv and cumm installed by pip
  2. install build-essential, install CUDA
  3. git clone, cd ./cumm, pip install -e .
  4. git clone, cd ./spconv, pip install -e .
  5. in python, import spconv and wait for build finish.


  1. uninstall spconv and cumm installed by pip
  2. install visual studio 2019 or newer. make sure C++ development component is installed. install CUDA
  3. set powershell script execution policy
  4. start a new powershell, run tools/msvc_setup.ps1
  5. git clone, cd ./cumm, pip install -e .
  6. git clone, cd ./spconv, pip install -e .
  7. in python, import spconv and wait for build finish.

Build wheel from source (not recommend, this is done in CI.)

You need to rebuild cumm first if you are build along a CUDA version that not provided in prebuilts.


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


  1. install visual studio 2019 or newer. make sure C++ development component is installed. install CUDA
  2. set powershell script execution policy
  3. start a new powershell, run tools/msvc_setup.ps1
  4. run $Env:SPCONV_DISABLE_JIT = "1"
  5. run pip install pccm cumm wheel
  6. run python bdist_wheel+pip install dists/xxx.whl


If you find this project useful in your research, please consider cite:

    title={Spconv: Spatially Sparse Convolution Library},
    author={Spconv Contributors},
    howpublished = {\url{}},



The work is done when the author is an employee at Tusimple.


Apache 2.0

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distributions

spconv_cu116-2.3.6-cp311-cp311-win_amd64.whl (68.1 MB view hashes)

Uploaded cp311

spconv_cu116-2.3.6-cp310-cp310-win_amd64.whl (68.1 MB view hashes)

Uploaded cp310

spconv_cu116-2.3.6-cp39-cp39-win_amd64.whl (68.1 MB view hashes)

Uploaded cp39

spconv_cu116-2.3.6-cp38-cp38-win_amd64.whl (68.1 MB view hashes)

Uploaded cp38

spconv_cu116-2.3.6-cp37-cp37m-win_amd64.whl (68.1 MB view hashes)

Uploaded cp37

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page