AutoAWQ Kernels is a new package that is split up from the main repository in order to avoid compilation times.

• Windows: Must use WSL2.

• NVIDIA:

  • GPU: Must be compute capability 7.5 or higher.
  • CUDA Toolkit: Must be 11.8 or higher.

• AMD:

  • ROCm: Must be 5.6 or higher.

The package is available on PyPi with CUDA 12.1.1 wheels:

pip install autoawq-kernels

For ROCm and other CUDA versions, you can use the wheels published at each release:

pip install https://github.com/casper-hansen/AutoAWQ_kernels/releases/download/v0.0.2/autoawq_kernels-0.0.2+rocm561-cp310-cp310-linux_x86_64.whl

You can also build from source:

git clone https://github.com/casper-hansen/AutoAWQ_kernels
cd AutoAWQ_kernels
pip install -e .

To build for ROCm, you need to first install the following packages rocsparse-dev hipsparse-dev rocthrust-dev rocblas-dev hipblas-dev.