Advanced installation guide

Install the aihwkit conda package

At this time, the conda package is only available for the Linux environment. You can use the following steps as an example for how to install the aihwkit conda package.

There is a conda package for aihwkit available in conda-forge. It can be installed in a conda environment running on a Linux or WSL in a Windows system.

Install any one of the conda packages as shown below.

  • CPU:

    $ conda install -c conda-forge aihwkit
    
  • GPU:

    $ conda install -c conda-forge aihwkit-gpu
    

Install the aihwkit using pip

AIHWKIT can also be installed using pip commands as shown below.

  • CPU:

    $ pip install aihwkit
    
  • GPU:

    $ wget https://aihwkit-gpu-demo.s3.us-east.cloud-object-storage.appdomain.cloud/aihwkit-0.8.0+cuda117-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
    

    then:

    $ pip install aihwkit-0.8.0+cuda117-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
    

Note

Please note that the instructions on this page refer to installing as an end user. If you are planning to contribute to the project, an alternative setup and tips can be found at the Development setup section that is more tuned towards the needs of a development cycle.