Build and install ExecuTorch

Overview

With the ExecuTorch source checked out and your virtual environment active, you can now build ExecuTorch and set up the Arm toolchain for Ethos-U cross-compilation.

For a full tutorial on building ExecuTorch, see the Learning Path Introduction to TinyML on Arm using PyTorch and ExecuTorch .

Install ExecuTorch

Upgrade pip and install build tools:

    

        
        
pip install --upgrade pip setuptools wheel

    

Build and install the executorch pip package:

    

        
        
./install_executorch.sh

    

After the installation finishes, verify the package is available:

    

        
        
pip list | grep executorch

    

Set up the Arm toolchain

Initialize the Arm-specific environment and accept the EULA:

    

        
        
./examples/arm/setup.sh --i-agree-to-the-contained-eula

    

Source the environment variables:

    

        
        
source ./examples/arm/ethos-u-scratch/setup_path.sh

    

    

        
        
pip install torch torchvision
pip install --no-build-isolation .
pip install --no-build-isolation third-party/ao

    

    

        
        
ps aux | grep buck
pkill -f buck

    

    

        
        
./install_executorch.sh --clean
git submodule sync
git submodule update --init --recursive
./install_executorch.sh

    

What you’ve learned and what’s next

In this section you’ve:

• Installed the ExecuTorch package and verified its availability
• Set up the Arm toolchain with Ethos-U support
• Configured the environment for cross-compiling to Ethos-U65

With ExecuTorch installed and the Arm toolchain configured, you can now compile .pte model files targeting the Ethos-U65 NPU.