Visualize Ethos-U NPU performance with ExecuTorch on Arm FVPs
Introduction
Overview
Understand the ExecuTorch workflow
Set up your ExecuTorch environment
Set up the Corstone-320 Fixed Virtual Platform
Deploy and run Mobilenet V2 on the Corstone-320 FVP
Enable GUI and deploy a model on Corstone-320 FVP
Evaluate Ethos-U Performance
Next Steps

  Log an issue

  Fork and edit

  Discuss on Discord

Visualize model execution using the FVP GUI

You’ve successfully deployed a model on the Corstone-320 FVP from the command line. In this step, you’ll enable the platform’s built-in graphical output and re-run the model to observe instruction-level execution metrics in a windowed display.

Find your IP address

Note down your computer’s IP address:

    

        
        
ip addr show

You’ll use the IP address of your active network interface (inet) later to pass as an argument to the FVP.

Note

For macOS, note down your en0 IP address (or whichever network adapter is active):

    

        
        
ipconfig getifaddr en0 # Returns your Mac's WiFi IP address

Configure the FVP for GUI output

Edit the following parameters in your locally checked out executorch/backends/arm/scripts/run_fvp.sh file, to enable the Mobilenet V2 output on the FVP’s GUI:

    

        
        
-C mps4_board.subsystem.ethosu.num_macs=${num_macs} \
-C mps4_board.visualisation.disable-visualisation=1 \
-C vis_hdlcd.disable_visualisation=1                \
-C mps4_board.telnetterminal0.start_telnet=0        \
-C mps4_board.uart0.out_file='-'                    \
-C mps4_board.uart0.shutdown_on_eot=1               \
  • Change mps4_board.visualisation.disable-visualisation to equal 0
  • Change vis_hdlcd.disable_visualisation to equal 0
  • Enter a --display-ip parameter and set it to your computer’s IP address
    

        
        
-C mps4_board.subsystem.ethosu.num_macs=${num_macs} \
-C mps4_board.visualisation.disable-visualisation=0 \
-C vis_hdlcd.disable_visualisation=0                \
-C mps4_board.telnetterminal0.start_telnet=0        \
-C mps4_board.uart0.out_file='-'                    \
-C mps4_board.uart0.shutdown_on_eot=1               \
--display-ip <YOUR_IP_ADDRESS>                   \

Deploy the model

Now run the Mobilenet V2 computer vision model, using executorch/examples/arm/run.sh :

    

        
        
./examples/arm/run.sh \
--aot_arm_compiler_flags="--delegate --quantize --intermediates mv2_u85/ --debug --evaluate" \
--output=mv2_u85 \
--target=ethos-u85-128 \
--model_name=mv2

Observe that the FVP loads the model file, compiles the PyTorch model to ExecuTorch .pte format and then shows an instruction count in the top right of the GUI:

Image Alt Text:Terminal and FVP outputTerminal and FVP output

Note

For macOS users, follow these instructions:

  • Start Docker. FVPs run inside a Docker container.

  • Make sure to use an official version of Docker and not a free version like the Colima Docker container runtime

  • run.sh assumes Docker Desktop style networking (host.docker.internal) which breaks with Colima

  • Colima then breaks the FVP GUI

  • Start XQuartz: on macOS, the FVP GUI runs using XQuartz.

    Start the xquartz.app and then configure XQuartz so that the FVP will accept connections from your Mac and localhost:

        

        
        
xhost + <YOUR_IP_ADDRESS>
xhost + 127.0.0.1 # The Docker container seems to proxy through localhost
Back
Next