`x-cube-n6-ai-power-measurement`

Computer Vision application to enable the deployment of object detections models on STM32N6570-DK board and to measure power consumption.

This application is prebuilt with a people detection model "TinyYOLOv2".

This top readme gives an overview of the app. Additional documentation is available in the Doc folder.

⚠️ Warning: this application requires a hardware modification of the DK board (see Power measurement environment).

Features demonstrated in this example
Hardware Support
Tools version
Boot modes
Quickstart using prebuilt binaries
- How to Program hex files using STM32CubeProgrammer UI
- How to Program hex files using Command line
Quickstart using source code
- Application build and run - Dev mode
- Application build and run - Boot from flash
Known Issues and Limitations

Doc Folder Content

Features demonstrated in this example

Sequential application flow
NPU accelerated quantized AI model inference
DCMIPP pipe
DCMIPP crop, decimation, downscale
DCMIPP ISP usage
Dev mode
Boot from External Flash
Wake-up from sleep using USER1 button
System requency scaling (switching betwing Overdrive and nominal modes)
De-init of unused IPs
CPU sleep during capture and during NPU HW epochs (ASYNC mode)

Hardware Support

The development platform:

STM32N6570-DK discovery board
- The board should be connected to the onboard ST-LINK debug adapter CN6 with a USB-C to USB-C cable to ensure sufficient power
- Set the boot mode in dev mode (BOOT1 switch to right position. BOOT0 switch position doesn't matter)
- OTP fuses are set in this example for xSPI IOs in order to get the maximum speed (200MHz) on xSPI interfaces

And one of the following camera modules:

Provided IMX335 camera module
STEVAL-55G1MBI
STEVAL-66GYMAI1

⚠️ Warning: for power measurement, only IMX335 camera has been tested.

STM32N6570-DK board with MB1854B IMX335.

Measuring Power Consumption using STM32N6 Discovery Board

To measure the power consumption of an STM32N6 Discovery Board (DK), particularly the core domain supply (IddCore/VddCore), which includes the CPU, NPU, interconnects, and CSI Phy digital supply, certain hardware modifications are required.

Specifically, these modifications involve removing components to allow the application of the VddCore voltage upstream of the SMPS using an STLINK_PWR and measuring the consumed current. Simultaneously, the firmware will log a timestamp of the different stages of the code to delineate them and accurately measure the consumption of each step. These data (consumed current and timestamps) are then retrieved by Python scripts, which process them to generate a detailed report of the power consumption.

Hardware Modifications

Component Removal

Remove the components R23, R24, and C27 from the DK board. This step is crucial to allow the reinjection of current using the STLINKPOWER_v3 tool through I_SENS_EXT_N pin.

Component to remove on board :

Current Injection

The STLINKPOWER_v3 tool will be used to inject current before the external SMPS (3.3V). Note that the efficiency of the SMPS must be considered when measuring power. For instance:

TPS62088: Efficiency is approximately 0.85% (for DK version C02), this value should be the one to be used in filed efficiency: 0.85 in file my_configuration.yml

Power Measurement Setup

STLink_Pwr to N6 DK Pin Connections (for VddCore)

To properly connect the STLink_Pwr tool to the N6 DK board, use the following pin mappings. Note that CNx-y refers to the connector header CNx on the DK board, where y is the pin number of the header.

STLink_Pwr Pin	N6 DK Pin
OUT	CN2-2 (I_SENS_EXT_N)
GND	CN2-8 (GND)
TGI	CN4-1 (PG5)
TVCC	CN2-12 (I_SENS_VDDIO_N)
GND	CN2-7 (GND)

Additional Information

For debugging in development mode, the embedded STLINK_v3 on the DK board is used. It is also utilized to receive power timestamps. It is crucial to use a USB-C to USB-C cable for this connection.

A second USB-C to USB-C cable is used to connect the STLINK_PWR to the computer to obtain power consumption measurements.

Both devices are interfaced with the power measurement Python scripts.

Python Scripts

Utilize the Python scripts located in the Utilities/pwr_scripts directory. These scripts not only measure power consumption but also provide timestamps from the application. This feature is essential for accurately identifying and locating all steps and sequences of the application. detailed instructions to use these scripts are listed in this README.md

Tools version

IAR Embedded Workbench for Arm (EWARM 9.40.1) + N6 patch (EWARMv9_STM32N6xx_V1.0.0)
STM32CubeIDE (STM32CubeIDE 1.17.0)
STM32CubeProgrammer (v2.18.0)
STEdgeAI (v2.1.0)

Boot modes

The STM32N6 does not have any internal flash. To retain your firmware after a reboot, you must program it in the external flash. Alternatively, you can load your firmware directly from SRAM (dev mode). However, in dev mode if you turn off the board, your program will be lost.

Boot modes:

Dev mode: load firmware from debug session in RAM (boot switch to the right)
Boot from flash: Program firmware in external flash (boot switch to the left)

Preliminary Steps before any power measurement

rework board as explained in Power Measurement Setup
Switch BOOT1 switch to right position
Connect the DK board using a USB-C to USB-C cable.
Connect the STLINK_PWR as explained in chapter Power Measurement Setup.
update Utilities/pwr_scripts/example_configuration.yml file with the serial numbers of used devices (see update_config_file)
run command python ./capture.py power -c my_configuration.yml on to provide VddCore (see power_on_VddCore).

Note: This application does not use LCD. It only sends data over a VIRTUAL COM port, which is processed by the consumption measurement scripts.

Quickstart using prebuilt binaries

Three binaries must be programmed in the board external flash using the following procedure:

Switch BOOT1 switch to right position
Connect the DK board using a USB-C to USB-C cable.
Connect the STLINK_PWR as explained in chapter Power Measurement Setup.
Power the VddCore (see power_on_VddCore).
Open STM32CubeProgrammer and choose the serial number corresponding to the STLINK embedded in the DK (STM32N6570-DK).
Program Binary/ai_fsbl.hex (First stage boot loader)
Program Binary/network_data.hex (params of the network; To be changed only when the network is changed)
Program Binary/x-cube-n6-ai-power-measurement.hex (firmware application)
Switch BOOT1 switch to Left position
Power cycle the board
Power cycle the STLINK-V3PWR python ./capture.py power -c my_configuration.yml off and then python ./capture.py power -c my_configuration.yml on
use command python ./capture.py capture -c my_configuration.yml -w capture_full.csv to get power measurement data
generate report using command python ./full_sequence_power.py capture_full.csv
generate plot using command python ./capture.py display -r capture_full.csv

Note: This application does not use LCD. It only sends data over a VIRTUAL COM port, which is processed by the consumption measurement scripts.

Note: The Binary/x-cube-n6-ai-power-measurement.hex firmware is built for MB1939 STM32N6570-DK with IMX335 Camera module.

How to Program hex files using STM32CubeProgrammer UI

See How to Program hex files STM32CubeProgrammer

How to Program hex files using Command line

Make sure to have the STM32CubeProgrammer bin folder added in your path.

export DKEL="<STM32CubeProgrammer_N6 Install Folder>/bin/ExternalLoader/MX66UW1G45G_STM32N6570-DK.stldr"

# First Stage Boot loader
STM32_Programmer_CLI -c port=SWD mode=HOTPLUG -el $DKEL -hardRst -w Binary/ai_fsbl.hex

# Network parameters and biases
STM32_Programmer_CLI -c port=SWD mode=HOTPLUG -el $DKEL -hardRst -w Binary/network_data.hex

# Application Firmware
STM32_Programmer_CLI -c port=SWD mode=HOTPLUG -el $DKEL -hardRst -w Binary/x-cube-n6-ai-power-measurement.hex

Quickstart using source code

Before building and running the application you have to program network_data.hex (model weights and biases). This step only has to be done once unless you change AI model. See Quickstart using prebuilt binaries for details.

More information about boot modes is available at Boot Overview

Application build and run - Dev mode

Make sure to have the switch to the right side.

STM32CubeIDE

Double click on STM32CubeIDE/.project to open project in STM32CubeIDE. Build and run with build and run buttons.

IAR EWARM

Double click on EWARM/Project.eww to open project in IAR IDE. Build and run with build and run buttons.

Makefile

Before running commands below be sure to have commands in your PATH.

Build project using the provided Makefile:

make -j8

Open a gdb server connected to the STM32 target:

ST-LINK_gdbserver -p 61234 -l 1 -d -s -cp <path-to-stm32cubeprogramer-bin-dir> -m 1 -g

In a separate terminal session launch a GDB session to load the firmware image into the device memory:

$ arm-none-eabi-gdb build/Project.elf
(gdb) target remote :61234
(gdb) monitor reset
(gdb) load
(gdb) continue

Application build and run - Boot from flash

Make sure to have the switch to the right side.

Build the application

STM32CubeIDE

Double click on STM32CubeIDE/.project to open project in STM32CubeIDE. Build and run.

Known issue: Some warnings are interpreted as error (displayed in red). Build is still successful.

Makefile

Before running commands below be sure to have commands in your PATH.

Build project using the provided Makefile:

make -j8

Program the firmware in the external flash

Once your app is built with Makefile, STM32CubeIDE, or EWARM, you can add a signature to the bin file:

STM32_SigningTool_CLI -bin build/Project.bin -nk -t ssbl -hv 2.3 -o build/Project_sign.bin

and then you can program the bin file at the address 0x70100000;

export DKEL="<STM32CubeProgrammer_N6 Install Folder>/bin/ExternalLoader/MX66UW1G45G_STM32N6570-DK.stldr"

# Adapt build path to your IDE
STM32_Programmer_CLI -c port=SWD mode=HOTPLUG -el $DKEL -hardRst -w build/Project_sign.bin 0x70100000

Note: Only the App binary needs to be programmed if the fsbl and network_data.hex was previously programmed. Put the switch to the left side.

Do a power cycle to boot from the external flash. ( Power cycle the STLINK-V3PWR python ./capture.py power -c my_configuration.yml off and then python ./capture.py power -c my_configuration.yml on )