Best practices for splitting tasks between M7 (GUI) and M4 (backend) on STM32H745I-DISCO Body:

[Jerrythottathil]

Hello,
I’m working with the STM32H745I-DISCO board and I’d like to use the Cortex-M7 core for GUI (TouchGFX/LVGL) and the Cortex-M4 core for backend tasks like sensor data handling, communication protocols (CAN, UART, I2C), and preprocessing.

I’m looking for guidance and resources on:

• Recommended project structure in STM32CubeIDE / CubeMX for dual-core applications.

• How to set up inter-core communication (M7 ↔ M4) (OpenAMP, shared memory, or mailboxes).

• Examples or reference projects where M7 is dedicated to GUI and M4 to background tasks.

• Any best practices for memory allocation (framebuffer vs shared RAM) and FreeRTOS usage across cores.

If anyone has worked on a similar setup or can point me to example projects, tutorials, or documentation, that would be really helpful.

Thank you!

[Osman SOYKURT]

Hello @Jerrythottathil ,

• We also recommend to use the Cortex-M7 core for TouchGFX on a dual core MCU (m4 + m7)
  
  
• You can find an example of core communication in the firmware pack of the STM32H7 family 
  (STM32Cube\Repository\STM32Cube_FW_H7_V1.12.1\Projects\STM32H745I-DISCO\Applications\OpenAMP for instance)
  
  
• Yes you can find examples of it in the TouchGFX Designer. These 3 TBS are using dual cores :
  
  
• We recommend to use the fastest memory with the CM7 core and modify your linker scripts to define memory regions explicitly for each core.