cancel
Showing results for 
Search instead for 
Did you mean: 

X-CUBE-AI not available in CubeMX for the STM32H747I-DISCO?

AUzpa
Associate II

Hello,

When I create a new project in CubeMX for the STM32H747I-DISCO board, I cannot see the X-CUBE-AI package when I click on "Additional Software". Is it because this board has 2 cores Cortex-M7 and M4? Somehow we need to specify on which core the network should run?

In the user manual of FP-AI-VISION1 (UM2611: https://my.st.com/content/ccc/resource/technical/document/user_manual/group1/a0/53/9b/b7/f9/b5/4a/6f/DM00630755/files/DM00630755.pdf/jcr:content/translations/en.DM00630755.pdf), it is written that:

"For users having selected dual-core MCUs for their application but running it on the Cortex®-M7 core only: when dual-core MCUs like the STM32H747 are not supported by the STM32CubeMX tool, they must select a single core MCU like the STM32H743 to be able to generate the code for the Cortex®-M7 core."

So, if I generate an X-CUBE-AI project in CubeMX for, let's say, the STM32H743I-EVAL2 or STM32H743I-EVAL board, would that project compile and work on the STM32H747I-DISCO as well, without any other modification? This seems counter-intuitive to me since the EVAL board contains more features than the DISCO one.

Should I select an MCU (STM32H743xx) instead of an EVAL board, but which one would be the best? I would still need to use the LCD screen (as well as the external Flash memory and RAM) of the H747-DISCO board for my project.

Thank you and kind regards

1 ACCEPTED SOLUTION

Accepted Solutions

>>..would that project compile and work on the STM32H747I-DISCO as well, without any other modification? 

Usually no, what one might consider is making a simple shell project on the prefered target, and then merge in the non-platform specific portions

Tips, Buy me a coffee, or three.. PayPal Venmo
Up vote any posts that you find helpful, it shows what's working..

View solution in original post

3 REPLIES 3
Julien CLOSS
ST Employee

Hello,

Dual-core STM32H7xx products (ex: STM32H747) are not supported by the X-CUBE-AI package at the moment.

So if you want to use X-CUBE-AI package in order to generate neural network code running on the Cortex-M7 core of the STM32H747, you have to select a single core STM32H7xx product (ex: STM32H743).

In X-CUBE-AI package: if you have selected application SystemPerformance or Validation, you need to perform the following changes in order to be able to run the application on the STM32H747-DISCO board:

  • In STM32CubeMX, in USART1configuration: you have to change the pin name (from PB6/PB7 for ex . on STM32H743) to PA9/PA10.
  • On the STM32H747-DISCO board: you have to rework the board (according to the description provided in chapter 5.2.3 of the UM2411) in order to have the hardware aligned with the generated software for the power supply part.

Best Regards,

Julien. 

>>..would that project compile and work on the STM32H747I-DISCO as well, without any other modification? 

Usually no, what one might consider is making a simple shell project on the prefered target, and then merge in the non-platform specific portions

Tips, Buy me a coffee, or three.. PayPal Venmo
Up vote any posts that you find helpful, it shows what's working..
victagayun
Senior III

Just curious, how do you init the GPIO on CubeMX for  STM32H747I-DISCO? I have the problem in the link below.

https://community.st.com/s/question/0D53W00000NUOGLSA5/cubemx-gpio-not-initialized-in-stm32h747disco