2026-05-07 11:24 AM - last edited on 2026-05-08 6:11 AM by Andrew Neil
Hi everyone,
I'm encountering a blocking issue while trying to start an inference on a NUCLEO-N657X0-Q using STM32Cube AI Studio and STM32CubeProgrammer.
The connection to the board via SWD seems successful, but the process fails during the memory programming phase.
2026-05-08 4:53 AM
This is a very common STM32N6 issue it seems. The main problem is that the STM32N657 has no internal flash. So STM32CubeProgrammer needs the proper external loader (.stldr) for the external XSPI flash on the NUCLEO board. Usually when you get the message “Your error flash loader can’t be loaded” it means the loader is missing, wrong or the board boot mode is wrong.
Give these fixes a shot:
Many users just solved the exact same problem by changing the boot jumper configuration and choosing the correct external loader.
2026-05-08 5:58 AM
Hi, thank you very much for the detailed response!
Since this is my first time using an STM32 board (and the N6 specifically), your suggestions were invaluable in helping me understand the issues related to the lack of internal flash memory.
Regarding this, I have two technical questions to ensure I proceed correctly:
Frequency Configuration in AI Studio: Is there a way to set the SWD speed to 1000-2000 KHz directly within STM32CubeAI Studio? Or is it better to bypass the tool's automatic validation and use STM32CubeProgrammer exclusively to manually manage the connection and the loader?
Flashing Procedure: If I understand correctly, once the board is unlocked and the External Loader is set, should I proceed with manually flashing the .bin or .elf file generated by AI Studio? In my case, the log indicates the Octo-SPI address is 0x71000000.
Thanks again for the support!