Does STM32H56xx 1MB device has extra memory for high-cycle data? Part II

[Dimlite]

I create this post as my own (and others) experience contradicts the solution to the follow ticket. More information found in the replies section.

Please look into it and see if you agree.

https://community.st.com/t5/stm32-mcus-products/does-stm32h56xx-1mb-device-has-extra-memory-for-high-cycle-data/td-p/849391

[AScha.3]

>Does STM32H56xx 1MB device has extra memory for high-cycle data?

short: yes.

 

long: No, but you might “find” some more flash, as there are 2MB on the chip: so the “extra” memory is always there,

all same die, so its just a part of the 2MB on the chip:

So the flash is 2MB , but with H563xGTxx  you pay for 1MB (in 2x 512K banks) tested and working.

More memory is there, but at your own risk to use it, no guarantee to work.

And the high-cycle setting is in this flash area:  --- if you enable it , some extra rules apply for this area ---

 

So you see: the “extra” flash is just a part of the 2MB flash area.

[Dimlite]

Hi,

I am not sure that I’ve “found” extra memory. I’ve followed the instructions and enabled high-cycle flash. The big question when using a 1 MB device is: does this re-purpose the last 8 sectors as HC (meaning that the space available for application code shrinks by 8 sectors), or does this enable another 8 sectors of flash (meaning that I still have 64 sectors available application code).

Some people claims the first (see link to another thread). But my own tests indicates the latter.

If the first claim was correct, I should the same data in sector 127 and 63. But I don’t.

I am aware that sometimes it is possible to use parts of a device in another way than intended (at your own risk), but in this case I’ve followed the instructions and the result indicates that the device contains 512 + 48 kB in each bank.

[AScha.3]

Hi,

>or does this enable another 8 sectors of flash (meaning that I still have 64 sectors available application code).

Right, as shown, the high cycle area is at the end of the 1MB bank;

so if you have payed for a 2x 512K flash, the 8 sectors look like an “extra”, but are just always there, as its on the die.

And if you have the “2MB version” , you see it as part of the 2MB flash.

On the “1MB” flash its still at its/same address :

 

[Dimlite]

Hi,

I understand what you are saying, but this is not what I am asking about. The thread that I link to in my post is marked as solved with the following answer:

“ Enabling EDATA on 1MB, 2MB, and 512KB devices always reduces the available code memory, as the high-cycle data area is mapped to the last sectors of the flash.
The 1MB device does not gain "extra" high-cycle data memory; the 96KB is part of the existing flash, not additional. ”

My point is that this answer is not correct. On a 1 MB device, you actually gain another 96 kB when enabling high-cycle flash. It may not be available for application code, but it is certainly available for data storage (which is the intended use).