[STM32] Understanding of FSMC

[Fever]

Hello,

I am using an STM32L496ZG, and 1MB of SRAM (this one : **broken link removed**)

My first idea was to use it with GPIO registers to control it. Like using GPIO_ODT to write the address I want to write to, and ODT again to write the data at this place.

I guess for SRAM it is a very slow way to do things, since I have to do everything manually.
I also guess that, to write 1 byte to the SRAM, I have to do two steps. Write the address, then write the data. It seems very inefficient to me.

I read about FSMC on STM32 devices, and I'm a bit intrigued. The reference manual explains how to use it well, but does not really explain what is it good for.
I saw on internet that a lot of people uses it for SRAM, FLASH, TFT ect, and I wanted to know if it is a good idea for me to use it.

How does it help to use FSMC ? Is it more efficient than using GPIOs as described before ? If so, why ?

Thank you very much for your help

 

[FvM]

FSMC implements a memory mapped interface, you'll access any memory location the same way as internal memory. I presume, you know how this works programmatically, so you'll know why it's more effective than using GPIO.

 

[Fever]

I guess I understand.

Having a given memory location (e.g 0x6000) means that I can write code like :

(*(char *)0x6000) = 'H';
(*(char *)0x6001) = 'i';
(*(char *)0x6002) = '!';

(this is just an example, I wouldn't code it like that)

So if I understand it correctly, it means that FSMC cares about chip select, OE, WE etc etc on the hardware directly. So no need to drive all the commands manually, saving quite a lot of instructions I guess.

Am I right ?

If so, how does the MCU knows that from the address 0x6000 it has to use FSMC rather than embedded SRAM from the MCU ?

Thanks

 

[FvM]

FSMC has individual memory range for each chip select, review the FSMC chapter in ST32 reference manual.

Rather than accessing individual bytes, you'll usually assign complete C structures or arrays at external memory range.