CCM-the ultimate tool for maximizing STM32F4 performance

Reproduced

Tags: stm32 ccm it

Category: Embedded

CCM (core coupled Memory) is dedicated to the F4 core full-speed 64KB RAM, they do not go through the bus matrix, the F4 core directly connected to it, the address space in the 0x1000_0000 ~ 0x1000_ffff.

Because of its address space and conventional SRAM is not continuous, coupled with DMA and peripherals can not directly use them, it makes the general user program completely ignore this 64KB full speed ram, sin ah ...

Anyway, since it is full-speed and dedicated SRAM, you can use this to give full play to F4 performance:

1. Make a stack.

Modify the link script to find ccmram_region, Block Cstack, and block HEAP keywords.

Change the order of the original place in Ram_region {...} to the following 2 sentences:

Place in ram_region {readwrite};

Place in Ccmram_region {block cstack, block HEAP};

Get!

Especially when running an rtos, such stacks can be managed as privileged resources, and security and performance are improved synchronously.

2. Do high-speed arithmetic buffering

For computationally intensive applications, a dedicated full-speed operation buffer can also be partitioned in CCM by customizing the link script.

For example: FOC model decoupling, network SSL, JPEG encoding/decoding ...

At this point, the F4 kernel is truly capable of performing the highest computational performance.

Let's talk about this.