Von Neumann structure and Harvard structure

The Harvard structure is that data and code exist separately.

Von Neumann structure was put forward after the Harvard structure, von Neumann proposed "code itself is a kind of data", solve the Harvard structure of a problem-bus take up too much resources.

The von Neumann structure is that data and code are put together.

First, all the programs (data and code) in the hard drive (Flash) are copied to ram through the BIOS (Uboot). So at this point

Ram internals are divided into segments-code snippets, date segments, BSS segments, and more. This saves a set of external data bus and address buses compared to the Harvard structure.

But the downside is that it slows down. Originally the data and code are stored separately, they correspond to the "two high-speed channel." Now, the data and code are all put together to correspond to "one high-speed channel."

With the development of the semiconductor industry, many things can be integrated into the chip inside. The wire inside the chip is much lower than the cost of the circuit outside the PCB.

In other words, the Harvard structure is well suited to be designed inside the chip. The von Neumann structure still exists on the outside of the chip.

As a von Neumann, there is now a mix of structures with the Harvard structure:

This structure is the current structure of arm, the two structures to the long, to avoid its short. Where the cache inside the chip indicates cache (the word cache is said to come from French, is the cache invented by the French?) Ha ha. Dcache is used to cache part of the code, Icache to cache part of the data. The cache will load new data into RAM only if it needs to be changed. So most of the time the CPU is through the Harvard structure and cache (cache) communication, this speed is very fast ~ ~

In this way, outside the chip, the use of von Neumann structure, saving the external PCB trace resources.

Inside the chip, the use of the Harvard structure increases the speed at which the CPU accesses data. It's the same with both worlds ~

Von Neumann structure and Harvard structure