In computing terms, what is the difference between a CPU core and a GPU core?

Source: The content comes from " Expreview Super Network ", thank you .

When the PC was first introduced, there was no GPU, and all graphics calculations were done by the CPU. Later, people realized that the CPU was too slow to do graphics calculations, so they designed a special graphics accelerator card to help handle graphics calculations. Later, NVIDIA proposed the concept of GPU, which promoted the GPU to the status of a separate computing unit.

Image courtesy of NVIDIA

Since the two are aimed at different calculations, the CPU faces a variety of people's needs. For example, when you open a web page, the CPU is responsible for network interaction, parsing HTML files and JS scripts, building DOM trees, and loading various resources. Finally, it is the GPU's turn to render the screen and output it.

To give another example, when playing games, the CPU is responsible for completing the logic of the game program, loading the game data from the hard disk, and then starting to pick out graphics-related things and handing them over to the GPU. However, various calculations in the game still require the help of the CPU. For example, how common NPCs move still requires the CPU to calculate.

From the above two examples, we can see that the CPU is responsible for a variety of calculations. In order to cope with so many types of calculations, its versatility requirements are very high, which will inevitably lead to complex designs. For example, to support conditional statements such as if-else, some control units must be added specifically. Since its birth, GPUs have been facing a single type of calculation. In order to better cope with single graphics computing tasks, early GPUs adopted a pipeline design. By stacking a large number of chips on a chip scale, the graphics processing speed can be improved, thereby providing better image quality. It is simple and crude.

So the answer to the question is obvious. In terms of computing, CPU is more versatile and can handle all kinds of work, but its speed is not very fast. GPU can only handle tasks such as graphics computing, and its speed is very fast, or it can also take tasks specially written for it to calculate. This is the origin of GPGPU (general-purpose computing performed on graphics computing unit).