Article directory
How does the computer monitor display images? What is the relationship between CPU and GPU?
In the computer world, all data is just 0 or 1. There are only two real computing hardware in the computer, one is the CPU, and the other is the GPU (image processing chip, the core of the graphics card).
Graphics card function
The graphics card is connected to the computer motherboard, mainly to convert the digital signal of the computer into an analog signal for display on the monitor. At the same time, the graphics card also has image processing capabilities, which can assist the CPU to work and improve the overall operating speed.
For historical reasons, we can say that CPUs do all the work at the software level, while GPUs do at the hardware level. Hardware acceleration refers to a technology that reduces the workload of the central processing unit by allocating very large computational tasks to specialized hardware for processing in a computer. Especially in image processing this technique is often used.
3D hardware acceleration: Refers to the DirectX or OpenGL acceleration borrowed from the graphics card.
Why do you need a GPU when you have a CPU?
1. The CPU is mainly responsible for multi-task management and scheduling. It is the leader and responsible for command. GPU is mainly used for a large number of repeated calculations. It is a computing expert, but protocols and management are not its expertise. The task of the graphics card is assigned by the CPU.
2. If all the calculations are handed over to the CPU, then the CPU will be under a lot of pressure, especially the huge graphics calculations, and if there is a graphics card, then the CPU can hand over graphics-related tasks to the graphics card (specially set to handle image tasks hardware) processing, the CPU itself can use more resources to handle other more important things
How does the computer monitor display images?
Generally, the CPU submits the calculated and displayed content to the GPU. Of course, there are also CPUs that directly issue commands to let the GPU process the calculated and displayed content (hardware acceleration). After the monitor receives relevant signals, the video amplifier circuit shoots to the screen of the picture tube through the electron gun of the picture tube. This is the image we see!
The specific point is:
1. From the bus (Bus) into the GPU: the data sent by the CPU is sent to the North Bridge (main bridge) and then sent to the GPU for processing (rasterization).
2. From Video Chipset (graphics chipset) into Video RAM (video memory): send the data processed by the chip to the video memory.
3. Enter the Digital Analog Converter from the video memory (= RAM DAC, random read-write storage digital-analog converter): read the data from the video memory and send it to the RAM DAC for data conversion (digital signal to analog signal). But if it is a graphics card with a DVI interface type, it does not need to convert the digital signal to an analog signal. And directly output the digital signal.
RAM and VRAM
The memory plays a very important role in the computer. All programs running in the computer need to be executed through the memory. The function of the memory is to read the data in the hard disk in advance, so that when the CPU reads, it reads directly from the memory. Much sooner.
The graphics card is mainly composed of video memory, GPU, and cooling elements (fans or metal sheets). The video memory of the graphics card is equivalent to the memory of the computer, and the displayed image data is stored in it! Discrete graphics cards have independent video memory, while integrated graphics cards use CPU resources and occupy memory, that is, divide a part of the memory for video memory.
Video memory is used to store instructions and data called from the hard disk. Since the reading speed of the video memory is much faster than that of the hard disk, if the required data is transferred to the video memory in advance, the processing speed of the GPU will be greatly improved, and it will rarely get stuck when running large-scale games. The larger the video memory, the more data it can store, and the less it depends on the hard disk, which is good for improving the smoothness of the game.
Can all calculations be handed over to the GPU for processing?
In general, we can do everything with the CPU (the CPU mostly does things at the software level), but for image processing, it is usually faster to use hardware, because the GPU is optimized for highly parallel floating point operations using images. Therefore, the work of screen rendering should be handed over to the hardware as much as possible. The real problem is that the GPU does not have unlimited processing performance, and once the resources are exhausted, the performance will start to drop. So most animation performance optimizations are about intelligent utilization of GPU and CPU.