01What Exactly Is a GPU?
The full English name for GPU is Graphics Processing Unit, usually called graphics processor in Chinese.
The reason it's closely related to "graphics" is because many graphics tasks require completing numerous, similar, and simultaneously doable calculations in a very short time, and this is exactly what GPUs are good at.
Process Lots of Graphics Data
Such as positions, colors, lighting, etc.
Process Many Pixels Simultaneously
Screens have many pixels that need calculation
Fast Repetitive Calculations
Suitable for many structurally similar tasks
02Why Are Graphics Tasks Particularly Suited for GPUs?
A single image might contain millions of pixels. Many pixels need color calculations, and in 3D games, many vertices, lighting, textures, and other graphics data also need to be processed.
03What's the Difference Between CPU and GPU?
Both CPU and GPU can perform calculations, but they excel at different tasks. We can't simply say "GPU is faster than CPU" or "CPU is stronger than GPU"; it depends on the specific task.
CPU
- Excels at complex, flexible control
- Suitable for many different types of tasks
- Often responsible for the overall flow of programs
- Individual processing cores usually more powerful
GPU
- Excels at many similar calculations
- Suitable for highly parallel tasks
- Graphics rendering is a typical scenario
- Has many computing resources oriented toward parallel work
In real computers, they usually work together: CPU runs game logic, operating system, and other programs, GPU handles many graphics calculations, and both sides exchange data.
04How Is a Frame Created?
The 3D world in games isn't directly "mounted" on the screen. The computer calculates what should ultimately be displayed based on scene data, step by step.
Scene Data
Object positions, shapes, materials, etc.
Graphics Calculations
Calculate positions, lighting, textures, etc.
Form Pixels
Determine the colors of many pixels on the screen
Display Image
Final image sent to display system
05What Is VRAM For?
Many dedicated graphics cards have dedicated high-speed storage space nearby, usually called VRAM (Video RAM).
Graphics Data
Textures, frame buffers, etc.
VRAM
Stores data the GPU needs to access quickly
GPU
Reads and processes this data
06Do GPUs Really Only "Draw"?
No. GPUs were initially developed for graphics processing, but now they are also often used for scientific computing, video processing, artificial intelligence, and other parallel computing tasks.
Graphics and Games
Render 2D, 3D images
Video Processing
Some video encoding/decoding and image processing
Scientific Computing
Process many parallelizable numerical calculations
Artificial Intelligence
Train and run many neural network models
So a more accurate statement isn't "graphics cards can draw," but: GPUs are particularly good at massive parallel computing, and graphics tasks happen to need this capability.
What You've Learned
- GPU is a graphics processor, and graphics rendering is one of its important uses.
- GPUs excel at many similar, simultaneously performable parallel calculations.
- Graphics tasks contain many vertex, pixel, lighting, and texture calculations, making them suitable for GPUs.
- CPU and GPU aren't about who is "absolutely faster," but about excelling at different types of tasks.
- A frame can be understood as scene data that goes through graphics calculations and finally becomes pixels on the screen.
- Dedicated graphics cards usually have VRAM, while integrated GPUs often share system memory.
- Besides graphics, GPUs can also be used for video, scientific computing, and artificial intelligence tasks.
Next Question: Inside the computer, data can already be processed, but how do keyboards and mice tell the computer about our operations? Next article: How Do Keyboards and Mice Tell Computers What We're Thinking?