What Is a GPU?
Author: Domeny.ForSale editorial team Published
In brief: A GPU is a graphics processor: a computing chip designed to perform many similar operations simultaneously. Among other things, it creates images in games and graphics applications, but it is also used for scientific computing and artificial intelligence. It is not the same as an entire graphics card: it can be the card’s main component or part of a processor or larger integrated circuit.
How does a GPU work, and what is it used for?
GPU stands for Graphics Processing Unit. Its strength is parallel processing: instead of performing all calculations in sequence, the chip distributes the work among many computing units. This works well when a task can be divided into a large number of similar operations.
When rendering a game, the GPU processes the geometry of objects, applies textures—the images covering their surfaces—and calculates lighting. This produces successive frames. The more complex the scene and the higher the resolution, the more work the graphics chip usually has to do.
GPUs are also used beyond gaming:
- In video editing, they accelerate supported effects and image processing. Video encoding and decoding are often handled by specialised blocks within the chip.
- In 3D graphics, they help generate scene previews and render, or calculate, the finished image.
- In artificial intelligence, they perform the many mathematical operations needed to train and run models.
- In science and engineering, they accelerate certain simulations and calculations on large datasets.
However, simply having a GPU does not make every program faster. An application must support the chip and be able to transfer an appropriate portion of its work to it.
GPU vs CPU: what is the difference?
The CPU, or central processing unit, executes instructions from the operating system and applications and coordinates the computer’s work. It handles varied tasks, successive decisions and calculations in which the next step depends on the previous one well. A GPU, by contrast, is optimised for large numbers of parallel operations.
You can compare this to working with photos: the CPU efficiently manages opening files, handling menus and saving results, while the GPU can calculate the same effect for many pixels simultaneously. This is a simplification, but it shows why these chips complement each other rather than replace one another.
A more powerful GPU will not fix every performance problem. If a game is waiting for CPU calculations, buying a faster card may make little difference. Likewise, if there is insufficient RAM or data is being read slowly, the bottleneck lies elsewhere.
GPUs, graphics cards and integrated graphics
A graphics card is a component containing a GPU and usually its own memory, power circuitry, cooling system and display connectors. In a desktop computer, it most often takes the form of a circuit board installed in a motherboard slot. The GPU is therefore the processor on the card, not the name of the entire device.
An integrated GPU, also called an iGPU, is built into a processor or larger integrated circuit and usually uses memory shared with the CPU. This solution takes up little space and is often sufficient for web browsing, office work, watching videos and less demanding games. Its capabilities depend heavily on the specific model and memory configuration.
A discrete GPU, also referred to as a dedicated GPU, is a separate chip and usually has its own graphics memory—VRAM. It often delivers higher performance in demanding tasks, but it can increase power consumption, heat output and hardware costs. In a laptop, this chip is usually soldered to the board, so “dedicated” does not mean it can be easily replaced.
How can you assess what GPU you need?
Start with your applications and tasks, rather than the number of computing units or memory capacity alone. More VRAM does not automatically mean a faster GPU. Memory stores things such as textures, scene data or an AI model; a shortage can limit performance or prevent a task from being completed, but spare capacity alone does not increase computing power.
For example, a laptop used for documents, video calls and watching videos generally does not need a discrete GPU. The screen, battery life and amount of RAM may matter more when choosing one. For regular work on complex 3D scenes or demanding games, you will need to assess the performance of the specific chip.
Before buying, take these three steps:
- Check the requirements of your applications or games. Use the official documentation to confirm supported GPUs, required features and recommended memory.
- Compare tests that match your intended use. For games, choose your resolution and similar settings; for work, look for the application you use and the type of project you work on.
- Check the whole computer. In a desktop, space inside the case, the power supply and connectors matter. For a laptop, check tests of the exact model: power limits and cooling can significantly affect the performance of GPUs bearing the same name.
If you want to speed up your current computer, first monitor CPU, GPU and memory usage during the task that is causing problems. This will help you determine whether replacing the graphics chip actually addresses your needs.
See also: What Is 2FA?
Read also
