If you’re considering building or upgrading your PC, you might be wondering what the difference between a GPU and CPU is. These two acronyms refer to two different internal PC components.
The CPU, or ‘central processing unit’, is a chip that acts as the brain of the computer. It runs all the PC’s program instructions, reads from and writes to memory, and tells other components what to do.
The GPU, or ‘graphics processing unit’, is a chip that handles—you guessed it—graphics processing, performing tasks like rendering game frames and encoding videos. In this way, the GPU has a more specific job than the CPU.
You can run a PC without a GPU, but a PC without a CPU won’t be able to do much of anything. Nevertheless, there are benefits of having a GPU in your system—for example, and probably most importantly for many who are reading, having a GPU should greatly increase your FPS in games.
What is a CPU?
A CPU, or ‘central processing unit’, is a physical chip inside your PC that computes and executes all the instructions that different programs on your PC tell it to. It is, for all intents and purposes, the ‘computer’ itself.
How a CPU Works
The CPU processes data that you input, via mouse and keyboard and so on, in whatever way the program instructs it to. Your CPU then tells different hardware components what to do based on this processing—perhaps play a sound, or display something on screen, for example.
It loads program instructions from memory and sometimes executes these instructions in parallel—one instruction at a time on each CPU ‘core’. It performs operations at a certain speed which is measured in MHz (megahertz) or GHz (gigahertz). CPUs today can execute millions of instructions per second.
Your CPU delegates tasks to other components, too. For example, while the GPU might be what renders game frames, it will only do this if the CPU tells it to do so—just like how your arm will only move if your brain tells it to do so.
You can learn more about how CPUs work by reading our CPU Specifications Guide.
What is CPU Architecture?
CPU architecture is just what it sounds like: how the CPU is designed. Just like a building architect designs the layout and structure of a building, a CPU architect designs how the internal components of a CPU come together to make the entire CPU.
Different generations of Intel and AMD CPUs often share the same architectures. For example, Intel’s current 12-gen CPUs are all based on the ‘Alder Lake’ architecture, and AMD’s current (though soon to be replaced) 5000-series CPUs are all based on the Zen 3 architecture.
What this means is that different 5000-series CPUs, for example, might have different core counts and clock speeds, but the way these chips are logically designed is the same.
Simplifying greatly, CPUs are mostly made up of arithmetic logic units (ALUs), control units (CU), registers, and caches.
The ALU is made of circuits that perform logical and mathematical operations, the CU fetches and decodes instructions and tells the ALU what to do, the registers act as super-fast short-term memory for the CPU to store operation data in, and caches are slightly slower memory stores for data that the CPU might need to use repeatedly.
Modern CPUs are much more complicated than this, but they all have at least these four things in some form.
A change in architecture might change how these parts themselves are designed and how they come together to form the CPU.
What is a GPU?
A GPU, or ‘graphics processing unit’, is a physical chip inside your PC that’s dedicated to processing graphics. A GPU is not the same as a graphics card, but graphics cards have GPUs in them—a GPU is the chip inside the graphics card that does most the rendering legwork.
CPUs can also process graphics, but GPUs can do it much quicker thanks to the way they’re designed. They’re much better at computing a whole bunch of simple tasks very quickly or simultaneously, such as the calculations required for graphics rendering.
How a GPU Works
GPUs perform many simple tasks—such as geometric calculations for rendering—quickly and simultaneously.
They can do this because they have a great many more cores than CPUs do. An NVIDIA GeForce RTX 3080’s GPU, for example, has 8,704 separate CUDA Cores across 68 Stream Multiprocessors which are used for graphics rendering.
After the CPU has told it to do so, the GPU takes in a large amount of data from memory and performs rendering calculations on all this data simultaneously or in very quick succession.
What is GPU Architecture?
GPUs are designed by GPU architects just like CPUs are designed by CPU architects.
Comparing CPU vs GPU architecture, we can see that the two components are designed for very different purposes. CPUs are designed for complex, ‘serial’ (step-by-step) processing, and GPUs are designed for simple, ‘parallel’ (simultaneous) processing.
GPUs share similar designs to CPUs in some ways, however. Both have shared memory caches and compute cores, for example.
GPU cores have ALUs just like CPU cores, but these are designed to undertake much simpler mathematical tasks, and are therefore smaller, which allows many more of them to be placed on one chip.
Most GPU cores are designed for simple integer or floating-point calculations, but some cores are designed to handle deep learning or ray tracing.
CPU vs GPU Rendering
GPU rendering is much quicker than CPU rendering. A CPU without an integrated GPU (or a dedicated graphics card to delegate the task to) will probably render graphics much slower than a GPU could.
When a game frame needs rendering, many things need to happen. For example, the entire game scene’s geometry needs to be calculated, and so does its lighting. A lot of data also needs to be sampled.
These calculations are pretty easy for a computer to perform, but there are so many of them to perform for just one game frame that a CPU can struggle, because it can only perform a limited number of operations at a time.
But because a GPU has thousands of cores that can perform these operations simultaneously, it can render a game frame in no time. As such, GPUs tend to render graphics much quicker than CPUs.
The situations where this doesn’t hold and CPUs render quicker than GPUs are when the rendering in question is very complicated. For example, simulation rendering often runs quicker on the CPU than GPU because the CPU can handle the complexity of the simulation logic much better.
Integrated Graphics vs Dedicated
Some CPUs have GPUs built alongside them in the same physical chip, which means they have ‘integrated graphics’.
For example, AMD ‘G’ processors like the Ryzen 5 5600G are Accelerated Processing Units (APUs) that combine a CPU and GPU onto one chip. Many Intel processors have iGPUs (integrated GPUs), too, for example the Xe graphics built into many current-gen Alder Lake CPUs.
When using CPUs with integrated graphics, rendering is usually performed by the integrated GPU. So, note that in these cases, even though the graphics processing occurs on the processor’s physical chip, it’s still a GPU that’s performing rendering, not the CPU.
When a GPU isn’t located in the CPU chip, it will be in a separate piece of hardware called the graphics card. These GPUs will be ‘dedicated’ just to graphics processing, and because of this they’re often much more powerful than integrated GPUs.
Because integrated GPUs must be small to be placed on the same chip as the CPU, they’re usually less powerful than dedicated GPUs. To fit a GPU onto a CPU processor die, GPU core counts must be reduced, and graphics processing power is therefore limited. The number of CPU cores are also often reduced to fit the GPU onto the chip.
So, when comparing, say, integrated graphics vs an NVIDIA graphics card, the latter will almost always be more powerful for gaming.
Read more about this in our Dedicated Graphics Card vs Integrated Graphics guide.
GPU vs CPU Performance
A GPU processes graphics much quicker than a CPU does. But GPUs can’t perform many other general tasks that the CPU performs, such as running an operating system and complex applications.
There are also things that the CPU does better than the GPU for gaming.
Amongst other things, the CPU:
- Tells the GPU what it needs to render
- Deals with complex in-game physics
- Processes NPC logic
- Calculates damage and other stats
- Detects collision
- Computes player-vs-enemy or player-vs-player interactions
The GPU, on the other hand:
- Creates and transforms shapes out of vertex data
- Fragments these shapes into individual pixels
- Computes the lighting for each pixel
- Renders the graphics, shading millions of pixels each game frame
CPUs perform their tasks very well, and GPUs perform their own tasks very well. But the GPU and CPU can’t perform the others’ tasks very well at all. Ultimately, both are needed to run a game smoothly at a comfortable frame rate.
How much the CPU affects FPS in comparison to the GPU depends on the game in question.
Games that require lots of complex simulations or logic, such as 4X games like Civilization VI, will rely heavily on the CPU. And games that have very simple graphics and lighting won’t require as much GPU legwork.
Most games, however, require lots of GPU legwork to compute and render graphics. Therefore, providing you have a moderately capable CPU, GPU upgrades are likely to improve your FPS more than CPU upgrades for most games.
CPU vs GPU Bottleneck
Bottlenecking is when one component inside your PC limits your PC’s overall capabilities by acting like the bottle of a neck that prevents liquid coming out at a faster rate. If one component is slow, this might prevent other components from utilising their full power.
A game that’s being bottlenecked by the CPU or the GPU will suffer with lower FPS than the system is theoretically capable of if the bottlenecking component were upgraded. Both CPU and GPU bottlenecks can have the same symptoms—namely, low FPS.
Both CPUs and GPUs can bottleneck a system, but, when gaming, a GPU bottleneck is much more likely, because games are usually more GPU-bound.
To check whether a component is bottlenecking your system, you can load up monitoring software like HWMonitor64—or just check the performance tab of Windows Task Manager—and see whether the GPU or CPU is running at 100% utilisation while gaming. If it’s at 100%, it might be bottlenecking your system.
Read more about this in our guide on How to Fix Bottlenecking.
Summary: What’s the Difference Between a CPU and GPU?
Comparing a CPU vs GPU is like comparing an apple to an orange: they’re very different in some ways, but similar in others, and having both together makes for a well-rounded fruit bowl.
CPUs and GPUs are made from the same kinds of basic components but perform different tasks, and both together make for a much better graphical application experience.
The CPU is like the brain of the computer, processing complex program instructions one after the other (but often, to a limited extent, also in parallel) and telling other components what to do.
The GPU, on the other hand, is often dedicated to performing graphics processing tasks, because it’s much better at doing many simple tasks quickly and simultaneously, such as rendering millions of pixels.
A GPU can either be built into the CPU’s physical chip, giving the CPU integrated graphics, or it can live on a separate PCB inside your system as part of a dedicated graphics card. The latter allows more physical room for a more powerful GPU with more cores.
GPU vs CPU for Gaming
For gaming, a CPU is technically more necessary than a GPU. Without a CPU there’s nothing to read game instructions—in fact, there’s no ‘computer’ at all.
But once you have a CPU, you’re likely to see a bigger gaming performance increase by getting a CPU with an integrated GPU, or getting a dedicated graphics card, than by upgrading your CPU. Most games are very GPU-bound, meaning they rely on the GPU’s graphics processing power much more than the CPU’s complex processing power.
GPU vs CPU for General Use
For general use, most PC users will see a bigger performance increase by having a powerful CPU than by having a powerful GPU.
Most PC applications aren’t graphically intensive, and even the most basic integrated graphics of modern systems can easily handle everyday tasks like word processing and browsing the internet.
But GPUs can affect the performance of general applications, too—even basic operating system animations can feel a bit snappier when done by a GPU instead of a CPU.
Furthermore, certain non-game tasks can benefit greatly from a GPU. For example, video encoding can often be performed much quicker with a GPU helping the CPU out. If you stream, you might notice that NVIDIA’s NVENC gives you better streaming performance than pure CPU encoding does.
Ultimately, the best performing systems are the ones that have a CPU and dedicated graphics card. These systems should have all bases covered, and will be able to perform render-heavy tasks like gaming with ease.