What is integrated graphics?

Modern computer graphics, with high-definition video and detailed 3D rendering in video games and professional applications, is a demanding job. While a CPU can create (“render”) graphics, it lacks the correct architecture to do it quickly and efficiently.

This is why we have GPUs, built from the ground up to be great at the sort of math you need to calculate the correct values of millions of pixels dozens or hundreds of times every second. A GPU is in almost every computing device, from PCs to smartphones.

The only computers that don’t have GPUs are usually “headless” servers that are remote operated and only do work suited to CPUs. However, even that’s changing, since GPUs are now used to do things other than graphics.  Any computer that outputs to a display today almost certainly has a specialised GPU.

With a laptop or desktop computer, the specification sheet will often outline that the computer comes with “integrated” or “dedicated” graphics. There’s a big difference between these two approaches to including a GPU within a computer.

A dedicated GPU is like its own self-contained computer, with its own independent processor package and separate cooling solution. Dedicated GPUs also have their own power management hardware and memory.

In a desktop computer, dedicated GPUs come on their own circuit board, known as a graphics card. The card slots into the computer’s motherboard and often needs more power than can be provided through the card slot. So it may have its own dedicated power connections from the computer’s power supply.

Dedicated graphics in laptop computers sometimes come as a removable package, such as the (now discontinued) MXM laptop GPU modules.  It is more common that they are soldered directly to the mainboard, but still dedicated as separate components from the CPU, with their own cooling, memory and power.

An integrated GPU, on the other hand, has to share everything with the CPU. This is why it’s often called an onboard GPU.  It sits in the same processor package, cooled by the same heatsink and heat spreader and shares the same system memory as the CPU. The motherboard provides the display output hardware that allows you to hook up a monitor, but all the “brains” of the GPU are inside the CPU’s package.

Nearly all computer CPUs today have an integrated GPU, with the exception of high-end CPUs.  The integrated GPU model is the most common GPU type for a few reasons.

The first is cost. It doesn’t add that much cost to a CPU, to etch a GPU into your silicon real estate. Including a GPU in every CPU reduces costs in other parts of the system by far more than it increases the cost of the GPU itself.  Therefore systems that use integrated GPU are significantly cheaper than those with a dedicated solution.

The second reason is complexity which is especially relevant to laptop computers, where every cubic millimeter of volume matters. Integrating the GPU into the CPU package, means that laptops can be alot smaller since you don’t need all the extra supporting hardware to cool, power and connect an entirely separate chip package.

The third pillar of integrated GPU design is power efficiency. It’s much easier to manage the power draw of a single integrated chip than to balance the needs of two separate ones. Since the GPU and CPU are tightly integrated, they can make sure they fit neatly into the TDP (Thermal Design Power) of the CPU package.

Laptops with dedicated GPUs usually also have integrated GPUs, the operating system will dynamically switch between the two depending on the application you’re using which means you aren’t burning through battery power, by using a high-performance GPU while doing spreadsheets.

Dedicated cards have far more performance potential than integrated.

The physical size of a dedicated GPU can be bigger, have a much larger power budget and reach higher temperatures in operation safley. It also has access to specialised, high-performance memory designed for graphics applications.

For a long time, integrated GPUs have been synonymous with poor performance, best off for basic productivity tasks and limited multimedia.

However, since GPU technology has improved integrated graphics performance, to the point where most users, including casual gamers, don’t need dedicated GPUs, with a caveat, dedicated GPUs today are still much faster than integrated ones.

Another type of integrated GPU can be in SoCs or System-on-a-Chip devices. SoCs don’t just integrate the CPU and GPU into the same package, integrating the system memory and often even storage.

Since SoCs are designed with specific total performance targets in mind, the GPUs in them can be large. The Apple M1 SoC, for example, has more graphical power than a Playstation 4, with the CPU still managing to compete with high-end laptop CPUs.

Modern computer-game consoles integrate high-performance CPUs and GPUs into the same package, but are designed to dissipate the massive amount of heat this causes from the ground up.

If performance is the most important thing for you when it comes to graphics, then you should buy a desktop computer with dedicated graphics. If battery life, cost, heat and noise matter most to you, an integrated solution is likely what you need.

Desktop users have the option to add a dedicated GPU to their system, assuming your motherboard has the correct slot and your power supply and chassis are up to it. So you can try out the integrated GPU on your CPU to see if it’s good enough for your needs.

Ultimately, the most important thing is that you research the performance capabilities of the integrated GPU your prospective CPU has. You may be surprised at how much power is in that tiny processor.