Both PS5 and Xbox Series AMD Ryzen CPUs and AMD Radeon GPUsneither of them has an Intel processor or an NVIDIA GPU. The same thing happened with the consoles of the previous generation, both the standard versions and the improved ones that came later, because they all repeated this configuration with AMD solutions.

However, the Nintendo Switch uses a NVIDIA GPUs, although its CPU is based on ARM architecture. It’s the exception to the rule, and it looks like it will continue to be the case, as everything points to the Nintendo Switch 2 once again featuring an ARM CPU and NVIDIA GPU based on a more advanced architecture.

In contrast, the PS5 Pro will once again use AMD Ryzen CPUs and Radeon GPUs, and it’s highly likely that next-gen consoles from Sony and Microsoft will repeat the same strategy, but do you know why? The answer is very interesting and gives us a lot of valuable information about the past, present and future of consoles. For that reason, I have prepared this special article where we will talk about this very interesting topic.

Consoles before the APU revolution

PS3 inside. Image courtesy of Copetti.org.

It seems like a long time ago, but the truth is that only three generations ago they still had consoles clear division at CPU and GPU level. The processor had its own silicon chip and its own packaging, and it was no different for the GPU. This interior design had its advantages because:

• Each component had its own TDP.
• They could take up more space at the silicon level.
• It was possible to incorporate more powerful and complex GPUs and CPUs.

This structure was used until the arrival of PS4 and Xbox One not because it was the best option, but because it was actually the only one. In the era of PS3 and Xbox 360, there were no solutions capable of integrating CPU and GPU in the same package with enough power and efficiency to create a gaming system with a good level of performance.

Although these designs had their advantages, they also had their disadvantages:

• The separate integration of CPU and GPU complicates the design internally.
• It also makes it difficult to design an efficient cooling system.
• It increases production costs and complicates supply, especially if you depend on two different brands.
• It can also complicate video game development, although this will depend on the architectures used.

There are many examples of what is possible with a separate CPU and GPU design, but one of my favorites is The original Xbox. This console was a true monster, the most powerful of its generation, and during the early phase of the PS3, when it was still very difficult to harness its true power, the games on the aforementioned Xbox had nothing to envy.

Original Xbox inside. Image courtesy of Copetti.org.

The Xbox had a processor Intel Pentium III custom capable of working at 733 MHz, and GeForce 3 Ti GPU customized and 64 MB RAM. Thanks to the optimizations that its development kits had and its operating system that was extremely lightweight, this console was able to run even games that were made for the next generation, such as DOOM III and Half-Life 2.

This design was carried over to the Xbox 360, which featured a triple-core IBM PowerPC CPU and a custom AMD GPU with 240 first-generation unified shaders, and the PS3, which had an IBM Cell CPU and an NVIDIA GeForce 7900 GPU GT, which still retained the split between pixel shaders and vertex shaders.

In previous generations, consoles did not differ much from this structure. The split between the processor and the graphics core was evident, although it was also normal to find support chips and co-processors. For example, the PS1 had a MIPS CPU and GTE GPU built on a 600nm node, which is a fairly conventional design, but the Sega Saturn used two Hitachi SH2 processors.

CPU and GPU integration: the first steps and the first APU

Wii U CPU and GPU in the same package but with different packages.

The first console to use a solution integrating CPU and GPU was Wii U. This console had an IBM PowerPC processor and an AMD Radeon graphics core with 160 shader units. Both shared covers, but They had their own encapsulationso this solution did not fit into the APU concept, where the CPU and GPU are on the same silicon chip and therefore in the same package.

A year after the launch of this console, the PS4 and Xbox One arrived, two consoles that marked a real revolution by fully embracing the APU as the definitive solution to remove the separation between the processor and the graphics core. Both components have been integrated on the same silicon chipand represented a huge evolution that was only possible due to AMD’s commitment to its own designs.

The PS4 APU managed to integrate the performance of a mid-range GPU of its generation in a package of only 348 mm2. Within this package we had a graphics core capable of more or less performance at the Radeon HD 7850 levelalthough it used a more advanced architecture (GCN 2.0) than this and supported more graphics features.

In the case of the Xbox One, we had the graphics core less powerful which approached the Radeon HD 7770 in raw performance and was limited to the GCN 1.0 architecture. The differences between the two consoles in terms of graphical performance were evident from the first moment, and the PS4 came out as the big winner of this generation.

The PS4 APU is highlighted in red. CPU and GPU in the same package. Image courtesy of iFixit.

In order to integrate such a powerful GPU into a silicon chip shared with the CPU, many advances and many changes were needed, although the most important were two: the jump to 28nm node enabled an unprecedented level of transistor miniaturization and performance was sacrificed at the CPU level give full priority to the graphics core, which is, after all, the most important part of the console.

By limiting the complexity of the CPU, this was possible reduce its silicon footprint as well as resource consumption. To do this, AMD used the Jaguar architecture, which was the equivalent of the Intel Atom processor architecture, a low-power, low-power solution that had nothing to do with the most advanced and powerful architectures of the time. .

If AMD had not decided on the Jaguar architecture, it would not have been able to design the PS4 and Xbox One APUs as we know them, as it would be possible to integrate such a powerful and complex GPU with a more powerful CPU based on, for example, the Piledriver architecture. It was configured with that Jaguar processor eight cores at low frequencyand development kits with low-level APIs did their magic.

At first they raised doubts, but they lived up to expectations. The PS4 APU allowed you to move games inside 1080p at 30 FPS in almost all cases throughout its life, and the Xbox One usually hovered between 900p and 720p and 30 FPS due to lower performance. Good results The fact that AMD APUs were acquired made both Sony and Microsoft decide to bet on them again to design the PS5 and Xbox Series X.

PS5 and Xbox Series X: the consecration of APUs and goodbye to NVIDIA

The PS5 APU, which once again integrates the CPU and GPU in the same package.

The move to an APU-based design allowed Sony and Microsoft to simplify the architecture and construction of the PS4 and Xbox One. reduced costs, It made the design and supply contract negotiations easier, and both companies depended solely on AMD to have a good level of APU inventory that they needed to build their consoles.

There is no doubt that with the new APU-based design there were all the advantages, both at the architectural level and at the cost level. Restoring this design with the PS5 and Xbox Series terminate to designs based on separate and independent CPUs and GPUs.

I have already explained the positives and negatives that both approaches have. Finally for Sony and Microsoft the question of cost is the most important during console manufacturing. It should also be taken into account that the APUs are used by the PS5 and Xbox Series

Portable consoles have greatly benefited from the evolution of APUs, a reality that has allowed the creation of very powerful systems that are capable of playing current games at 1080p resolution. even better than the previous generation of consoles. I could give many examples, but without a doubt the most famous ones are ASUS ROG Ally and Steam Deck.

The Nintendo Switch was a very special exception, as it was the only portable console to opt for an ARM CPU and NVIDIA GPU, based on the veteran Maxwell architecture. There is an explanation for this, and that is NVIDIA could provide an SoC solution that integrated everything Nintendo needed at that time and at a very reasonable pricebecause the green giant needed to release its Tegra X1 chip.

If NVIDIA only offered Nintendo a GPU solution for its hybrid console, it wouldn’t accept it and would look for another alternative that would at least include a CPU and GPU on a single chip. In any case, it is important that although the Nintendo Switch uses an NVIDIA graphics core In the end, your bet is the same than what we’ve seen in the last two generations, combining CPU and GPU on a single chip to simplify design and reduce costs.

So why don’t the PS5 and Xbox Series X use Intel processors and NVIDIA GPUs?

Well, it’s very simple to do They would have to go back to the PS3 and Xbox 360 era design, This means they would have to implement a separate Intel CPU and NVIDIA GPU on two custom-packaged chips, as there is no APU that integrates both components from these companies.

This would make the design of future consoles with two independent chips more expensive and complicated, forcing Microsoft and Sony to do the same negotiate two contracts customization and supplies, one for the CPU and the other for the GPU. Finally, it should also be considered that switching to an Intel processor and NVIDIA GPU would complicate maintenance backward compatibility new consoles with previous generations.

At that time I shared an article about how could it be Xbox Next with an Intel processor and NVIDIA GPU, where I also cover this topic, and I invite you to look at it if you have not had the opportunity to read it, because it is very interesting and will not leave you indifferent.

If Intel and NVIDIA were to reach some kind of agreement work together to design a solution that integrates CPU and GPU in a single package Things could get very interesting, especially since NVIDIA has the most advanced graphics architecture that exists for working with ray tracing, and because it also has unique technologies within the NVIDIA DLSS 3.5 ecosystem.

I know some of you will think it’s impossible, but the truth is that if Intel and AMD could work together to create Intel Kaby Lake G with Radeon graphics I think we can’t rule out that the chip giant and the green giant could work together one day. I’m not saying it will happen, I’m just saying it’s not impossible and with how important the console market is, it could be a worthwhile adventure for both companies.

It will surely be the only next-gen console to have an NVIDIA GPU Nintendo Switch 2. It won’t directly compete with the PS5 and Xbox Series