With PS5 and Xbox Series X, Sony and Microsoft have maintained the same foundation and approach that we've seen on PS4 and Xbox One, but the two companies' strategy in this new generation It was very different, and in the end the production problems, and the shortage of consoles, have ended weighing down a transition period which, under normal conditions, would now be very close to closing.

Today, with that almost year and a half of context on the table, we can appreciate the work that Sony and Microsoft did with PS5 and Xbox Series X, how the ecosystem of the two consoles has evolved and choose whether, in the end, they were up to the task or not. Personally, I think that Not only have they not met expectations, but deep down they remained quite below what was supposed to be.

It's a topic we've covered in other articles, but in the end it's not at all difficult to argue. It's enough to mention that PS5 and Xbox Series X They are not systems capable of working perfectly with native resolutions in 4K and 60 fixed FPS, while they continue to resort to rescaling or active resolution, and that ray tracing is something that it's too big for them.

I don't want to repeat myself on issues we've covered, but all this information we've seen is what is required to have the context we need to investigate five things that Sony and Microsoft did wrong with PS5 and Xbox Series X. This is not supported by an opinion, but rather in truth that we saw over the nearly year and a half that the two consoles have been on the market, and also in the five problems that we have seen in the article.

1.-Using an APU in PS5 and Xbox Series X was a mistake

It is becoming increasingly clear to me that the change that was caused by going from PS3 and Xbox 360 to PS4 and Xbox One was a serious failure. I am referring to the abandonment of a design with central processing unit and GPU distinguished in different packages to an APU, a solution that integrates the central processing unit and the GPU in a single package, and I will explain why.

Using a semi-custom APU makes design easier of new consoles, and can also contribute to reducing costs, that is not open to discussion and is a good thing. However, when integrating the CPU and GPU in the same package, fundamental restrictions come into play:

• The two elements They must communicate space at the silicon level, which means that in order to fit a CPU and GPU into a single casing, essential elements have to be cut. This results in a loss of performance and essential specs, to the point where, for example, the PS5's Zen 2 CPU only has 8MB of L3, and its RDNA2 GPU comes with no limitless cache.
• In addition They distribute the free TDP, something that our Sony explained during the official presentation of PS5, when it mentioned that the GPU speed of that console can reach 2.29 GHz of maximum speed, but that this is active and that it adjusts depending on the workload, and that it can be reduced if the central processing unit has a greater weight.

As I said at the beginning of this chapter Xbox 360 and PS3 had a very different design, in which Both the central processing unit and the GPU were integrated into completely different packages, which allowed those two essential restrictions we mentioned to be overcome. So much so that both consoles had GPUs that, at the time, were among the most cutting-edge today, something that did not happen with PS5 and Xbox Series X.

2.- Use architectures that exist in the PC world

Don't get me wrong, there are no drawbacks to the architectures that exist in the PC world, but they are designed to be incorporated in a way that looks and has power capabilities that are not possible in a console. On the other hand, using these architectures prevents consoles from achieving have superior technologies or elements to those who are in the PC field at each and every precise moment.

Think, for example, of PS3 and its Cell central processing unit, in which XDR memory of that console, or on the Xbox 360 GPU, which It used an early unified shader architecture in 2005.. It was striking because both consoles had much more advanced features than the PCs of the time. This allowed developers to find essential advances as they learned to exploit the true potential of those consoles, and it also had a positive impact on their useful life. You only have to think, for example, at the quality of the latest games that PS3 and Xbox 360 have received.

PS5 and Xbox Series X use architectures that exist in the PC planet, which means that truly They do not contribute anything revolutionary. That differential value was eliminated with PS4 and Xbox One, and the same has happened with the new generation of consoles. RDNA2 architecture and ray tracing? Hardware-accelerated ray tracing arrived on PC in 2018, and budget 8-core, 16-thread processors have been a thing since 2017.

The same applies to SSDs and 4K resolution. The GTX 780 Ti from 2013 could now run games at 4K, and in the end we are left with a picture that It does not represent any real progress compared to PCs from several years ago. If we add to this what we have said in the first point, those restrictions derived from using an APU, we will realize that PS3 and Xbox 360 marked "the culmination" of the term console, and that from there we have gone "downhill."

3.- Associate ray tracing with PS5 and Xbox Series X

Strictly speaking, PS5 and Xbox Series X have dedicated hardware to speed up ray tracing, but they are not able to work with this technology in a truly seamless way. The main reason is in the very restrictions that architecture shows that use the GPUs of the two consoles, and that are also present in the Radeon RX 6000, based on the RDNA 2 architecture.

Ray tracing acceleration is performed by applied cores that are distributed in a ratio of one core per computing unit. Thus, PS5 and Xbox Series X They have a total of 36 and 52 cores for ray tracing, respectively. The downside is that all of these ray tracing cores have some essential shortcomings that put them below the level of the RT cores found in the GeForce RTX 20:

• Each core for ray tracing Shares elements with texturing entities, which means that they cannot work simultaneously.
• Ray tracing cores work with ray-triangle intersections and box-bounding intersections. It is true that these are the most intensive and consume the most resources, but BVH cross-intersections, which are one step before those, are performed in the shaders, which means that these are not completely free from ray tracing.
• Finally, they are useless to work asynchronously (independent of shaders), which can produce a bottleneck at several levels.

If we add to this that the PS5 and Xbox Series X GPUs do not have unlimited cache, we realize the problem we have at hand, and it is very easy to understand that the developers they should have limited the implementation of ray tracing in games for both consoles.

I could give several examples, but in the end the titles that best prove that ray tracing is too big for PS5 and Xbox Series X are Dying Light 2, which runs on 1080p and 30 FPS on both consoles with ray tracing limited to shadows and ambient occlusion, and Cyberpunk 2077, which also runs at 1080p and 30 FPS with ray tracing similarly limited to shadows and ambient occlusion.

4.-Talk about performance scenarios that they do not have the capacity to achieve

This isn't the first time Sony and Microsoft have exaggerated the capabilities of a specific generation's consoles, but with PS5 and Xbox Series X the situation has been raised to levels that They bordered on the absurd. Remember, for example, when the PS5 was billed as an 8K console, a resolution that a powerful RTX 3090 isn't capable of comfortably scrolling at, or when it was claimed that the Xbox Series X's GPU would be on par with an RTX 2080 Ti.

This type of action represents a major drawback because, in the end, They develop hopes among the players that are not going to be fulfilled., and this is not going to do much more than produce more or less deep disappointments. In the end, PS5 and Xbox Series X are consoles that do not feel really comfortable working with 4K resolutions and very high qualities, since they generally do not reach 60 FPS unless they use rescaling or active resolution.

Again, I could give several examples, but I'm going to select one of the most popular titles currently, and one of the worst optimized on PC to best illustrate this problem, Elden Ring. Said game runs in 4K and at 30 FPS on PS5 and Xbox Series X, a level that equivalent to a PC with a 6GB RTX 2060 with the game set to its limit.

We're not getting into ray tracing here as we've already dedicated a separate section to it, but it would be interesting to see how the performance of PS5 and Xbox Series X evolves when it's released. the transition to the Unreal Engine 5 graphics engine and intergenerational developments are abandoned. I think we are in for a surprise, and not a good one.

5.-Use only 16 GB of unified memory

Memory represents an essential section of any system. Xbox 360 had an essential advantage over PS3, and it happens that that used a total of 512 MB unified, while the PS3 had two blocks of 256 MB each, one of the XDR type available for the central processing unit and another of the GDDR3 type available for the GPU. This strict division complicated the work of the programmers, and was the trigger for the jump to a unified architecture.

Having a single unified memory block has an essential advantage, and it happens that programmers They have the possibility to distribute the free memory as they see fit.. Thus, a specific game may require a larger proportion of memory allocated to data for the central processing unit, which would be the work of the RAM in a PC, while another may require much more memory allocated to textures and other elements used by the GPU, what we know as graphic memory or VRAM in the PC world.

A PC that has 8 GB of RAM and 8 GB of VRAM memory It will be less flexible than a console that has 16 GB of unified memory, because it will be able to work with both types of memory as there is no such strict division. However, we must not forget that, in the end, The proportion of free memory has a fundamental weight, and that this tends to be among the elements that begins to show the passage of time first.

PS4 and Xbox One had 8 GB of unified memory, but of that amount an essential section was reserved for the system, which meant that, in the end, the programmers only had about 5 GB of unified memory free to save all of the game's data. This meant that it is still possible to play on PCs with just 8GB of RAM today, but it affected the development of games for consoles and has greatly hampered innovation and the creation of much more complex games with much larger worlds.

With PS5 and Xbox Series X the proportion of free unified memory has increased to 16 GB, but a section of this is also reserved for the system, which means that in the end the developers only have about 13.5 GB free, Exactly. As a result, some next-gen exclusive titles like Ghostwire: Tokyo now require 12GB of RAM on PC, but will that amount of memory be enough for the next-gen to age gracefully? Personally, I don't think so, and what's happened with PS4 and PS4 Pro, and especially Xbox One and Xbox One X, are two precedents that strengthen my position.