PlayStation 5 and Xbox S
eries X have a common goal: to set as standard the game at 4K and 60 FPS. Sony and Microsoft have sold us on this ambition since they started telling us about their next-gen consoles, and it makes sense. 4K UHD televisions are slowly becoming popular, so it is understandable that gamers demand that the hardware of the new consoles provide us with a satisfactory experience when we use them. And the sustained cadence required for that experience to be possible is precisely 60 frames per second.
On paper, the hardware of the new Microsoft and Sony consoles should be able to offer 4K graphics and a sustained frame rate of 60 FPS. PlayStation 4 Pro and Xbox One X already do it with some games, and the hardware of their successors is much more ambitious. However, if we turn our gaze to the PC world we will realize that the promise of PlayStation 5 and Xbox Series X is no piece of cake at all, especially when ray tracing technology comes into play (ray tracing). And this has a consequence that users are interested in knowing in order to keep our expectations under control: it is very likely that new generation consoles do not offer us native 4K graphics with all the games.
The promise of PlayStation 5 and Xbox Series X is no piece of cake at all, especially when ‘ray tracing’ comes into play
A state-of-the-art gaming PC equipped with a high-end graphics card, such as an NVIDIA GeForce RTX 2080 Ti, can suffer when forced to render 4K graphics by applying ray tracing. So much, in fact, that those sustained 60 FPS are not possible in some games, as monstrous as this graphics card is, it is. It’s certainly a bleak outlook looming over next-gen consoles as well, but it’s not all bad news at all. And it is that 4K graphics at 60 frames per second, and even with higher cadences, are possible both with the current PC hardware and that of the next consoles.
The solution to this challenge lies in the reconstruction techniques that did not yield the spectacular results that NVIDIA described when it unveiled this innovation, but DLSS 2.0, its revision, is very promising because it introduces very notable improvements on which the first one does not rely. version. This technology, without going into complex details, renders game images at a resolution lower than 4K to relieve the GPU of a significant part of the workload. Then in, an NVIDIA supercomputer specialized in the execution of artificial intelligence algorithms, to process them through deep learning, taking 64 samples per pixel and returning images in which the maximum possible level of detail has been recovered. The result of all this process should be a graphic finish very close to native 4K and with a sustained image cadence significantly higher than that offered by the hardware without applying DLSS.
DLSS 2.0 introduces very important improvements that the first version of this technology did not have, such as new temporal feedback techniques or the generalized artificial intelligence network
Some of the improvements that DLSS 2.0 introduces compared to the first version of this technology are the new techniques of temporal feedback, which allow to recover a higher level of detail; a more efficient use of the Tensor Core of NVIDIA RTX GPUs, and, what is undoubtedly one of the most promising enhancements, a widespread artificial intelligence network able to work with any game, and not only with those for which it has been previously trained. This video from Digital Foundry is very interesting because it illustrates very well how DLSS 2.0 works on a PC and the technique checkerboard on a PlayStation 4 Pro about the same game: ‘Death Stranding’.
Sampling using technology checkerboard has thrown very convincing results in some PS4 Pro games, like, for example, ‘Horizon Zero Dawn’, although it resorts to a somewhat simpler strategy than DLSS. Roughly its philosophy is to render the images at a resolution lower than 4K (usually 1440p), once again in order to relieve the GPU of some of the computational effort, and expand them to 4K by filling in the missing pixels. To do this, an algorithm analyzes the pixels that surround each missing pixel and applies a smoothing procedure to the jagged edges to minimize the presence of saw teeth (an effect known as aliasing).
‘Checkerboard’ technologies, dynamic resolution and DLSS are just some of the techniques that allow balancing graphic quality and image cadence
Xbox One X is noticeably more powerful than PlayStation 4 Pro, and this has made it possible for it to render a reasonably wide range of games at 4K natively. However, both the current Sony and Microsoft consoles have at their disposal a wide range of techniques that help them deal with 4K resolution in a more or less satisfactory way. One of the innovations with which it is also good to be familiar is dynamic resolution, which, as we can guess, consists of temporarily reducing the resolution at which the images are being rendered within predefined margins to lighten the workload of the GPU and ensure that the cadence of images per second does not drop when the graphic load she’s very tall. This technique is currently being used by some PC games as well as on Nintendo Switch, Xbox One and PlayStation 4.
Technologies checkerboard, dynamic resolution and DLSS are just some of the techniques that game developers have at their disposal to balance the graphic quality and the cadence of images per second of their titles, but there are other innovations that also pursue this objective and that can work side by side with them. In any case, we cannot ignore that DLSS is a proprietary technology of NVIDIA, and Xbox Series X and PlayStation 5 incorporate a CPU and a graphics engine developed by AMD together with Microsoft and Sony, so they cannot use it. This limitation seems to complicate the outlook for next-generation consoles, but it doesn’t have to be that way because we have reason to think that RDNA 2, the architecture on which both AMD’s upcoming graphics cards and PS5 and Xbox Series X GPUs, will put all the meat on the grill.
The future of PS5 and Xbox Series X is tied to AMD’s RDNA 2
This is official: the graphics processor of the next consoles from Microsoft and Sony is designed on RDNA 2 architecture. This has been confirmed by both AMD and these last two companies. However, this does not mean that their GPUs are identical. In fact, they are not at all. Sony and Microsoft engineers have worked together with AMD to introduce their own ingredients into the GPU of their consoles that allow them to distinguish themselves from the hardware of the competition, and for the moment Microsoft’s console has an advantage because its graphics processor is noticeably more powerful, at least if we stick to its ability to calculate floating point operations: the Xbox Series X GPU slightly exceeds 12 TFLOPS, while the PS5 is satisfied with a more modest 10.28 TFLOPS. This is not the only ingredient that determines the power of a graphics processor, but it is a measure that helps us get a fairly accurate idea of the power that these chips have.
Unfortunately, we don’t know everything about RDNA 2 yet because AMD hasn’t revealed all the details about its upcoming graphics architecture. For several weeks the possibility of this company presenting its high-end graphics cards with this architecture during the months of September or October, a little before the arrival of PS5 and Xbox Series X, has been gaining weight. And it makes sense because it is reasonable that prefers to unveil its new architecture on its own and before the next-generation consoles arrive. Logic also invites us to assume that RDNA 2 is AMD’s bet to compete with the upcoming NVIDIA GeForce RTX 30, which is also expected to arrive in September, or, at the latest, in October.
RDNA 2 is AMD’s bet to compete with the next NVIDIA GeForce RTX 30
All this promises us that the next few months are going to be exciting for gamers, especially for those of us who enjoy digging into hardware, but what tools does AMD have to deal with image reconstruction on RDNA 2 architecture graphics processors? What both Sony and Microsoft have revealed in their technical sessions dedicated to their new consoles seems to reflect that both PS5 and Xbox Series X will have VRS technology (Variable Rate Shading), something very reasonable if we take into account that this innovation will be supported natively by RDNA 2. This technique allows developers to deal with the shading of the pixels of each frame, prioritizing those regions that will have a greater impact on the quality of that will allow Xbox Series X to use the standard cores of the GPU of this console to make possible the efficient execution of deep learning algorithms that will seek to increase the quality of which, on paper, describes a strategy similar to the one it has deployed NVIDIA with its DLSS. We still cannot be sure what the experience that Xbox Series X and PS5 will offer us when they arrive at our homes, but it will be interesting to check if they really measure up of the expectations they are generating.