NVIDIA has released an interesting job offer that suggests the successor to the Nintendo Switch might be equipped with a next-generation Tegra SoC, and that it would be prepared to work with DLSS 2.0 technology.
Sounds good, of that there is no doubt, but let’s stop for a moment to see the key points of this new information. The first thing to keep in mind is that the successor to the Nintendo Switch has not yet been officially confirmed, but obviously this does not mean that it cannot have been in development for a while. You know how these things are, normally when a company launches a console it starts working on its successor, something completely normal since it is a process that requires a lot of time, and a lot of effort.
Now we go with the NVIDIA job offer. In it we can see that the company is looking for an experienced software engineer for its Tegra division, and it clearly refers to «Work on solutions based on artificial intelligence like DLSS 2.0». I don’t believe in coincidences, and uniting “Tegra” and “DLSS 2.0” in the same job offer is certainly a statement of intent.
What value would DLSS 2.0 offer in the successor to Nintendo Switch?
Well, the truth is that a lot. Think, for a moment, about the potential offered by the Tegra X1 SoC that integrates Nintendo Switch. It is a low consumption chip, very modest and with a fairly weak power by current standards. Its CPU uses ARM architecture, and its GPU is a Maxwell based solution with 256 shaders. For comparative purposes a GeForce GTX 750 It has 512 shaders and dedicated GDDR5 memory.
In order to move demanding games with a minimum of fluency, the Nintendo Switch has to resort to dynamic resolution, which means that even though a game renders, for example, in 720p, it can drop to 540p or even 360p in certain scenes to maintain good performance. This greatly affects the graphic quality, since reducing the pixels produces an unpleasant blur.
DLSS 2.0 technology would end that problem, and in the best possible way. Thanks to it, the successor to the Nintendo Switch would no longer have to pull dynamic resolution, but could apply smart rescaling that would allow, for example, render a game in 540p and rescale it to 1080p while maintaining excellent image quality, thanks to the combination of different images to obtain the perfect frame.
Obvious to say that this would require the new APU Tegra from the successor of Nintendo Switch have tensor nuclei, since it is these that are in charge of taking forward the workload that represents the DLSS 2.0 technology, based on artificial intelligence.
Personally, I think that this move by Nintendo would be a success, and it could make it easier for them to maintain a certain backward compatibility with the original Switch.