Originally Posted by Eurogamer
So, for Nintendo and indeed game developers, the Switch OLED model is very much business as usual, to the point where the platform holder is telling game makers that no new development kits are essential for making games and there are no new technical requirements that require changing standard Switch iconography in-game to accommodate the new model. On top of that, games running on Switch OLED have no idea they are running on anything other than a standard unit - Nintendo's developer documents reveal that there is no way for their games to query the system, to figure out whether they are running on Switch OLED or not. With that said, the firm does recognise the need for developers to test their games on the new screen. With that in mind, a new ADEV development model is being made available to co-exist alongside the existing SDEV and EDEV versions. For reasons undisclosed by the Nintendo, this machine ships with 8GB of onboard memory compared to the 6GB in the other development models and the 4GB of all retail units.
(...)
With a new Nvidia chip ruled out, could the existing model have provided any form of improved performance for docked users? Absolutely, but not to a genuinely game-changing degree. Overclocking (or indeed underclocking) the Switch has been a component of the homebrew scene for years now, with system hacks able to boost CPU from its standard 1.02GHz to the full spec 1785MHz, while the docked clock of 768MHz can easily be boosted to 921MHz. On paper it's a 75 percent boost to CPU performance and a 20 percent uplift for the GPU. The 'Mariko' Tegra X1 graphics core can even run at 1.267GHz - a sizeable 65 percent improvement to the Switch spec. However, despite the presence of extra venting on the new dock, it seems unlikely that Nintendo will open up full performance to developers. In theory it is possible, but there's certainly nothing in the SDK to make it happen for game makers. And in truth, based on our overclocking tests, you gain extra stability and a cleaner picture on games that support dynamic resolution scaling, so while there is an improvement, it's not enough to comprehensively solve the docked play problem.
Of course, the truth is that only a fundamental redesign of the Switch itself can address this - something that does not really make business sense to Nintendo in the here and now, and would only cause headaches to developers in supporting a third performance profile. Any kind of smart upscaling solution capable of making truly low resolution images into something acceptable on a living room display requires new silicon, and while Nvidia DLSS has been mooted as a solution, this technology requires tensor cores baked into the hardware - and these are not present in Tegra X1. DLSS cannot work as a standalone 'add-in' chip in its current iteration, and neither can it be liberally applied to any game - it needs to be plumbed into the game engine.
And both factors, combined with Nvidia's recent improvements to DLSS, make the technology a better fit for an actual Switch successor based on more recent GeForce architecture. The potential is certainly mouth-watering. DLSS performance mode runs internally at a quarter of output resolution: 720p becomes 1440p, 1080p becomes 4K. The results aren't perfect, but would certainly work well on a TV viewed at range in living room conditions. DLSS ultra performance mode can actually scale 720p to 4K. Yes, there's a quality hit, but it works. DLSS or a technology similar to it is the missing piece of the puzzle for making docked play from a mobile chipset viable, but the reality is that this is next generation stuff. So, is this the actual Switch Pro that has been talked about for months now? Would it still be a Switch Pro - or should we start referring to it as Switch 2 or Super Switch instead?
