It should have happened a long time ago. Nintendo originally partnered with semi-conductor giant Nvidia for a successor to the DS which nevertheless never went into production although prototype development kits based on preliminary versions of Tegra processors have been produced. However, only now with Switch do we realize how powerful such a partnership can be. Nintendo’s idea of a hybrid gaming console combined with Nvidia’s performance and efficiency (parameters where the Santa Clara house is second to none) to create the best-launch console ever in the history of the game. Kyoto house. The question that now arises is therefore: what will be the next step in this collaboration?
First of all, let’s be clear: the collaboration between the two giants is not a one-off agreement, at least according to what Nvidia declares. In an investor conference held last year, Nvidia CEO Jen-Hsun Huang said the partnership with Nintendo “is likely to last two decades,” which is not surprising considering the deal suits both companies.
Nintendo needs the technology of Nvidia, an industry leader with a particular emphasis on energy efficiency, for its portable devices. Nvidia’s Tegra division, for its part, has never been a very successful business, and a deal for a large volume of chips is pure gold for Nvidia, now that we know that the custom Tegra processor it supplied for the Switch is identical. to the Tegra X1, already on the market since 2015.
But looking ahead, what kind of hardware could Nintendo take advantage of based on what looks like exclusive access to Nvidia’s technology? It is certainly interesting that the technology needed by Nintendo to create a second and more refined version of the Switch is already available, following the lines drawn with the mid-gen refreshes of DS and 3DS. Greater performance and battery life were the highlights of both the DSi and New 3DS, and that’s exactly what Nintendo could achieve by switching to Nvidia’s Tegra X2 processor.
This is the processor that we would have hoped to find behind the Switch technology, designed with the most efficient manufacturing process (16nm FinFET versus the planar 20nm of the X1) and which also boasts an upgrade to the Nvidia Pascal architecture for the GPU (which it actually has an almost identical number of cores to that of the Switch chip) along with double the memory bandwidth. The Tegra X2 retains the processor’s four ARM Cortex A57 cores, along with Nvidia’s proprietary 64-bit Denver CPU architecture. This is an interesting refresh of the Tegra X1, but it does not represent a generational leap. We can get an idea of its potential by taking a look at the only product on the market that mounts it, the Jetson TX2 embedded system kit.
It is a PCB solution designed for developers who intend to create their devices based on Tegra processors, and offers two performance modes. Pushed to its maximum, the GPU reaches the frequency of 1300MHz while the CPU cluster runs at 2GHz or more. In efficiency mode, however, the A57 cores run at 1.2GHz and the GPU is set to 854MHz. In this configuration, Nvidia claims that the Tegra X2 is twice as fast and efficient as the X1. And most importantly, its maximum efficiency mode also disables the Denver CPU cores, leaving only the Cortex A57s active, and that’s really all Nintendo would need for full compatibility with the current Switch title library.
Tegra X2 meets all of Nintendo’s needs for a possible second-generation Switch: even taking into account additional downclocks, it should be able to easily outclass the performance of the current hardware, increasing battery life and giving a significant boost to performance when the console is connected to the dock. It is difficult to imagine that the Denver cores can be of much help (and the fact that they are disabled in efficiency mode speaks volumes enough) and the mere fact that a new processor like the Tegra X2 only uses A57 cores suggests that the processor in issue was built with the purpose of maintaining compatibility with the X1. Long story short, with the subsequent improvements to the CPU’s ARM cores, many think that keeping the A57 cores in the Tegra X2 design could bring more benefits to Nintendo than to Nvidia.
In addition to the Tegra X2, Nvidia has already announced its latest and most innovative Tegra processor codenamed Xavier. It’s still built with 16nm FinFET technology, but the GPU’s CUDA cores have doubled and there’s also an octa-core for the CPU cluster. How well this chip is suitable for a Nintendo console at the moment is not very predictable (we would likely be facing a level of performance in the same sphere as the base PS4). The size of this chip would make it difficult to fit into a hybrid device like Switch, and the processor would obviously also be much more expensive. The power consumption of a double GPU than the one fitted to the current Switch would also be problematic, but there is one factor to consider: Nvidia has already confirmed that the new Volta GPU architecture used in Xavier is 50 percent more efficient from point of energetic point of view with respect to Pascal. In a few years, when the production costs of the chip have decreased, maybe Xavier will be able to be used in a tablet-type device?
There are other options to explore of course, and anything could happen in this window of collaboration between Nintendo and Nvidia that on paper should last for two decades. Considering a particularly innovative concept, there is nothing that can prevent the Kyoto house from returning to the purely home console sector. Pairing the next generation of 64-bit CPUs with a GPU at the level of the GTX 1060 could result in a console powerful enough to handle 4K displays, based on our recent tests in which we employed PS4 Pro rendering techniques on hardware. Mainstream gaming PC. In addition to the impressive results seen on that occasion, let’s remember how good Nvidia’s API – NVN graphics console has proven itself. Tegra X1 is based on (far from recent) technology from 2015 and is achieving outstanding results from its 256 CUDA cores. The GTX 1060 instead boasts 1280 that run at around 2.3x the clock speed, paired with much faster memory. In practice, Nintendo has historically not pushed its platforms to the technological limits available, but now it has the potential to do so if it will.
But maybe there will be a different approach. Last year, there was a lot of discussion about Nintendo’s patent for an additional computational device (SCD) that essentially added computing power resources to the Switch, and the patent indicated that this computing power would be distributed through the cloud. But the same patent opens the door to a more powerful hardware device that could integrate the current processor. It could be in the form of a larger dock, which connects to the Switch via the USB-C port, the interface that takes care of conveying high-bandwidth data between the tablet and the dock. But what kind of hardware would we find inside such a device?
Well, at that time, a leak had emerged from the foxconn factory where the Switch was produced. In some ways, the information provided proved to be inaccurate (it supported the idea of adopting a much more powerful processor), but in others (especially with regard to the physical description of the machine and its interior) it proved to be 100% reliable. percent. And there is a discussion regarding that leak about an “advanced development kit”, which appears to have 8GB of RAM, and a processor that is about the same size as Nvidia’s GP106 adopted in the GTX 1060. And beyond that, yes speaks of another processor (comparable in size to the Tegra mounted on the Switch).
Assuming that it is not the result of fantasies, this dev kit looks like a device that integrates the current Switch in combination with an additional computational device, a very “user-friendly” “all-in-one” gadget. for developers. The retail version of this device would simply be a replacement for your Switch dock. The Switch GPU could be completely disabled, allowing for higher CPU performance capable of creating more detailed images, with additional hardware resources behind it via the cloud.
This is an interesting theory on a potential future path, which would certainly represent a good chance to compete with PlayStation 4 Pro and Project Scorpio, at least from the point of view of graphics power. But the challenge the developers would face would be twofold. First of all, the titles should support Switch in three configurations: portable, docked, and with SCD enabled. Secondly, even with the CPU clock raised, we would be faced with an underpowered processor compared to the graphics unit that we might find in the SCD unit. Cross-platform ports could therefore be, in some respects, cut off.
Finally then, the main point of this discussion is this: the device described by the Foxconn leaker is a graphical upgrade, pure and simple. Historically, Nintendo is not used to launching large upgrades solely aimed at increasing graphic quality: there would be a need to have a convenient and innovative concept behind it, and if this project will ever become real we can’t wait to see how it will be designed. .
But in conclusion, even if the speculations on the Tegra X2, Xavier and the SCD unit will prove to be just a hole in the water, Nintendo still has access to some of the most powerful and efficient hardware on the market, with development tools and API perfectly. synchronized by one of the brightest minds in the industry. For a company that has long insisted on ancient architecture compounded by extremely poor development environments, working shoulder to shoulder with Nvidia as a technology partner is truly a game changer for Nintendo.