The next Nvidia graphics card in the 20-series range of Turing-based GPUs must surely be the RTX 2060. Or GTX 2060. We’re still not entirely sure exactly how the naming scheme is going to play out, especially given that Nvidia itself hasn’t said word one about the next-gen mainstream GeForce graphics cards.
Which isn’t all that surprising considering there are still thousands of current-gen mainstream Nvidia cards still clogging up the channels and making it very difficult for the green team to realistically think about replacing them. Hell, Jen-Hsun’s stopped shipping any mainstream GPUs until next year because there are so many damned GTX 1060s still sitting on the shelves.
But one day next year we’ll have a shiny new 20-series card that will actually be affordable for us normal PC gamers who can’t afford to drop some $500 on an RTX 2070, the lowest priced Nvidia Turing graphics card around. But what will that GPU look like, how will it perform, how much will it cost, and will we still be able to ray trace the fun out of Battlefield V?
The first thing we’d like to see regarding the Nvidia RTX 2060 would be a release date. And if Nvidia is being smart, and as ruthlessly tactical as we know it can be, then we’d expect that RTX 2060 release date to be set sometime after team Radeon has released the new AMD Navi graphics cards.
Navi is expected to be AMD’s direct replacement for Polaris – despite what Lisa Su has said about competing at the high end – and will be dropping into the mainstream as a next-gen 7nm alternative. That’s the end of the market the RTX 2060 is going to be making its home, and it would make sense for Nvidia to wait and see what the performance of the flagship Navi card ends up being before finalising the speeds and feeds of its own mainstream offering.
Given that it needs to wait for at least one whole sales quarter – and probably two – for the existing mainstream 10-series GPUs in the channel to shift, holding off until after Navi wouldn’t be too much of a stretch. That could put the RTX 2060 release date as sometime in the summer, potentially June 2019. We’re expecting AMD’s new GPUs to launch in the second quarter of 2019, which would potentially set up an exciting, competitive, combative Computex.
RTX 2060 or GTX 2060?
What we’d also like to see is an end to the speculation over the naming scheme of the new mainstream GeForce graphics card. Ever since the RTX moniker was first announced there has been a lot of chatter about whether Nvidia would stick with the RTX branding right across the stack, or have the lower-end GPUs reverting to the GTX prefix without the power to ray trace in real time.
For my money, I think Nvidia is likely to stick with the RTX branding for consistency, and to provide a link between its absolute best GPUs – that super-smash RTX 2080 Ti – and the low-end of its 20-series stack.
Even if the RTX 2060 doesn’t have the silicon cojones to power real-time ray tracing in games, Nvidia has already set out its stall saying that RTX-support does not automatically equate to ray tracing. The RTX tag used for games is also being seen as denoting support for the other new features baked into the Turing GPUs, such as the AI-powered deep learning super-sampling (DLSS) and variable rate shading.
Both of those are performance-enhancing features and could be genuinely effective technologies if used correctly at the lower-end of the graphics card market.
With Nvidia switching back to the GTX naming it would likely be seen as devaluing those 20-series cards, and that’s something I don’t see Jen-Hsun and co. wanting to do.
What’s it going to be capable of?
There have been some early RTX 2060 benchmarks leaked by the Final Fantasy XV database. Yes, that ever-reliable FF XV benchmark. It’s not really a benchmark we use when we’re testing graphics card capabilities due to its less-than-consistent results. You can see that from the database itself where you’ll find RX Vega 56 cards beating out AMD’s flagship RX Vega 64, for example.
Still, it’s the first vague notion of RTX 2060 performance we’ve seen and shows the test RTX 2060 delivering JRPG performance that’s 30% higher than the GTX 1060, and 22% faster than the new AMD RX 590. Which isn’t that exciting. We’d have maybe hoped that it would at least a touch faster than the GTX 1070, yet it’s actually 6% slower according to this.
That performance level maybe lends more credence to the possibility of an RTX 2060 Ti card stepping into the sizeable price and performance gap then left open between the RTX 2060 and RTX 2070.
But will it ray trace? That’s a tough one to answer. Nvidia has told us that tracking 5 billion rays per second is about the minimum for a meaningful ray traced experience, and the current RTX 2070 is rated at around the 6 billion rays per second mark. So potentially Nvidia could introduce an RTX 2060 Ti using the same TU106 GPU as the RTX 2070, but with a little logic lopped off, and make sure it’s capable of nailing 5 billion rays.
If that comes to pass then I’d have to say I doubt the RTX 2060 would have any hope of a ray tracing future. Though UL has recently said it expects more graphics cards are going to get DirectX Raytracing support in 2019 which could indicate at least one more Turing GPU getting ray happy next year.
But how much?
Pricing for the RTX 2060 is going to be key, but Nvidia has already set out its pricing strategy at the high-end. With the RTX 2070 costing at least $499 there’s potentially a big gap between that and the $250 – $300 price range you’d want a mainstream graphics card to appear in. Though, perhaps as much as the release date, that pricing level might just depend on what AMD’s Navi GPUs look like when they arrive next year.
Given the price hikes for the different levels of GPUs – from RTX 2070 through to RTX 2080 and RTX 2080 Ti – compared with their 10-series brethren, it wouldn’t be too much of a stretch to expect the RTX 2060 to be priced way higher than the GTX 1060. That might be difficult if the Navi GPUs arrived around the $300 price point with decent mainstream gaming performance. If it manages a 20% performance boost over the RX 590 then that matches the FF VX RTX 2060 benchmark.
The solution? Nvidia could just add another GPU into the mix. There has already been speculation that it might have an RTX 2060 Ti waiting in the wings for a CES 2019 unveiling in January, along with the other Max-Q mobile GPUs. And that would allow Nvidia to have a high-cost mainstream Turing chip somewhere around $399, as well as a more reasonable $299 option. And then it would still be able to have the RTX 2050 knocking around the $199 mark.
So, what about the pie-in-the-sky daydreams we might harbour around the RTX 2060, what would we ideally like it to be? Obviously ridiculously cheap with GTX 1080 Ti levels of gaming performance would be grand… but unfeasible.
We also can’t help but long for the days of the gorgeous single slot graphics card. Again, that’s unlikely from the temperatures some Turing GPUs have been posting under load, but that doesn’t stop us also hoping for some really shiny small form factor designs too.
And NVLink. We know Nvidia has nixed it from even the RTX 2070, but go on, give us the chance to jam a pair of RTX 2060 GPUs into our rigs and get high-end, sometimes frustratingly awkward performance. Again, it’s not going to happen, but a nerd can dream…