The new king of 4K gaming.
Like clockwork, Nvidia has announced a new architecture for its GPUs, and it’s debuting it in a range of Founder’s Edition graphics cards that feature an all-new design befitting a new direction for the company. While the RTX 2080 could be considered the standard card in the 20XX-series lineup, the RTX 2080 Ti is the top dog in terms of raw performance (and its unusually high $1,200 price tag). All of the new GPUs from Nvidia feature specialized “Turing” hardware that will allow them to eventually offer features like real-time ray tracing and artificial intelligence-powered super sampling, so there’s more to this GPU than just sheer horsepower. I’ve spent the past week putting it through its paces, so let’s dive in.
RTX 2080 Ti – Design and Features
Just as Nvidia’s previous GPU architecture, Pascal, provided a significant performance boost over its Maxwell predecessor, the new Turing architecture in the 20-series cards take things even further, but not in the way we’ve come to expect based on previous launches. Instead of simply offering more performance, Turing offers entirely new features never before seen in a GPU. Though we’ve already covered the potential of ray tracing previously, both with articles and hands-on gaming, the other big innovation with Turing is called Deep Learning Super Sampling, or DLSS. This technology uses a neural network to allow Nvidia to essentially feed game images to the network, which it then uses to learn how to recreate the images for either better image quality, or faster performance.
DLSS is handled by Turing’s Tensor cores, so The result is less rendering load placed on the GPU, which allows for increased performance. Again, this technology isn’t ready for prime time as of launch, unfortunately. But the implications are profound, assuming Nvidia can get enough developers on board. As of this writing, 25 games have announced upcoming compatibility with DLSS.
The first RTX 2080 Ti I’m taking a look at is the Founder’s Edition, which is a first-party card released by Nvidia. As usual, the Founder’s Edition cards tend to carry a slightly higher price tag than “partner” cards, but demand has made them equivalent for now. Much like the RTX 2080, the RTX 2080 Ti does feature some design improvements from the 10-series Founder’s Edition.
First, there’s dual fans and a full-length vapor chamber, which is obviously a better setup than the previous single-fan, “blower” style cooler. Nvidia also has encased the entire card in an aluminum shroud, which looks very sleek and clean. The RTX 2080 Ti has expanded I/O connections, including three DisplayPort 1.4, an HDMI 2.0b port, and a VirtualLink USB Type-C connector. The latter is increasingly important if you’re using VR headsets.
RTX 2080 Ti – Specs
The RTX 2080 Ti is built on the Turing TU102 GPU and includes 4,352 CUDA cores; an almost 20% increase over the GTX 1080 Ti Founder’s Edition. There’s 11GB of all-new GDDR6 memory, and the GPU boost clock has increased slightly from 1582 MHz on the 1080 Ti to 1635MHz. Nvidia claims the Turing TU102 chip in the 2080 Ti, with its 18.2 billion transistors, is the most powerful GPU ever placed in a GeForce graphics card, and as I’ll discuss later in my benchmarks, that claim seems to hold up. For context this chip has about five billion more transistors than the RTX 2080, so it’s a massive chip indeed, and makes the price seem a bit more reasonable when you consider its size. Though it’s not named Titan, it easily could have been.
Powering the previously mentioned DLSS and ray tracing functionality are two new sectors in the Turing architecture, with 68 RT cores and 544 Tensor cores. Obviously, neither of those are comparable with the GTX 1080 Ti, but for comparisons sake, the standard RTX 2080 features 46 RT cores and 368 Tensor cores. How much that difference impacts ray tracing or DLSS performance, I have no idea at this point. But it stands to reason the RTX 2080 Ti will have the edge.
The thermal design power, or how much average maximum power the GPU will draw while running applications, for the RTX 2080 Ti is set at 260 watts, just a 10w increase over the GTX 1080 Ti. Accordingly, this card is features two 8-pin power connectors. As you’ll see in the benchmarks below, technical jargon aside, the RTX 2080 Ti is just an absolute monster of a graphics card. Where the higher-end 10-series cards were specced to near 60fps in 4K resolutions, and the standard RTX 2080 finally makes that dream come true, and the RTX 2080 Ti pushes most modern games past the 60Hz mark even at ultra settings.
RTX 2080 Ti – Benchmarks
I spent some time—a lot of time, actually—putting the RTX 2080 Ti through its paces in both real-world and synthetic benchmarks. I ran all the tests on a hand-built system that was used for the Pascal GPUs, and it consists of a Skylake Core i7-7700K CPU, 8GB of DDR4 memory, an Asus Z270 Prime motherboard, Intel SSD, and EVGA PSU. I tested this card (and all others) at 3840 x 2160, 2560 x 1440, and 1920 x 1080 resolutions with all games and benchmarks set to their highest graphical settings, but without anti-aliasing activated. You can see the results in these charts below:
Clearly, the RTX 2080 Ti is at the head of the pack in almost every single test, at every single resolution. When compared to the GTX 1080 Ti, this card decimates its predecessor in regards to pure 4K speed. By and large, the RTX 2080 Ti performed around 25 to 35 percent faster, and in some cases—Monster Hunter: World, for example—as much as 58 percent faster. The only game where this card fell under 60fps in 3840 x 2160 was Shadow of the Tomb Raider, and it was close enough to almost be a moot point. Moreover, with just the slightest bit of settings tweaking, achieving rates well above 60fps is quite easy.
Those ratios stayed fairly consistent in regards to 2560 x 1440 resolutions, but the span between quad HD and 1080p resolutions was much narrower in most real-world gaming scenarios. I can’t give a fully technical explanation of why beyond my assumption that some CPU bottlenecking is taking place.
Of course, when compared to a Zotac GTX 1070 Ti, the span between 4K refresh rates is kind of incredible. In most cases, the RTX 2080 Ti is 100 percent faster. Yes, the 1070 Ti was already a lower-rated card, but the jump between Pascal and Turing is well on display with pretty much anything thrown at the RTX 2080 Ti. I’ll be taking a look at more third-party cards soon, but for the time being, if you’re the type of person who’s interested in a Founder’s Edition, the RTX 2080 Ti is a monster.
I did run into a couple technical issues while using the RTX 2080 Ti. While running Far Cry 5 benchmark, the system would crash completely to reboot after the initial loading screen. This happened every time I attempted to run the game with this card. Also, while running PlayerUnknown’s Battlegrounds, there were large black artifacts on the screen when entering or just after exiting a building. I could not recreate this issue with any other card on the bench, with the exception of a Zotac 2080 Ti, so it’s possible it’s due to the pre-release drivers.
I haven’t attempted to do any overclocking with the RTX 2080 Ti at this point, purely due to time constraints. I’ll be taking a further look at this card in the near future with even more expanded real-world gaming benchmarks. I can tell you though that in my testing under full load the RTX 2080 Ti reached a max temperature of 77C and a boost clock of 1875MHz. As was the case with the RTX 2080 Founder’s Edition, and frankly its first-party predecessors, this card ran a bit warmer than I would have liked, but not so hot that it caused me any concern. Overall 77 degrees is very decent considering the amount of power on tap here.
Finally, though this card is clearly the most powerful GPU available today, it’s as future proof as you can get with ray tracing and DLSS support arriving in the future, but we don’t know how that will all shake out. It’s disappointing not being able to take a deeper look at those features right now as the card is launching, but even excluding those technological advances, the RTX 2080 Ti is just an incredibly beefy graphics card. As I mentioned earlier, if you want blazing fast 4K GAMING, this GPU is the best game in town. But price is a serious factor in this equation, and the RTX 2080 Ti is, to put it bluntly, just ridiculously expensive at this point. When a card costs more than plenty of people spend on an entire rig, that’s an issue. At some point, you have to ask yourself if the bump in performance is enough to justify a $400 or more increase from the RTX 2080 (which still includes ray tracing and DLSS, too).
The GeForce RTX 2080 Ti Founder’s Edition has an MSRP of $1,199 and is now available, but good luck finding one as it looks like they are all sold out. However, we put together a guide of where you can theoretically order one, and your best bet is to get one with a pre-built system. Barring that, you can sign up on Nvidia’s website and it’ll notify you when more are in stock.