It’s really not much of a spoiler now to say the Nvidia RTX 2080 Ti is the fastest consumer graphics card on the planet right now. If it wasn’t there’d be villagers with flaming torches and pitchforks all the way up Walsh Avenue to Nvidia’s Santa Clara HQ. It’s lucky then that the GeForce RTX 2080 Ti is a beautiful piece of silicon and technical engineering, both in the design of the Turing GPU itself as well as the physical Founders Edition card.

And it’s even knocked the Volta-powered Titan V into second place – if you’re looking for any kind of value judgement for this supremely expensive GPU there it is – the $1,200 RTX 2080 Ti can beat the $3,000 Titan V. Though if you’re looking for a value judgement on a $1,200 graphics card then you’re honestly looking in the wrong place.

We’ve already checked out the second-tier Nvidia RTX 2080, a card that just about manages to justify itself as an upgrade to the GTX 1080 Ti it’s ostensibly priced against, but the RTX 2080 Ti is another level up again.

It’s this generation’s ultra-enthusiast card and has no peers, the only card close to it in terms of price is the Titan Xp which is essentially a slightly beefier 1080 Ti. In other words, no threat at all…

So the RTX 2080 Ti is a standalone monster of a GPU. A graphics processing powerhouse that is designed as the standard bearer for 4K PC gaming and the foundation for all the work that Microsoft and Nvidia have done to bring real-time ray tracing and AI-powered gaming to the desktop.

Sure, it’s an aspirational card that few of us will ever be able to afford. But think of it like a preview of the sort of power that you’ll be able to find in a far more mainstream GPU of the next generation of graphics cards. That may be cold comfort at this, the start of the Turing generation, but fingers crossed we don’t have to wait as long for its successor.


Nvidia RTX 2080 Ti specs

The top-spec GeForce RTX graphics card comes bearing the TU102 GPU at its heart. It’s the same graphics silicon which powers Nvidia’s $10,000 Quadro cards – another note in the unnecessary value column there – albeit with a few SMs lopped off. So yes, this isn’t the full-fat version of the TU102, but with 68 SMs (four fewer than the complete Quadro GPU) and 4,352 CUDA cores it’s not exactly a featherweight chip.

In fact it’s a definite heavyweight in the GPU silicon world, with 18.6 billion 12nm transistors packed into its massive 754mm2 footprint. That’s a monster chip, and more than twice the size of the GTX 1080’s GPU.


Alongside the graphics processor itself is 11GB of brand new GDDR6 memory, running at 14Gbps across a 352-bit aggregated memory bus. That enables it to deliver 616GB/s of memory bandwidth, again almost twice that of the GTX 1080.

The Founders Edition cards are overclocked versions this time around, and so come with slightly higher boost clocks than the reference models. We’re talking about only by around 90MHz, but they’re also the finest versions of the Turing silicon to come out of the factories, so they’ll have even more overclocking headroom to offer.

Of course, as an RTX graphics card it’s got to have some RTX silicon inside it, and so the RTX 2080 Ti has 68 RT Cores, and 544 AI-focused Tensor Cores. It’s the highest-spec consumer chip built for the rigours of real-time ray tracing and the awesome potential of using AI processing to make our games look better, run faster, and feel far more realistic. I for one welcome our new robot overlords, especially if they can make my games look super-pretty. And hell, they can hook me up as a battery if they can also give me genuine AI in my beloved Football Manager.

Sadly none of that is going to be used at launch. We’ll have to wait for Microsoft to drop the October Windows update before we can enjoy real-time ray tracing in games, likely kicking off with Shadow of the Tomb Raider’s shadows, and for developers to get the AI-powered Deep Learning Super Sampling into games. That’s not tied into the Windows update, so could happen before October, potentially with Final Fantasy XV first to the punch.


This RTX 2080 Ti Founders Edition uses the same almost unibody shroud design as the RTX 2080 Founders Edition we’ve already tested. And it’s just as good here on the Ti version. It’s a monstrously weighty card, thanks to the twin axial fans and full-length heat exchanger below them, but the design of the rigid shroud ensure that all the weight is on the bracket, and therefore the chassis. That means there’s no danger of it warping your PCIe slot and messing up your motherboard.

If you want to sell both kidneys, however, you can always buy a second RTX 2080 Ti and hook them up NVLink. It’s the same high-bandwidth link Nvidia has been utilising on its professional cards, ported down to the consumer space. NVLink is kinda like SLI on steroids, only it isn’t. It’s a peer-to-peer link with 100GB/s of potential bandwidth hopefully ushering in a new age of multi-GPU cooperation. Though maybe not for a while… a familiar theme with this nascent Turing generation.