Loading build...
Your Shops
Nvidia GeForce RTX 20 Series GPU Family Revealed - Pricing, Specs and Performance Details
By Jon Sutton on August 21st, 2018 at 11:20am - original article from game-debate

After months of teasing, Nvidia last night revealed the GeForce RTX series to the world. This includes the GeForce RTX 2080 Ti, the GeForce RTX 2080, and the GeForce RTX 2070. All three of these GPUs are ready to launch on September 20th and are available to pre-order right now.

These are the first ever gaming graphics cards capable of real-time ray tracing (RTX). The GeForce RTX series is based on the Nvidia Turing GPU architecture, utilising RT Cores (Ray-Tracing) and Tensor Cores (AI) alongside the traditional CUDA Cores.

Nvidia boasts that GeForce RTX represents “the biggest generational leap ever in gaming GPUs.” As for whether there’s any truth to those claims, well, the presentation was suspiciously light on benchmarks. Despite this, Nvidia claims that Turing delivers 6x the performance of previous-gen Pascal, performance primarily utilised for near-photorealistic real-time ray-traced lighting.

“Turing opens up a new golden age of gaming, with realism only possible with ray tracing, which most people thought was still a decade away,” said Jensen Huang, founder and CEO of NVIDIA, speaking before Gamescom. “The breakthrough is a hybrid rendering model that boosts today’s computer graphics with the addition of lightning-fast ray-tracing acceleration and AI. RTX is going to define a new look for computer graphics. Once you see an RTX game, you can’t go back.”

From the looks of things, the ‘RTX’ capabilities are being reserved for the RTX 2070 and upwards, so the likes of the 2060 and below are going to be sticking to the ‘GTX 2060’ nomenclature. If you want ray-traced lighting, the GeForce RTX 2070 is your entry point, although this entry certainly doesn’t come cheap.

The cheapest ray-traced RTX graphics card is the $599 GeForce RTX 2070. The Founders Edition commands a $100 price premium but is overclocked right out of the box. Nvidia has said the starting price for AIB models of the GeForce RTX 2070 will be $499. Good luck finding it at that price though as Nvidia’s ‘suggested’ pricing is rarely ever adhered to.

Further up the chain, we have the $799 GeForce RTX 2080, followed by the absolutely monstrous GeForce RTX 2080 Ti at $1199. That’s almost double the price of the GeForce GTX 1080 Ti, which launched at $699.

But what does frightening amount of cash get you? Here are the full specs or the GeForce RTX 20 Series GPUs.

Nvidia GeForce RTX 20 Series Specs


GeForce RTX 2080 Ti

GeForce RTX 2080

GeForce RTX 2070

GPU Architecture








Process Node

12nm NFF

12nm NFF

12nm NFF

CUDA Cores




Tensor Cores




Core Clock

1350 MHz

1515 MHz

1410 MHz

Boost Clock

1545 MHz

1800 MHz

1620 MHz





Memory Speed

14 GHz

14 GHz

14 GHz

Memory Interface




Memory Bandwidth








Power Input

2x 8-pin

1x 6-pin & 1x 8-pin

1x 8-pin

Founders Price




Normal Price




Yet again we see Nvidia shifting up the price brackets for each of its GPUs. The RTX 2070 will be $599, although AIB partners are free to price lower. This seems unlikely though, and we know these lower prices are typically marketing guff from Nvidia to mask the true pricing. If we compared this to the previous generation, the GeForce GTX 1070 launched for $379, the GeForce GTX 1080 for $549, and the GeForce GTX 1080 Ti for $699.

So, from Pascal to Turing, we are seeing price increases of between 45-71% for like-for-like GPUs. This is a huge price hike but a necessary cost if you want to be in on the ground floor, using brand new tech such as ray-tracing. 

In terms of actual performance of the RTX 20 Series, Nvidia was coy. Rather the pop up the usual bar charts, Jensen simply said they needed an entirely new metric to convey the performance of these graphics cards. This is a roundabout way of not comparing them to the previous generation at all.

What appears likely, however, is that we're looking at roughly 20-25% performance gains over the previous gen in terms of core count, while clock speeds fluctuate marginally. However, performance gains could improve significantly if Nvidia has improved the architecture and efficiency.

The big difference though is the RT Cores. In supported games, all shadows and lighting can be offloaded to the RT Cores, leaving much more computing power for the rest of the GPU. This could prove a hefty boost to frame rates, although is going to be largely dependent on how efficient the RT Cores are working in parallel.

However, in games without Raytracing support, and we're reading between the lines here, the difference in performance between Pascal and Turing will not be so pronounced.

What is ray tracing?

Real-time ray-tracing, or RTX, is a method for physics-based rendering of lighting and shadows. A developer can put a light source in a game and RTX will calculate how far the light travels, how it bounces and refracts off objects, and how it reflects off surfaces. Different light sources also impact one another, requiring some incredibly complex computations as the RTX GPUs calculate exactly how this would look in real life. For example, how an explosion in Battlefield V would look in a shop window when reflected in a puddle. Each and every beam of light is traced in real-time. It's the sort of tech Pixar has been using but, rather than rendering a single frame for days, it now can all happen in an instant.

What about the rest of the GeForce 20 Series?

All evidence so far indicates the GeForce 2060 and below will not be capable of real-time ray-traced lighting. Because of this, they should be known as the GeForce GTX 2060, GeForce GTX 2050 Ti, GeForce GTX 2050, etc. They will have cut-down Turing GPUs which will run faster than their Pascal equivalents, although we expect it will be an evolution rather than a revolution. The performance gains are likely to be comparable to the benefits we saw moving from the GeForce GTX 700 series to the GeForce GTX 900 series.

With that in mind, we've taken a stab at the expected pricing of the lower-end GeForce Turing models. This is based upon the price differential between tiers of GPUs during previous GeForce gens, as well as taking into account the GeForce GTX 2060 and below aren't capable of ray-tracing.

GeForce GTX 20 Series Expected Pricing

  Expected Price Previous Gen Price (Pascal)
GeForce GTX 2060 $300-350 $249
GeForce GTX 2050 Ti $200-250 $139
GeForce GTX 2050 $150-200 $109
GeForce GT 2030 ~$120 $80
GeForce GT 2020 ~$80 N/A

On top of this, we would expect there to be at least one new RTX 20 Series GPU, possibly even more. The GeForce RTX 2070 Ti at around $600 seems a safe bet to launch next year, although it seems fairly unlikely that Nvidia has a chip bigger than the GeForce RTX 2080 Ti waiting in the wings. They may perhaps have a Quadro RTX 8000 equivalent flipped into a new Titan, but that $10,000 workstation GPU only has a 5.8% higher core count than the RTX 2080 Ti, so that seems a push.

What's changed on GD?

The arrival of the RTX series has brought with an entirely new metric for performance. Nvidia's RTX GPUs have three distinct types of cores used within each chip and our hardware pages have been updated to reflect this. Ray tracing support is now indicated on each graphics card page, as well as the number of Tensor Cores. There's still a question mark of how this will affect gaming frame rates and we will continue to update the hardware pages to reflect this.

As well as this, games will shortly have an icon on their system requirements page indicating whether it supports ray tracing.

So, what do you make of Nvidia's next-gen reveal? More importantly, what do you think of the sky-high pricing, are you still planning to buy a GeForce RTX GPU?

What is RTX Real-Time Ray Tracing And Why is it Good for Gaming?