The Nvidia GeForce GTX 1070 is a popular upper mid-range graphics card, around 35 per cent faster than the most popular GTX 1060 but around 20 per cent slower than the enthusiast GTX 1080. Despite its status in the Nvidia stack, the 1070 somewhat incredibly outperforms the previous-gen Titan X Maxwell and GTX 980 Ti, demonstrating just how far we’ve come in terms of GPU power in the last few years.
As things stand, the GTX 1070 is one of the best graphics cards, alongside the similarly powerful AMD RX Vega 56, to pair with monitors that exceed the usual 1080p and 60Hz standard, such as higher resolution 1440p monitors or high refresh rate 144Hz displays. It can also be a good choice for other scenarios that require just a little bit of extra graphical grunt, such as VR gaming or monitors with an ultra-wide 2560 by 1080 resolution. It’s overkill for 1080p60 gaming, but the most extreme games running at max settings may still require this level of horsepower.
Different people will have different ideas about what represents a playable or ideal frame-rate, so the best option is just to show you exactly what level of performance you can expect from the GTX 1070 to see if it will meet your needs. That’s why you’ll find comprehensive benchmark results for nine different games below, showing how the card performs against its closest rivals when paired with a suitable processor with high, very high or ultra settings selected.
This time, we’re looking at five competing cards. That includes three from Nvidia – the GTX 1060 6GB, 1070 Ti and GTX 1080 – and two from AMD, the Vega 56 and Vega 64. We’ll also examine the card’s performance against a wider range of Nvidia cards, showing you exactly how much progress has been made from from generation to generation, and how the 1070 sits in the last-gen Pascal stack.
Unless you’re viewing this page on a mobile device, each benchmark can be played through in real-time, allowing you to see how the GTX 1070 and the other cards handle the scene as it unfolds. Start the video, then use the controls to the right to add or remove cards until you’ve got the ones you’re most interested in. You can also opt to see just results from 1440p or 4K. Below the live results, there is a static chart showing the average results from the entire benchmark for each of the cards listed. Click on the bar chart to swap between fps measurements and relative performance metrics, expressed as percentages.
If you want to get more information on the system and how it was developed, you can learn about how the Digital Foundry benchmarking system works right here. In any case, we hope you find it useful!
Our first test for the GTX 1070 is Assassin’s Creed Unity, a vintage 2014 game that we employ as one example of legacy title benchmarking. The French Revolution was a challenging time in many ways, and so too is this particular benchmark which stress-tests both compute power and VRAM allocation. The GTX 1070 manages to hit 60 frames per second at 1440p during most of the benchmark, but some dips move the average down to around 53fps. That’s about 14 per cent behind the GTX 1070 Ti, and 20 per cent behind the GTX 1080. Meanwhile in AMD land, the Vega 56 turns in a near identical average result as the GTX 1070, but with much greater variation – it seems that AMD’s graphics cards really don’t like the depth of field effect employed in this test scene.
Ashes of the Singularity is our next test. This DirectX 12 title seems to play better on AMD cards and so we’d expect the GTX 1070 to be outgunned by the Vega 56 here. That expectation is upheld, with the Vega 56 providing a score that’s 10 per cent higher. Regardless, the GTX 1070 still provides a comfortable 65 frames per second at 1440p and scrapes a playable 50 at 4K. For that ideal 60 frames per second, you’ll need to step up to a GTX 1070 Ti or turn down some settings.
Battlefield 1 remains a favourite, thanks to its frantic implementation of Great War combat a century after the fact. This 2016 title doesn’t come with a built-in benchmark, so we’re using a section of the game’s single-player campaign for testing purposes. That means you can ignore the frame-time spikes, which come due to random close-range explosions during the level. The 1070 is able to handle 1440p well here, just 10 per cent behind the GTX 1070 Ti, but AMD’s Vega cards are the overall winners.
But can it run Crysis 3? In the case of the GTX 1070, the answer is a resounding yes! Our train-bound benchmark from the 2013 game shows a comfortable 68 frames per second at 1440p, dropping to a cinematic 32.5 frames per second at 4K. That 1440p result is a healthy 27 per cent faster than the GTX 1060, but 12 per cent behind the GTX 1070 Ti. Meanwhile, the Vega 56 is just one or two per cent slower than the GTX 1070 at both 1440p and 4K.
The Division from 2016 is another challenging benchmark, set on the snowy streets of post-pandemic New York City. AMD cards tend to do better on the DirectX 12 version of this benchmark, but we’re using the DirectX 11 renderer here so Nvidia should have more of a chance here. Regardless, the Vega 56 still outmuscles the GTX 1070 by 13 per cent at 1440p. The GTX 1070’s result of just under 60 frames per second at 1440p is still quite playable, however. Meanwhile, the GTX 1070 Ti offers an 16 per cent better result than the GTX 1070 at 1440p.
Stone Age title Far Cry Primal was released in 2016, meaning it came out between Far Cry 4 and 5. Primal tends to perform a little better on Nvidia cards, so we’re expecting a good result here for the GTX 1070. Indeed, the mid-range Nvidia card is only eight per cent behind the Vega 56 here, which is a closer gap than we’ve seen on other tests. The GTX comfortably gets more than 60fps at 1440p, but this drops to 34fps at 4K.
Ghost Recon Wildlands is the most demanding test in our benchmarks to date, particularly at the 1440p and 4K resolutions we’re looking at here. At 1440p, the GTX 1070 sits 10 per cent behind the GTX 1070 Ti, but only seven per cent behind the Vega 56. Ghost Recon Wildlands will be playable at the 40 frames per second we’re getting, but it might be an idea to use a G-Sync display to ensure it is smooth too.
2016’s Rise of the Tomb Raider benchmark (tested here under DX12 rather than DX11) contains three scenes: a snowy mountain pass, a mystical tomb and a verdant forest. The GTX 1070 scrapes 73 frames per second on average across the test at 1440p, with the GTX 1070 Ti boasting a 14 per cent better result and the Vega 56 ahead by 10 per cent. The game is even somewhat playable at 4K, though the benchmark flatters to deceive – careful settings tweaks should still keep you above 30fps, however.
2015’s multiple game of the year winner, The Witcher 3, is our final standard benchmark test. The GTX 1070 comfortably handles 1440p here at 68 frames per second on average, dropping to 38 frames per second at 4K. Both the Vega 56 and GTX 1070 are around eight per cent faster at the lower resolution, while at 4K the gap becomes smaller with the GTX 1070 Ti (six per cent) but larger with the Vega 56 (nine per cent).
In this special cross-generational matchup, we’ll use Assassin’s Creed Unity to show how the GTX 1070 compares to older Nvidia cards, all the way back to the GTX 700 series. Use the Kepler, Maxwell and Pascal buttons to see just how the Nvidia cards of the same generation compared to one another, or you can keep things as they are to see the generational leap. The jump offered by Pascal is pretty big, right?
Finally, we’ll look at how the GTX 1070 compares to other cards in the Nvidia 10-series lineup. The gap between the cards tends to expand with higher resolutions. We’re using 1080p here as it’s the most popular display type in use right now.
OK, that brings our benchmarks to a close! For more on the GTX 1070 and other strong video cards, why not read the Digital Foundry GTX 1070 review?
Now that you’ve seen the benchmarks for one card, why not check out see which GPUs we recommend? Click through to see Digital Foundry’s updated selections for the best graphics cards and for the best gaming monitors of 2018.