Skip to main content

AMD RX 6800 and 6800 XT review: revisiting the super-performers

Doom Eternal, Control, Borderlands 3, Shadow of the Tomb Raider.

Three new titles and one returning favourite kick off our inaugural set of RX 6000 series game benchmarks. All four are modern games using advanced graphics APIs, like DirectX 12 and Vulkan, that showed some of the biggest gen-on-gen performance improvements in our RTX 30-series testing. These modern interfaces allow proper CPU utilisation, minimising processor bottlenecking at lower resolutions, and therefore should run well on the Big Navi cards too.

We've run our benchmarks at three resolutions: 1080p, 1440p and 4K. (We have been asked to run ultra-wide benchmarks too, but you can extrapolate expected performance here pretty easily - 2560x1080 is somewhere between 1080p and 1440p, while the more common 3440x1440 is almost exactly between 1440p and 4K.)

Our benchmark results are presented a little differently to what you might be used to elsewhere on the web. On mobile, you'll get a basic overview, with metadata from the video capture of each GPU being translated into simple bar charts with average frame-rate and lowest one per cent measurements for easy comparisons.

On a desktop-class browser, you'll get the full-fat experience with embedded YouTube videos of each test scene and live performance metrics. Play the video, and you'll see exactly how each card handled the scene as it progresses; you can even choose exactly what GPUs at what resolutions you're interested in and it'll update in real time. Below the real-time stuff is a bar chart, which you can mouse over to see different measurements and click to switch between actual frame-rates and percentage differences. All the data here is derived from video captured directly from each GPU, ensuring an accurate replay of real performance.

Right, let's see how big Big Navi really is.

Doom Eternal

Doom Eternal kicks off the proceedings, and as a Vulkan title built by id Software's legendary designers it's a clear choice for anyone looking for what next-gen GPUs can do with the right hardware at their disposal. Nvidia's Ampere cards do well here, so does Big Navi also make the jump to next-gen with aplomb?

The RX 6800 does well right out of the gate, holding a 15 per cent advantage against the RTX 3070 at each three resolutions we tested - just in line with its 15 per cent higher launch price (assuming a perfect world where cards were actually available at the prices that their manufacturers announced).

The RTX 3080 manages to fend off the 6800 XT, but its advantage at 4K is a manageable seven per cent - again, reasonable given that the Nvidia card costs about that much more. At 1440p, the RX 6800 XT is better value, losing the head-to-head against the 3080 by only three per cent, and at 1080p we actually see the AMD card take the lead. So what's behind this improved performance at lower resolutions? It might be down to the difference in memory subsystems - the 3080 packs fast GDDR6X, while the 6800 XT uses slower GDDR6 augmented by its 128GB Infinity Cache, so perhaps the AMD system gets a little overloaded with 4K textures in play but copes better at 1080p and 1440p? It's not clear, and there could be other architectural differences in play too.

As well as offered competition to the RTX 3080, the RX 6800 XT also represents a massive jump forward in peak AMD performance. We see a 91 per cent jump in 4K performance from the (originally $400) 5700 XT to the ($650) RX 6800 XT - and the RX 6900 XT's still waiting in the wings to stretch that advantage even further.

Doom Eternal: Vulkan, Ultra Nightmare, 8x TSSAA

Borderlands 3

Borderlands 3 is another recent release, one that also ticks the oft-requested 'test an Unreal Engine game' box for us. It's operating under DX12 for these tests, which should provide the best CPU utilisation on high-end processors like the Core i9 10900K in our test rig (and we do compare this to the Ryzen 9 5900X later!). The game's Bad Ass graphical preset is also one of the heaviest tests in our suite, but the RX 6800 and 6800 XT perform brilliantly.

The 6800 manages 54fps at 4K with all settings turned up, a big 61 per cent improvement over the 33fps of the RX 5700 XT. Compared to the RTX 3070, the 6800 is more than 20 per cent faster - a big lead.

Looking now at the RX 6800 XT, and the AMD card also beats out its Nvidia competition at the high-end with a three per cent advantage - despite the 6800 XT costing $50 less. While the 4K results are impressive, the lower resolutions are slightly better, with the 6800 eclipsing the 3070 by 22 per cent and the 6800 XT beating the 3080 by five per cent; these leads become 20 and seven per cent at 1080p.

When compared to the RX 5700 XT, the 6800 XT is 88 per cent faster at 4K. So across the board, that's a big win for the Radeon team.

Borderlands 3: Bad Ass, DX12, TAA

Control

Control is a brilliant game from Remedy, but it's sadly one that doesn't suit the RX 6800 series cards - even with all RT features disabled. Our run comes from the very start of the game, with a brief in-engine fly-by, a run around the offices of the mysterious Federal Bureau of Control and a run-in with Ahti the janitor. Unfortunately, AMD's latest cards fail to live up to expectations with stable but low frame-rates throughout.

The 6800 falls two per cent behind the RTX 3070 at 4K, with the advantage flipping at 1440p and becoming a six per cent margin for the 6800 at 1080p. That's great, but if you're paying $80 extra for the RX 6800 you'd expect to see a greater advantage. The RX 6800 XT faces an even tougher time of it, with the 3080 beating it by a comprehensive 23 per cent at 4K - enough to drag a near-60fps game on Nvidia hardware to the mid forties on AMD. Things do improve at lower resolutions, where the gap lowers to 17 at 1440p and then 12 per cent at 1080p, but it's still a bad loss for the RX 6000 series line-up - an outlier compared to our results so far, so potentially one that might be fixable with driver updates?

We'll look at Control's RTX performance later on in this same review, so stay tuned for that!

Control: High, DX12, TAA

Shadow of the Tomb Raider

Shadow of the Tomb Raider closes out this page, and once again it's a modern DirectX 12 title with a challenging benchmark, this time of the built-in variety. Unusually, this benchmark is comprised of three distinct sections, requiring a little more 'cutting out' for us but showing how different scenes can vary in performance substantially. It's well worth playing back the video to see how the cards are grouped closer in some sections than others.

Let's start with the 6800 vs 3070 battle, and thankfully it's a return to form for AMD here with the 6800 holding a 14 per cent advantage over its nearest competitor. The gap grows to 17 per cent at 1440p, but shrinks down again to six per cent at 1080p where we're much more CPU bound. It's rare to see a performance trend like that, and again hints that 4K textures may be slightly overwhelming AMD's Infinity Cache system. Still, a 17 per cent lead at 1440p is great if that's the resolution of your monitor, and even at 1080p you're still getting a highly playable 177fps experience on RX 6800.

Now for the heavyweight bout: 6800 XT vs 3080. The margin is eight per cent this time, again just in line with the cards' price points. The RX 6800 XT and RTX 3080 effectively tie at 1440p and the 6800 XT narrowly wins at 1080p, making the Radeon challenger the better value option overall for Shadow of the Tomb Raider.

Looking back at the last-gen RX 5700 XT, we see a big boost but not as big as we saw on other games so far - the RX 6800 manages a 46 per cent lead over the older card, while the XT model delivers a 57 per cent higher frame-rate.

Shadow of the Tomb Raider: Highest, DX12, TAA

It's a competitive start from the Red Team, but it's far from the final word. Let's take a look at some other games where Nvidia's RTX 30-series cards didn't do as well - will this be where the RX 6000 series makes a greater case for itself?

AMD Radeon RX 6800 and RX 6800 XT Analysis