AMD says their new Radeon RX 6800 graphics card beats the Nvidia RTX 3070, but it also costs more money. I’m going to compare both in 13 games at 4K, 1440p, and 1080p resolutions as well as content creator workloads to find out what the differences are and see if the 6800 is worth paying more for.
Let’s start with the spec differences, I think the main thing to note is that the RX 6800 has double the total memory capacity of the RTX 3070. The 6800’s memory is also faster, it uses more power, and costs more money, so I’m expecting it to win in games. It will come down to how much it’s ahead by, whether or not the extra money is worth it, and of course if either are actually in stock.
Test PC Setup
This is the system that I’m using to test out both graphics cards, it’s got an overclocked i9-10900K at 5.2GHz with 32gb of DDR4-3200 CL14 memory in dual channel. This should keep things fair for each GPU, however Radeon 6000 series GPUs can get a boost with an AMD 5000 series processor, but that’s a topic for another time. We’ll get into the gaming results first, then check out cost per frame, power draw, and content creator workloads afterwards.
Assassin’s Creed Valhalla saw one of the biggest differences out of all games tested with the 6800 coming out significantly ahead of the 3070. The 6800 was over 30% higher in average FPS at 1080p and 4K, but 25% higher at 1440p, so great results for the Radeon card, but it will be interesting to see if future Nvidia drivers optimise for this newer title. Microsoft Flight Simulator was tested in the Sydney landing challenge, and although the 6800 was winning again, the differences are lower compared to most other games tested, with just a 5% lead at 1440p and about the same average frame rate at 4K, the smallest differences out of all 13 games we’ll be looking at. Red Dead Redemption 2 was tested using the games benchmark, and also saw below average differences between the two, granted the 6800 was still ahead, but not really by large enough margins to make spending $80 particularly worthwhile.Battlefield 5 was interesting, because at 1080p the 6800 was actually hitting the games 200 FPS frame cap, and as a result the 1% low throttles back, something I often saw in my recent Zen 3 CPU testing. The 6800 was able to do much better at higher resolutions where we’re not limited though, coming out 31% faster at 1440p and 16% faster at 4K. Shadow of the Tomb Raider was tested with the game's benchmark, again the 6800 was ahead regardless of resolution and was able to reach a nice average frame rate of 69 FPS at 4K, putting it 13% ahead of the 3070, and about 14% faster at 1440p. Watch Dogs Legion was also tested with the games benchmark tool, but this one saw below average differences between the two with not too much of a difference. The 6800 was just 6% faster at 1440p with no real change at 4K, likely not one you’re going to notice in any case. For Control I’ll start with ray tracing off results. This title tends to favour Nvidia, though the 6800 was still ahead, granted at smaller margins compared to most other games tested. The 6800 was 8% faster at 1440p and just 3% ahead at 4K, which in this case is about 1 frame, so again not exactly a difference you’ll notice. I’ve tested ray tracing with the high profile enabled. No DLSS was used here, as that’s an Nvidia only feature, however that would also be an advantage to the Nvidia platform in the games that support it.
Anyway the 3070 sees its first win now, and it’s doing a fair bit better compared to the more expensive 6800. To be fair though, neither were playable at 4K, and the 6800 can still surpass 60 FPS at 1080p. This isn’t too surprising given it’s been made with the Nvidia implementation in mind as that’s what’s been available for two years now. Dirt 5 on the other hand was a massive win for the 6800. This game saw the largest difference out of all 13 games tested with the biggest performance gap at all 3 resolutions. Even the 1% lows from the 6800 smashed passed the averages from the 3070. I’ve tested this one with ray tracing on too, however as it was designed with RDNA 2 in mind we’re not seeing the 3070 doing better as we did in Control, so basically it’s really going to depend on the specific game and how it’s optimized to support different types of ray tracing. I’ve tested Metro Exodus with the game's benchmark tool. Not too big of a difference here in averages, the 6800 was 7% ahead at 1440p and then 10% higher at 4K, however often there was a more notable difference to 1% low. The differences in The Witcher 3 were a bit larger, the 6800 was 10% faster at 1440p and 12% faster at 4K, though again the 1% low performance was often seeing larger improvements than the averages with the 6800. Death stranding had some interesting results, although the 6800 was ahead in average FPS at all resolutions, the 1% lows from the 3070 were ahead at 1080p and 1440p, though it’s not by too much compared to the average FPS differences. Call of Duty Modern Warfare was tested in campaign mode. Similar differences to average frame rate in this game too, with the 6800 11% faster in terms of average FPS at 1440p and 10% faster at 4K. Rainbow Six Siege was tested using the games benchmark tool with Vulkan. This was another where the 3070 was doing a little better in the 1% lows, however the averages generally saw bigger gains with the 6800 comparatively.
1080p Game Results
At 1080p on average over all 13 games tested the 6800 was around 13% faster than the RTX 3070 in average FPS. Dirt 5 and Assassin’s Creed Valhalla are definitely boosting the overall average, but even if we consider these as outliers the 6800 is still 8% faster in the rest of the games, the 3070 wasn’t winning in any titles.
1440p Game Results
Stepping up to 1440p and the 6800 now has a larger lead over the 3070, coming out 14% faster on average in the 13 games tested, however results can still vary significantly based on the specific game.
4K Game Results
At 4K the 3070 is able to catch up a little, so the 6800 now has a lower 10% lead. The 6800 was basically ahead in all games, the exception being flight simulator down the bottom where the difference is basically negligible and within margin of error.
Cost Per Frame
Although the 6800 is performing better in pretty much all games, it’s important to remember that it also costs 16% more money too. As a result, the 3070 ends up offering slightly better value in terms of cost per frame, however at 1080p and 1440p the differences are quite close together, then at 4K the gap is a little bigger.
Which for gaming?
Honestly if I was looking to spend more than $500 USD on a graphics card, I’d probably be leaning towards the 6800, and not just because it performs better today. Although the 3070 offers slightly better value in terms of dollar per frame, the differences aren’t that big.
Basically I’d be willing to spend the extra $80 for the extra VRAM which is going to be more useful in future, especially at higher resolutions. If I was playing games at 4K it just seems worthwhile paying the $80 extra to get double the VRAM capacity. It will definitely be interesting to revisit this comparison a year or two from now when games start using more VRAM. It wouldn’t surprise me at all if the lead the 6800 currently has gets bigger over time.
Currently the 3070 does offer a better experience for ray tracing, though it does depend on the game, but this makes sense given this has 2nd generation ray tracing hardware while AMD is only just getting in on it. Again it’s going to be interesting to compare both in future once upcoming games start optimizing for both.
The 6800 was using a little more power in Control at 4K, 3% more wattage for around 3% higher frame rate, which sounds pretty reasonable to me.
Performance Per Watt
When looking at performance per watt over all 13 games tested, the 6800 comes out a little ahead as it’s hitting higher frame rates, so the 6800 appears to be more efficient with the power available. Contrary to popular belief, these things are used for more than just gaming.
Content creator benchmarks
I’ve also compared both in content creator workloads. I’ve tested DaVinci Resolve with the Puget Systems benchmark, however the 6800 was scoring more inconsistently. Although I’m taking an average here, the scores ranged between 932 and 1042 for the 6800, so best case similar to the 3070s average, perhaps the software will need an update to better work with Radeon 6000 graphics. I’ve used Adobe Premiere to export one of my review videos at 4K, and this time lower is better, and the 3070 was around 6% faster here, still another win for the 3070. Blender was tested with the Open Data benchmark. Interestingly the 6800 was completing the longer classroom test a little faster, however it was slower in the shorter BMW test.
So all things considered, the Radeon RX 6800 is the better performing option in games, at least outside of ray tracing, where the 3070 does have an edge, while also typically doing a little better in the content creator workloads. The 6800 does cost more money, granted it’s difficult to buy either at the moment due to supply shortages.
As mentioned, the extra VRAM capacity of the 6800 is probably going to make it perform a bit better years from now compared to the 3070. That’s just my best guess based on what we’ve seen in the past from cards that have more VRAM. So that could potentially shift the value proposition towards the 6800, if it starts performing even better in games then it might take the lead in terms of cost per frame despite costing more, but that’s something we’re just going to have to wait and see on. I think it’s impressive that AMD are able to offer higher performing options like this, this thing would easily outperform a 2080 Ti too for far less money.
Let me know if you’d be interested in seeing a full comparison between the 6800 and 2080 Ti. I’ve already compared the 3070 with 2080 Ti, definitely worth checking out if you want to see just how much the 2080 Ti gets destroyed these days.