50 Games Tested: GeForce RTX 3080 vs. Radeon RX 6800 XT

[W1zzard]

In this article, we have tested 50 games on the NVIDIA GeForce RTX 3080 and AMD Radeon RX 6800 XT to conclusively determine which card offers better performance. Our RTX 3080 vs. RX 6800 XT performance testing has results at 1080p Full HD, 1440p, and 4K Ultra HD.

Show full review

 

[Exilarch]

I like the fact that there are a lot of games there to help us make an informed decision. I dislike the fact that you can't buy either of them for a normal price.

 

[human_error]

Nice article - loving the wide range of games and splitting them up by age and API. As you alluded to it must have taken a long time to put together so thank you

Liking the fact you've got Guardians of the Galaxy in there - have been playing it a ton the last couple of days. I'm wondering if you have tried it with RT on with the 3080 as on my 6900 I get very noticeable micro stutters with it enabled, even with no obvious reflective surfaces in the scene. I'm not sure if it's just a problem for anyone, or just an issue on AMD GPUs (I am at 4k seeing good frames so I guess I'm being fussy, just wish the stutters didn't force me to keep RT off).

 

[KarymidoN]

I love reading those reviews so much, keep it up plz
This gigantic test its just amazing, i can't imagine the amount of time and work that went in to it.

 

[Siph]

I really wish some site would include HUNT:SHOWDOWN in their games catalogue. Why is it overlooked constantly??

 

[Outback Bronze]

Huge thanks W1z!!

Massive task. I'd have buckleys finding the time to do this. Nice work

Edit: I suppose a good one to add in for comparison sake would be Power Consumption. RTX performance could be too but I'm pretty sure we know where that leads..

 

[Rowsol]

I think you're wasting time doing multiple resolutions. Just stick to 1440p and call it a day.

 

[Lianna]

Just an idea: games sorted by fps instead of names (may be the fps of card shown at the graph's middle point), and showing actual fps (for both cards) beside the graph.
Reason: e. g. +27% advantage is important if the slower card shows 35 fps, may be null if slower card show 10 fps (or, depending on your monitor and v-sync preferences, 70fps? or 150 fps?).
If this is to be regular feature, you may test slower cards, too.

It would be great if the minimum fps (or preferably 0.1%, or 1%, or whatever similar metric you choose) could be incorporated into results - as an option? additional graphs? It's the metric that may better represent gaming experience especially for average fps below 60.

 

[Metroid]

Incredible article, 6800 for vulkan and dx12 and 3080 for everything else.

 

[chispy]

Excellent content @W1zzard , thank you so much for this. Great choice of different games , apreciate it the hard work.
I would like to see at least one or two 3dmark benchmarks just for comparison reasons , 3dmark timespy or firestrike extreme and Unigine superposition 1080p extreme would be a very nice addition to this content . Again thumbs up for the hard work.

 

[Debaser]

Thanks for the nice work, I think Days Gone is a DX11 title but it's listed as DX12/Vulkan here.

 

[Cryio]

Regarding the last paragraph: I've seen on YT multiple 6900 XT that indeed hover around 320W, yet 3090 are doing 400-450W constantly. That's ~50% more power for not even 10% performance difference, yet people are not really calling Nvidia out on this. It's only a minor, passing mention, always, for some reason.

 

[wheresmycar]

I love (and massively appreciate!!!) what you've done here... its always "super" useful to see how well these cards compare with individual titles. I only play a couple of games and both are listed in the benchmarks @ 1440p. I guess now I have a more informed decision before considering my next upgrade.

 

[Richards]

Some games on the list are not high quality

 

[Totally]

Damn, GTA V carried the 3080. Makes me wish the graphs were interactive so that entry could be dropped to see how much it skewed the results.

 

[R-T-B]

I think you're wasting time doing multiple resolutions. Just stick to 1440p and call it a day.

Sounds like someone games at 1440p.

I for one appreciate the 4K data.

 

[Makaveli]

This was nice and alot of work thank you.

 

[phanbuey]

Great review - DLSS 2.0 is the only reason I went with the 3080 -- It's been a lifesaver in the games that have it at 4k 120. Its the only reason Cyberpunk is even remotely playable, and nice to have for the RDR2 playthrough.

RDNA 3 is going to be interesting -- I think nvidia is losing the next round...

 

[Sybaris_Caesar]

Nice review! Regarding last paragraph, I think you should do a separate page for graphical settings used and FPS/frametime histogram for each game. No writeup needed. Just an extra set of data to check out.

 

[Caring1]

Why did clicking the link to the "review" lead me to a front page article about this, instead of a review as per usual?

 

[Salvo39]

Awesome coverage. Thank you. I had some suggestions/ideas on potential additions.

• It would be great to see a multi-point curve. Something like this perhaps every 6months of the driver release. It will clearly show the Nvidia's or AMD driver improvements over time. For the same 50 games off-course
• I would love to see if RAM speeds would matter. I could see this being a very long test however it might be super-duper useful. Perhaps you only do this once a year.
  As DDR5 rolls in we are welcoming a world where there are 2 different RAM topologies being used for gaming, it would be interesting to see how
  • DDR4 vs DDR5 RAM speeds work vs
  • DDR4 different RAM speeds work vs
  • DDR5 different RAM speeds work
• Windows 10 vs Windows 11, As W11 matures, it would be nice to see same suite of games being run with W11 scheduler, Is the new scheduler more sensitive/faster compared to RAM speeds? Would be an interesting take.

Just some of my thoughts, keep the good working coming

 

[wolf]

What an excellent article and I really appreciate the insane amount of time and effort that went into this. Not that much changes honestly, the 6800XT was always the direct competitor to the 3080 and both trade blows constantly. The margins changed ever so slightly, but so did the games tested, the test system itself and configuration options (like ReBAR), so pick whichever is the least worst value if you're in the market to buy today, while considering the unique aspects both bring to the table.

 

[chrcoluk]

I feel bad about posting this comment as the amount of effort for this testing must have been a lot.

But I wish there was DX9 games on that list which would have felt like you tested a reasonable amount of the market.

DX9 games still make up over 1/2 of the games I play on my PC.

 

[bhappy]

I think testing with Ray Tracing turned off is a big oversight. Most people who buy these cards will be using ray tracing features, otherwise why would you even need to buy a RTX 3080/6800XT for, unless you are in the minority trying to achieve super high FPS? Just buy a cheaper card if you're not using Ray Tracing and save your money towards your next upgrade. TPU benchmarks should be reflective of how people will most likely be using these high end cards and that's with Ray Tracing features turned on and all the graphical settings set to max. These benchmarks aren't reflective of how most high end graphic card owners myself included would use their cards.

 

[R-T-B]

Most people who buy these cards will be using ray tracing features, otherwise why would you even need to buy a RTX 3080/6800XT for, unless you are in the minority trying to achieve super high FPS?

I personally feel those shooting for super high fps probably exceed the interest in raytracing, but fully admit I have no data to back that up.