Radeon RX 6800 XT benchmarks leaked, beats RTX 3080 at 4K, but worse Ray tracing performance

Written by Chad Norton on Fri, Oct 23, 2020 2:30 PM

As we gear up for the official announcement, more and more rumors surface regarding the upcoming Radeon RX 6000 series of GPUs by AMD. This time however, instead of focusing on the specs of RX 6000 series we have some alleged benchmarks for the Radeon RX 6800 XT graphics card, and it looks to be pretty on par with RTX 3080 performance.

And by on par I mean the results are kind of all over the place, but eventually balance out. On the one hand you have the RX 6800 XT absolutely destroy the RTX 3080 Founders Edition at 4K, but then loses pretty poorly at ray tracing performance. The good thing though is that the RX 6800 XT seems to be better across the board than the RTX 2080 Ti Founders Edition

There’s a number of factors that can contribute to this, including the fact that this is reportedly just an engineering sample of the Radeon RX 6800 XT for board partners, and so performance is likely to improve. Additionally, drivers have not been finalized, which could explain the poor ray tracing performance as it would not have been properly optimized by the time it gets shipped out to reviewers and consumers.

Okay so let’s jump into the rumored benchmark results. The new leaks come from several different sources and so they’re all slightly different from each other, but relative performance also seems to stay the same.

-------------

Radeon RX 6800 XT 3DMark benchmarks comparison vs RTX 3080 FE and RTX 2080 Ti FE

Source 1:

Benchmark RX 6800 XT RTX 3080 RTX 2080 Ti
Firestrike Extreme 4K 13066 (100%) 10688 (81.8%) 8551 (65.4%)
Time Spy Extreme 4K 8232 (100%) 7977 (96.9%) 7028 (85.4%)
Port Royal 4K (Ray Tracing) 4438 (100%) 5415 (122%) 4285 (96.6%)

According to the first source we can see that the Radeon RX 6800 XT GPU consistently beats the RTX 2080 Ti in terms of performance at every test, though the results get very close on the Port Royal 4K ray tracing benchmark.

The RX 6800 XT scores a 13066 in the Firestrike Extreme 4K benchmarks, whilst the RTX 3080 and RTX 2080 Ti both score 10688 and 8551 respectively. In the Time Spy Extreme 4K benchmark the RX 6800 XT scores an 8232, the RTX 3080 scores a 7977, and the RTX 2080 Ti scores 7028. Finally, on the Port Royal 4K ray tracking benchmark the RX 6800 XT scored a 4438 compared to a 5415 by the RTX 3080, and a 4285 by the RTX 2080 Ti.

In terms of relative performance, that pits the RX 6800 XT at 22% faster than the RTX 3080 and 53% faster than the RTX 2080 Ti at Firestrike Extreme 4K. On Time Spy Extreme 4K, the RX 6800 XT is only 3% faster than the RTX 3080 and 17% faster than the RTX 2080 Ti. On the Port Royal 4K ray tracing benchmark, the RX 6800 XT is actually 18% slower than the RTX 3080, but is still 4% faster than the RTX 2080 Ti.

RX 6800 XT vs RTX 3080 & RTX 2080 Ti relative performance:

-------------

Source 2:

Benchmark RX 6800 XT RTX 3080 RTX 2080 Ti
Firestrike Extreme 4K 12871 (100%) 10531 (81.8%) 8210 (63.8%)
Time Spy Extreme 4K 8230 (100%) 7978 (96.9%) 6933 (84.2%)
Port Royal 4K (Ray Tracing) 4387 (100%) 5351 (122%) 4214 (96.1%)

RX 6800 XT vs RTX 3080 & RTX 2080 Ti relative performance:

-------------

Source 3:

A third source had some slightly different results, but we thought to provide the graph of information anyway so you can have a look. The previous 2 sources are more interesting to compare but this one has more GPUs to compare with, including the entire RTX 30 series so far as well as the RTX 2080 Ti.

-------------

Radeon RX 6000 GPU lineup rumored specs:

Model GPU Compute Units Stream processors Game clock Boost clock Memory Memory bus Memory speed
Navi 21 XTX 80 5120 2040MHz 2330MHz 16GB GDDR6 256-bit 16Gbps
RX 6800 XT Navi 21 XT 72 4608 2015MHz 2250MHz 16GB GDDR6 256-bit 16Gbps
RX 6800 Navi 21 XL 64 4096 1815MHz 2105MHz 16GB GDDR6 256-bit 16Gbps

*Specs listed are entirely speculated and not confirmed, therefore they are subject to change.

-------------

Interestingly, if you compare the 2 different sources, the relative performance of the RTX 3080 compared to the RX 6800 XT stays exactly the same as the first source above. However, the RTX 2080 Ti’s relative performance shifts slightly, but not by much.

So looking at this information, considering that these are 2 different sources that have provided very similar results, we can assume that these results are indeed true. And if that is the case, then we can expect performance to actually increase over time before the official announcement.

It’s also great considering that this is RX 6800 XT which is rumored to be an ever so slightly cut down version of the flagship Radeon RX 6900 XT, which will presumably offer even greater performance. Obviously there’s no way to fully determine how these numbers will translate into gaming performance, but it certainly shows promise.

One thing still remains a bit fishy though, and that’s to do with the specs, as the sources detailed the RX 6800 XT having 80 Compute Units, but according to previous rumors the RX 6900 XT will have 80 CUs whilst the RX 6800 XT will feature 72 CUs.

So either the estimations for what graphics card is being benchmarked here is wrong, or the specs leaked aren’t exactly correct. We’ll have to wait and see when the official announcement begins next week, but it's certainly interesting to look at and speculate performance before then.

So what do you think? Are you excited for the Radeon RX 6000 series? Based on these results, would you consider an RX 6000 GPU? Let us know your thoughts!

Login or Register to join the debate

Rep
-1
Offline
02:55 Oct-28-2020

if the price is right and the power consumption isnt that much of difference than my current gear ill finally able to say goodbye to my decade old gear in peace and go full red

0
Rep
4
Offline
admin approved badge
05:42 Oct-26-2020

NOW I FEEL AMD WINING TOO MUCH, LIKE WHERE DID THIS PERFORMANCE COME FROM,THEY DID 5YRS BACK TO BACK REBRANDS NOW HIGH END GPUS LIKE ITS NOTHING THERES A CATCH, THEY MADE A DEAL WITH THE DEVIL OR THEY PULLING LIKE 1000WATTS. SOMETHING SEEMS FISHY

3
Rep
-1
Offline
02:56 Oct-28-2020

lisa su promise illuminati to take a cut of 80% of their profit to access alien technology obviously

1
Rep
14
Offline
16:26 Oct-25-2020

Hype train found a broom and now its flying

0
Rep
8
Offline
10:50 Oct-25-2020

chill amd
its just chips not drugs Lol

3
Rep
36
Offline
15:21 Oct-24-2020

I would take a higher rasterization performance from 6800XT over 3080 Raytracing performance in a heart beat, especially if the price is right.

5
Rep
76
Offline
admin approved badge
14:29 Oct-24-2020

Looks nice, but those are benchmarks and one thing history thought us, benchmarks never equate to gaming performance. They have their place, sure, but in the end, I won't play Firestrike, Time Spy or Port Royal, I will play games, be it Hitman 2, Red Dead Redemption 2, Outer Worlds, or upcoming Cyberpunk 2077,... so I can't wait for independent reviews.

2
Rep
34
Offline
14:37 Oct-24-2020

Port Royal IS a game, though. But yeah, I agree. Have to see more benchmarks from games. I wouldn't feel bad if it beat my new purchase because here it would probably be more expensive (since the 3rd parties and taxes dictate price, not AMDs MSRP) and more frames for higher price just feels like a natural upgrade.

0
Rep
-19
Offline
15:02 Oct-25-2020

port royale pc game, port royal board, and port royal benchmark are on completely different ends of the pc requirements spectrum. no one is using the first two to benchmark anything these days.

0
Rep
356
Offline
08:00 Oct-24-2020

Good stuff but ill hold on with gpu upgrade I play on low settings on 1080p res. Need better cpu First but looking damn good for next year finally competition on high end to

0
Rep
28
Offline
21:18 Oct-26-2020

You should have been on Intel if you wanted the highest framerates possible at 1080p that was a hell of a gamble to take to think that Ryzen 5000 would beat AMD at gaming and you could slot it in though it's paid off well played my friend.

0
Rep
356
Offline
07:22 Oct-27-2020

At Time Got ryzen for 330eur motherbord 150eur intel 9900k was around 600eur motherbord same tier 220eur

0
Rep
179
Offline
admin approved badge
23:05 Oct-23-2020

Though I'm not in the market for a GPU for this generation, I'm still excited to see how things turn out and if AMD is going to is going to price close to Nvidia(which I suspect they will) or if they're going to claim the value crown again and force a price war(doubtful this time around, but we can atleast hope)

1
Rep
6
Offline
21:42 Oct-23-2020

Still, how is it that in TimeSpy the 3080 has better score than the Big Navi, but not at Firestrike when the first one is supposed to be "5 times more demanding on the GPU" according to their developers? Ampere still has not received a driver to optimize for DX11 games?

0
Rep
179
Offline
admin approved badge
23:00 Oct-23-2020

Its suspected that the last benchmark which shows that inconsistency was conducted on an earlier prototype.

0
Rep
6
Offline
11:04 Oct-24-2020

According to the source of that benchmark it was indeed an ES sample, but with 80CU. So it doesn´t make a lot of sense to me as the 6800XT is supposed to have 72CU and the 6900XT is the 80CU one.

0
Rep
50
Offline
18:35 Oct-23-2020

Man, AMD and Nvidia have really got a thing for thousand numbers, Rs, Xs, and Ts


RTX 3080
RX 6800 XT


If these weren't enthusiast cards I could imagine quite a few people getting confused by these naming conventions.

2
Rep
28
Offline
16:48 Oct-23-2020

I dont understand the fascination with 4k gaming performance when 1440p has double the market share according to steam hardware survey. If big navi can out do rtx 3080 in 1440p i think it will turn quite a few heads.

6
Rep
16
Offline
17:13 Oct-23-2020

It might, as Ampere's architecture is such that favours higher resolutions more than lower ones. That's why the gains against older cards are way higher at 4K than they are at 1440p.
True, 4K gaming isn't quite there yet, but with the arrival of the new Xbox Series as well as the PS5, this might change within a year or two. Better be prepared.

3
Rep
-19
Offline
17:19 Oct-23-2020

also, Ampere seems to favor dx12 over dx11. that is the reverse of what has been traditional performance between the brands.

1
Rep
-19
Offline
17:14 Oct-23-2020

because it shows the raw power of the GPU. if it can push consistent performance (high fps and max settings) then its performance at lower resolutions will be greater for the people that want it. if its beating the 3080 at 1440 but fails over at 4k, then it leaves no upgrade path to a 4k screen and that means some people wont buy it.

1
Rep
-19
Offline
17:15 Oct-23-2020

with my screen setup, i wouldn't buy it if its 4k performance was less than a 3080 because my performance will be lower.

0
Rep
28
Offline
17:38 Oct-23-2020

People dont upgrade their monitors as offen as they do their whole rig maybe 7+ years for one monitor and a new rig in about 4/5 years. Also 1080p is insanely popular to this day.

4
Rep
-19
Offline
17:53 Oct-23-2020

but they could be looking at upgrading the monitor shortly after the GPU and a lower performing GPU at 4k might sway their decision to get a different card only for the fact they know they are going to get a new monitor soon. plus with the holiday coming up and the new GPUs, people might be more apt to purchase a new monitor and GPU this year.

0
Rep
58
Offline
admin approved badge
00:43 Oct-24-2020

Im kinda with Zenmaster on this one. 4K gaming is really over rated..I mean max quality in 1080P with some AA looks great!

1
Rep
-19
Offline
04:52 Oct-24-2020

it was just an example. regardless of what anyone wants to run with their system, we want to see the best the GPUs can do. pushing them at 4k is a way to see that compared to 1440 or anything else.

0
Rep
272
Offline
admin approved badge
17:15 Oct-24-2020

1080p does NOT look great, sorry. I've been playing for far too long at 1440p, 4K and 5K, I suppose, but 1080p doesn't look good anymore.

0
Rep
191
Offline
junior admin badge
18:04 Oct-25-2020

I think we really need an article that goes into depth on that topic.
Xquatrox, you'd be uniquely qualified to show a comparison between the different resolutions.

0
Rep
191
Offline
junior admin badge
18:23 Oct-23-2020

I'm still at 720p, so you know where I stand in that regard.

0
Rep
28
Offline
18:32 Oct-23-2020

The thing about lcd monitors is that the native resolution on any monitor looks great but turning it down any resolution below native looks like tetris/shimmering lines. Old CRT never had this issue.

3
Rep
191
Offline
junior admin badge
23:55 Oct-23-2020

I only had a single CRT before I switched to an LCD monitor, also I rarely changed my screen resolution on my CRT monitor so I can't personally attest to that claim. Tho I can confirm that if you go below the native resolution on an LCD monitor gives you a pathetic img

2
Rep
38
Offline
16:02 Oct-23-2020

If the actual product ends up like this, for like 500-650€ I'll get one. Screw Nvidia

6
Rep
16
Offline
16:38 Oct-23-2020

So will I. Raytracing is something I can do without, but my loyal 1080 Ti is now struggling at 4K. If a Big Navi card delivers an FPS improvement of at least 60-65% at 4K for up to 650-700€, I'm sold.

4
Rep
-21
Offline
15:57 Oct-23-2020

RX 6800 XT beats RTX 3090, PS5 and of course XBSX

1
Rep
58
Offline
admin approved badge
15:36 Oct-23-2020

Makes sense, I mean Nvidia is on second generation RTX so give AMD some time guys!

12
Rep
2
Offline
19:24 Oct-23-2020

Exactly, Nvidia's first iteration of RTX wasn't very impressive and when they focused on improving that the most they made great progress, especially with DLSS 2.0, nobody would expect AMD to beat NVIDIA in something they've been longer at

1
Rep
-19
Offline
15:11 Oct-23-2020

So from what can be viewed here, Navi does great in dx11 but loses a lot of performance in dx12. Why did AMD take such a huge step backwards? Hopefully the released cards will do better from driver and bios fixes.

0
Rep
3
Offline
16:57 Oct-23-2020

Judging from your comment I'm guessing Firestrike is a DX11 performance tool and Timespy a DX12 one? If so is there a Vulkan equivalent that would maybe help provide insight as to whether AMD had been optimising it more for that than DX12? Considering the RX6800 appears to be able to compete with a 3080 is still great for them and us if true anyway.

1
Rep
191
Offline
junior admin badge
14:37 Oct-23-2020

I'll be waiting for game benchmarks before I come to any foregone conclusion.

12
Rep
111
Offline
admin approved badge
14:47 Oct-23-2020

Agreed! Either way i hope the 2 compete just for consumers sake!

11
Rep
191
Offline
junior admin badge
18:21 Oct-23-2020

Hear hear

1
Rep
272
Offline
admin approved badge
15:05 Oct-23-2020

Yeah, benches first!


But this looks like it would match or even beat a 3090 in certain workload and/or gaming scenarios. Juicy!
Maybe Nvidia will finally drop their damn prices a bit... Though the more likely scenario is AMD increasing the prices, because why sell yourself short when you have a competitive product...

4
Rep
3
Offline
17:05 Oct-23-2020

As these are AMD's first Ray Tracing cards I'd expect the prices to be higher than previous gens but still lower than Nvidia's. I guess it depends on the R&D costs of the new tech and how much of that AMD are willing to swallow to provide a cheaper alternative.

3

Can They Run... |

| 30FPS, Low, 720p
Core i5-4200H 2.8GHz GeForce GTX 860M 2GB 16GB
Core i7-5500U 2-Core 2.4GHz GeForce 940M 2GB 8GB
0% No [1 votes]
| 60FPS, Ultra, 1080p
Ryzen 5 2600X 6-Core 3.6GHz GeForce RTX 2060 6GB 16GB
100% Yes [1 votes]
| 60FPS, Ultra, 1080p
Ryzen 5 2600X 6-Core 3.6GHz GeForce RTX 2060 6GB 16GB
| 60FPS, High, 1080p
Ryzen 5 3600XT 6-Core 3.8GHz GeForce GTX 1080 Ti Gigabyte AORUS Xtreme Edition 11GB 16GB
100% Yes [1 votes]
| Low,
Core i5-8300H 4-Core 2.3GHz GeForce GTX 1050 8GB
100% Yes [1 votes]
| 60FPS, High, 1080p
Ryzen 5 5600X 6-Core 3.7GHz GeForce RTX 3070 Gigabyte Eagle OC 8GB 32GB
| 60FPS, Ultra, 1080p
Core 2 Duo E7500 2.93GHz Radeon HD 5450 1024MB 4GB
0% No [1 votes]
| 60FPS, Medium, 1080p
Core i5-9400F 6-Core 2.9GHz GeForce GTX 1050 Ti MSI OC 4GB 16GB
100% Yes [2 votes]
| 60FPS, Ultra, 4k
Core i7-11700K 8-Core 3.6GHz GeForce RTX 3090 32GB
100% Yes [2 votes]
935
| 60FPS, Ultra, 4k
Core i7-11700K 8-Core 3.6GHz GeForce RTX 3090 32GB
100% Yes [4 votes]
935
Core i5-11600K 6-Core 3.9GHz GeForce RTX 3070 Zotac Gaming Twin Edge OC 8GB 32GB
Core i7-4790 4-Core 3.6GHz GeForce GTX 1050 Ti Palit StormX 4GB 16GB
| 60FPS, High, 1080p
Ryzen R5 1600 Radeon RX 570 4GB 16GB
| 60FPS, High, 720p
Core i5-4590 3.3GHz GeForce GTX 1050 Ti Asus Pheonix 4GB 10GB
100% Yes [1 votes]
| 60FPS, Low, 1080p
Core i5-9400F 6-Core 2.9GHz GeForce GTX 1050 Ti MSI OC 4GB 16GB
100% Yes [3 votes]