AMD RX 6800XT

I need a new graphics card soon to pair with an x570 motherboard and ryzen 5900x.

The rx5700xt seems to perform at or around a 2070 super in most games but suffers badly in ACC and is closer to the 2060 in performance I believe?

Do people think the 6800xt will show a similar performance loss to Nvidia cards In ACC? I do play other games but If 3080 holds a sizeable advantage in ACC I might have to stick with nvidia

 

I'm waiting for the reviews of the AMD 6000 graphic cards before I build my opinion. Judging by the launch event it looks promising, but there were some crucial information missing, like what's about DXR (Raytracing), how does it perform vs. a RTX 3000 series or what about Super Resolution in comparison with DLSS ...

Still waiting before I want to decide which would be better to get.

 

The card is different, we can't predict how features like the infinity cache affect ACC.

But with AMDs combined research budget for this design, getting developed for PC, XBOX and PS, the chances should be pretty good to have a competitive cards.

 

Hi,

nobody knows atm we need to wait for independent tests.

I am also in for a new build and tend to 5900X/6800XT/6900 XT.
I think we need to wait till the end of the year how things evolve.

AMD´s SAM (Smart Access Memory) is interesting in combination with Zen 3. https://www.amd.com/de/technologies/smart-access-memory
It´s maybe minor but every frame counts and if you can get it as a bonus i´ll take it.

For me one thing i am waiting for is if they will introduce DLSS 2.1 into ACC, then i will probably go with NV, although i doubt it, because it limits the userbase.

More interesting is what they will do with DirectML/FidelityFX/Super Resolution where everybody could benefit with an DX 12 compatible card, if i have understand that correctly.

Early next year i think it will be a lot more clear what makes the most sense, at least for me in the regard of ACC.

cheers

 

At 4K on EPIC settings, the 3080 is a whopping 29% faster than 6800xt. The 3080 averages 84FPS while 6800xt averages only 60FPS.

Source: https://babeltechreviews.com/vr-war...80-15-vr-games-performance-benchmarked/?amp=1


There is also VR benchmark on Vive pro where 3080 is only 4% faster. However, since the average FPS was 138FPS, they probably faced some CPU bottleneck with 10900k.

In any case, it seems that Nvidia is still a way to go if you upgrade to make the most out of ACC.

Based on this, even the upcoming 6900xt will struggle against lower tier 3080.

Rather disappointing, as I really wanted to upgrade from 3080 to 6900xt...

 

They used an Intel, so no SAM for the radeon, in combination with a loaned nvidia FE card.
If you load their main page, take a look at those huge RTX banners left and right: https://babeltechreviews.com/
On the about us page, the founder wears an nvidia badge around his neck.

I don't expect a completely unbiased review on that page It might be accurate, it might be not, definitely waiting for other benchmarks.

 

SAM is not that relevant as the same feature will soon be enabled on nVidia as well.

More benchmarks will definetly be needed, but their results from other games are in line with results of other reviewers.

 

In that case DLSS is irrelevant as well, because something similar will be available on AMD as well. However, it seems that anything AMD still sucks big with ACC, and that is ridiculous. I wish someone would explain why, and if it is just the Unreal engine, then that engine is crap and needs to go.

 

What i'm confused about is the hughe gap between the VR High Preset and 4k Epic Preset benchmark.

Vive Pro has a display resolution of 2880x1600. Render resolution of 4032x2240. Plus x1.2 Pixel density. Thats 4838 x 2688 (13M pixel)
>3080 is ~10% faster with slightly less consitent Frametimes

4k is either 3840x2160 or 4096 × 2160. (8.3 / 8.8M pixel)
>3080 is 39% faster

There are a lot more pixel to render at VR High Preset on a vive pro compared to 4k. Logically the 3080 should be even better off in vr compared to 4k. So either:
1) i'm missing something
2) they messed up the benchmark
3) One of the graphics settings absolutly destroys amd cards

 

Take a look how many games are using Unreal engine https://en.wikipedia.org/wiki/List_of_Unreal_Engine_games#Unreal_Engine_4
Hundreds of titles, from Borderlands and Fortnite to Valorant.

 

But it isn't made for (Sim) Racing scenario as I understand.

 

Well, many games use Unreal, also version 4, and although I don't know how AMD performs in all of them, but you can find many statements that say that Unreal 4 does favor Nvidia over AMD. That this engine might not made for the racing scenario is irrelevant, this is about Nvidia performing too much better in ACC. So is the explanation really just that the Unreal 4 engine doesn't like AMD, or is it ACC, if the difference is bigger in ACC than in other engine 4 games?

If the problem is indeed this engine, then the next question is why Kunos chose such a crap engine in the first place, assuming Kunos didn't deliberately want to spit in AMD cards owners faces? IMHO a game should work within reasonable margins with both major videocard producers, or otherwise warn people not to buy their game if you don't have a specific brand videocard.

 

If you use iRacing at all the SMS and SPS tech from the NVIDIA cards is worth another 15-30% so NVIDIA is still a no-brainer at this point for me.