• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

AMD Radeon RX 6900 XT

Very impressive. It's great to see competition at the high end of the GPU market once again!
 
Not really... BUT yeah I agree, 3090/6900XT are intended for ppl with more money then sense :)
View attachment 178709

View attachment 178710

View attachment 178711

Yes, really.

Equal to 3080:
relative-performance_1920-1080.png


Equal to 3080 Again:
relative-performance_2560-1440.png


Equal to 3080 Again:
relative-performance_3840-2160.png


Don't post random graphs from an unnamed review in a review for the darn card. We trust the results of TPU, not random graphs you made up.
 
Yes, really.
Don't post random graphs from an unnamed review in a review for the darn card. We trust the results of TPU, not random graphs you made up.
These numbers are based on 18 games average from Steve's HardwareUnboxed video, one of the most trusted reviewers. Same conclusion was made by Gamers Nexus, 1080p and 1440p win for 6900XT, 4K and RT/DLSS win for 3090. Both said you should not buy either.

Here are full reviews:
Radeon RX 6900 XT Review, AMD's Fight For the Top - YouTube
AMD Radeon RX 6900 XT GPU Review & Benchmarks: Undervolting, Gaming, Power, Noise - YouTube
 
Which is about as "equal" to a 3090 using the same logic. :kookoo:

Don't know what you're on about.

A 5-8% is a difference, a 1-2% is not. It's a pretty simple concept to grasp for most people.

These numbers are based 18 games from Steve's HardwareUnboxed video, one of the most trusted reviewers. Here's the full review. Same conclusion was made by Gamers Nexus, 1080p and 1440p win for 6900XT, 4K and RT/DLSS win for 3090.

Steve and GamersNexis are both hacks compared to TPU when it comes to performance testing.
 
I expected a bit more too frankly, I think the 3080 or the 6800 XT are better buys.
 
Ignoring the fact that you can't buy one anyway, this is the most pointless card in the world. It's the card that the 6800XT should have been, and thanks to poor optimisation or drivers doesn't even match the 3080 that the 6800XT is supposed to rival.

IMO, the 6900XT should have been a larger chip - it doesn't have enough shaders to compete with the 3090 at sensible power or cooling limits. At $999 MSRP it's a joke, because it's only even a match for the $699 3080FE in traditional raster-based loads anyway. The minute you turn on DXR it's fighting for survival between the $399 and $499 cards from Nvidia.

Once all the dust settles from the scalping and stock problems, I'd expect to see the 6900XT sell at $699 since it's still down on DXR, DLSS, and NVENC compared to the 3080 to name just a few shortcomings. The 6800XT belongs at the $579 price point the 6800 launched at - it's quite clearly fighting the 3070 in many of the tests and falls short of it with any mention of DXR or DLSS, something that's definitely gaining traction in 2020 with both new consoles including it.

As for the vanilla 6800, it's a runt using massively-defective silicon. 25% of the card disabled? Jesus! Hopefully that model will vanish altogether when yields improve and when the 6800XT comes down to it's $579 price point. If you were after raytracing performance, just get a 2018 $349 RTX2060 6GB. That's honestly how bad it is. In fact, the $299 2060KO is about equal :\
 
Last edited:
you know I really hoped this increased competition would have led both brands to open up everything in terms of overclocking to get the advantage, disappointed honestly.
 
you know I really hoped this increased competition would have led both brands to open up everything in terms of overclocking to get the advantage, disappointed honestly.

What would have "increased competition", seeing AMD at the top of chart with 1% more performance or something like that ?

Nothing else cuts it ? Is that really what the average consumer is looking for ?
 
What would have "increased competition", seeing AMD at the top of chart with 1% more performance or something ?

Nothing else cuts it ?
i was just thinking of removing limits imposed by companies is all. is someone who spends $1000 dollars on a GPU the average consumer?
 
You know what? I don't care if this doesn't beat the 3090 across the board or if its RTRT performance kind of sucks. I hope this is enough to curb Nvidia's trend of putting an arbitrary price tag on cards offering minute performance improvements.
I understood why Nvidia did it and I didn't mind much, because I'm not buying into that price bracket anyway. But I'd love to see that practice put to an end.
 
is someone who spends $1000 dollars on a GPU the average consumer?

I don't know but I imagine he does not seriously expects that the 1000$ product handily outperforms the one which is 1500$. Is 8 % less performance for 50% less money not competitive ?
 
Last edited:
5-8% for 110~150% more price? Great logic there!
I didn't say the 3090 was a good buy either. But, if I'm buying something that is actually better than the 3080, the only option is the 3090 right now. And paying $1,000 for a 6900XT just to get identical performance to a 3080 for $300 cheaper is an even worse buy than the 3090.

I hope this is enough to curb Nvidia's trend of putting an arbitrary price tag on cards offering minute performance improvements.

And yet this is a perfect example of AMD putting an arbitrary price tag on a card offering minute performance improvements.
 
Slightly better performance for the same price. The 3080's 700$ price is just fake.

They're all fake, arguing that one is better than the other is pretty bewildering.
 
Yes, really.

Equal to 3080:
relative-performance_1920-1080.png


Equal to 3080 Again:
relative-performance_2560-1440.png


Equal to 3080 Again:
relative-performance_3840-2160.png


Don't post random graphs from an unnamed review in a review for the darn card. We trust the results of TPU, not random graphs you made up.

There are 5 games where the AMD DX11 driver is causing a lower fps ceiling than the NV DX11 driver. AC:O, B3, D:OS2, FC5 and PC3. In all cases the fps decrease for the 6900XT and 3080/3090 from 1080p to 1440p and is 0 but the 3080/3090 are between 10% to 30% faster in these titles. Doing a simple average of those 5 games alone @ 1080p the 6900XT gets 113fps and the 3080/3090 get 131 fps giving the 3080/3090 a 16% performance advantage. For the 3080 to be 1% faster on average in the summary despite this 16% advantage in 5 of the 23 games tested shows that for the other 18 games the 6900 is actually faster at 1080p like the HUB numbers show.

In all cases the 6900XT is getting 95+ fps average so it is not exactly performing badly and if any of those games are a persons primary game then it is an entirely valid reason to choose an NV card over an AMD one but it does skew the summary a bit and is not really representative of future titles considering Series X is DX12 only.

EDIT: Give it a few years and when these DX11 titles drop off of the review suite some will claim AMD fine wine.
 
They're all fake, arguing that one is better than the other is pretty bewildering.
Not that I am defending AMD, but Nvidia's real prices are much more "fake" and Nvidia has been systematically doing this since 2xxx series. While AMD claims that prices will settle in early 2021. On the other hand Nvidia's 2xxx series never settled and were never sold for the announced prices.
 
We are once again battling GPUs not according to their pure rasterization power[that will diminish over time because of die/price ratio, or become very expensive], but like in the old times- features. In the early days, directx version changed quite often and brought many features, people were buying GPUs that supported new tech. As it is currently -why buy a 1k usd gpu that doesn't have useful features?
5nm GPUs will be very expensive, hence the importance of VRS, DLSS if RT become mainstream.
 
Back
Top