Review: Sapphire Radeon RX 6800 NITRO+

Why did you stop disassembling video cards? And why does the card have a temperature of 72+ degrees, while the reference has 64. I think this is a failure.

 

stupid me.
I thought this is XT
but it's 6800 non xt with XT msrp

 

never really understood why if you can buy a 500$ card you can't a 600$ just wait a few months until you have the money
this hardware crysis makes it even more dumb who'd want this expensive inferior 6800 rather than a XT ?
it's pretty obvious not only at amd but nvidia also that aib release extremely cheap reskins of the reference design this year, which they can only get away with because some people can't wait and buy whatever they will find, personnally no way I'm buying a card like this
I'm checking prices and availabilty every day and I hesitated quite a few times on no less than a 3090 when I saw one in stock, but when I saw for 2140USD or 1700€ that it only had 2x8 pins...no I'm not spending 2k on a low quality gpu (and thus) power limited card, weird how they all seem power limited this year

p.s.
on my fav webshop an msi 6800 is 755USD and 6800XT 822USD see what I mean ? or 616€ and 671€ if you can't afford 55€ more for this you're in the wrong price range for your purse
(none are in stock of course)

 

there has to be a limit somewhere for each person.
6800 is not great value compared to 6800xt and to 3070 especially.

 

Nice cooling but bad (average) silicon lottery :-(

 

Man, this is the 6800 i want.
Sucks I just bought a $1000 monitor lol

 

First monitor then gpu. I hope its ultrawide btw.

 

monitor should always come first

I recommend checking out pcgh if you're looking for 21:9
about the only tech source that tests 3440x1440 as part of the procedure

https://www.pcgameshardware.de/Rade...er-RTX-3090-Grafikkarten-Vergleich-1362845/3/

it is a demanding resolution for new games, 6800 can't maintain 60 fps in as many as 9 titles out of 20 while at 1440p only in 2 out of 20.
a VRR monitor is really a necessity for 21:9 qhd

 

Hilbert why You didn't put reference card on thermals graphs and noise graphs?.
I have reference Sapphire and it seems i have better thermals? How is that possible.

 

another fantastic, detailed review.

a thing to remember about "paper" launches - it's standard.
and not because some nitwit decided to irritate enthusiasts but because of old print media. while old print media is deader than the proverbial doornail they had to have the object of review well in advance of availability because the object had to be tested, the article written, proofread, edited, printed, and then delivered to brick and mortar stores or subscribers.

and yeah the 6800 (vanilla) is not the value of the 6800xt, but then the value of the 3060 isn't the same as the value of a 3070 either and the 3090 gets slapped every time someone utters the word "value".

 

That´s no healthy Hilbert... Quite frankly, i think you should scale down the size of reviews. Instead of using 15 games test 10 or less. Instead of using 15 or 20 synthetic benchmarks, test half of them and so on.

No point on going overboard if you´re not enjoying yourself and risking your health in the process.

Take care of yourself because the world needs you! And i need your christmas prizes...

 

@Hilbert Hagedoorn
I could be incorrect but think the labeling on some graphs (GPGPU) might be wrong.

 

This is rather disappointing. Hotter, uses more power and more expensive than the reference but performance ins the same or lower. I was looking forward to this model too. Guess I'll stick to my tried and true only buy refence with AMD (though that XFX MERC looks nice).

 

Well it's better than the recent MSI Gaming X Trio 6800XT at least, MSRP of 850 USD / EUR and then what store prices will do to that due to demand and lack of availability.

What it will do is make it easy to choose between that or a 3080 the way I'm seeing it if the custom boards are going to blow right past the 700 estimated pricing range and into the mid 800's.
(The two models are already close to the same price or due to pricing fluctuation almost identical so this feels like a very poor decision to price the cards this high.)

EDIT: AMD's locked little OC limitations are a bit of a thing too but the variants do make a good number on the cooler effectiveness and noise level of said cooling solution at least.

For Sapphire what I'm curious to see is how the Pulse will perform when actually available and what reviews will do for that, they're all kinda the same performance wise for the 6000's and 6800XT's and 6900XT's due to the cap and where they start encountering stability so without allowing additional wattage or voltage it's mostly just the cooler and some PCB tweaks potentially but not with a steep increase in pricing over the defaults.

Even if AMD was a tad aggressive with the reference product pricing which probably doesn't help margins for the board partners but it is what it is.
(And if pricing is increased and due to that closes in on the Ampere 3080 in pricing then it's going to be tougher to recommend the 6800XT over that.)

 

I doubt Hilbert would want to reduce the quality of GPU reviews as game benches are the most important part of them. But if anything can be cut down, it might be the quantity of reviews. Just this month H did 7 3060Ti reviews! Thats a lot of work for basically the same card. Doubt any other review site on the web would invest that much time and effort on one card.

https://www.guru3d.com/articles-categories/videocards.html

 

Well, in my opinion this is the best choice if you don't want the performance of the XT or 3080 while having a ~15% extra performance compared to the 3070, want to get into 4k gaming and don't want to worry about 8GB of VRAM.

 

where'd you get that ?

6.3 % on TPU

Spoiler: tpu

3% on computerbase

Spoiler: cptbs

7% on ppc

Spoiler: ppc

2% on pcgh

Spoiler: pcgh

4.5% on avg. across those 4 sites

edit:sorry,I missed the 4K point.
I checked at 4K it's 7.1%, 6-8% on the sites I mentioned
I still think both are too weak for 4K and held back by bandwidth

 

10C hotter than reference , and gets recommended ? Any explanation ?