• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

First Alleged AMD Radeon RX 7900-series Benchmarks Leaked

As good as this looks, it's all conjecture until Tuesday.
 
When AMD announced N31 specs, I thought 7900XTX would be -10% from RTX 4090 in TPU's 4K testbed, with the assumption that by making dual-issue the SPs it will bring around only +25% (/clock/SP) performance gain.
Sadly AMD already announced that the performance gain is even less, only +17.4%.
So probably 4K raster in TPU's testbed will end like that:
If RX 7900XTX performance is 100% then:
RTX 4080 is 100% (best case for Nvidia)
RTX 4080 is 95% (preferred scenario)
RTX 4080 is 90% (best case for AMD)
The performance gain for 7900XTX vs 6950XT is actually disappointing compared to resources allocated, but at least is priced right!
 
If the 7900XT trades blows with the 4080 that'll be a huge win for AMD @ $899.
 
Monday 2pm uk time is the review embargo
Oh nice! That means I'll be able to look at the reviews right after I get my COVID booster! Thanks for the clarification! :)

I've reported your account because the only thing you post is anti-AMD rumours and FUD. I don't understand the kind of person who is willing to shill for a company like NVIDIA, but you're clearly one of them.
Yeah, I was wondering what bugs they were referring to. I haven't heard anything about bugs and I've had no stability issues with my RX 6800 XT.
 
Yeah, I was wondering what bugs they were referring to. I haven't heard anything about bugs and I've had no stability issues with my RX 6800 XT.
There may very well be bugs.
 
Last edited by a moderator:
Pre-launch leaks are unreliable. They're leaked for clicks, no other reason. I'll wait for the real reviews.
 
you people have to realize that they know exactly if these numbers are fake or not, because, you know, they have the cards
there is no point posting this article if it's fake
 
As good as this looks, it's all conjecture until Tuesday.
most likely how things will end up. reviewers received their media packages last week and monday morning (US EST time) is when the embargo lifts. I expect most reputable media outlets by now to have gone through a big portion of the testing and ofc some leaks show up since i suppose not everyone follows nda rules lol.
 
you people have to realize that they know exactly if these numbers are fake or not, because, you know, they have the cards
there is no point posting this article if it's fake
Yep ^^
 
Looks good for AMD if true!

BTW, how'd they get a 4080 to run 80C? My FE card struggles to reach 60C, lol
 
How can the cpu temp is around 10° lower with the 4080 ?
Should it not be the same ?

There are all sorts of strange outliers in these 'benchmarks'. Completely untrustworthy. Ballpark correct, possibly, but nothing to base any conclusions on
 
As most predicted, 4080 will lose badly in raster from XTX and will be matched or even lose slightly from XT. Let's hope that will initiate a price war that will end up well for us consumers.

It strangely looks as though the 4080 is much more efficient with racing styled games. Perhaps related to some underlying technology / frame-to-frame strategies, and it is a stark contrast to this one where it seems to lose hard. Although it also looks like a broken run for that card, with 1%/0,1% lows there. Clearly a more erratic/variable benchmark seeing the other lows as well, but its also clear AMD handles it better.

The CPU hit also seems to be higher on Nvidia all over the place. I'm gonna look out for that one in real tests... going to make the 7900XTX vs 4090 perf interesting because it would be limited much harder.
haha misread the gpu temp for cpu :D Its an opposite situation, in fact!

1670699924509.png
,
 
Last edited:
How the hell do you make a news from some fake video?
That yt channel had 4090/4080 vids like two weeks before they were out...
It's obviously fake
Except that the numbers were correct for those 4080/4090 benchmarks, so...
 
Looks good for AMD if true!

BTW, how'd they get a 4080 to run 80C? My FE card struggles to reach 60C, lol
Try taking it out of the fridge. :roll:
 
you people have to realize that they know exactly if these numbers are fake or not, because, you know, they have the cards
there is no point posting this article if it's fake
yeah seems like a very cheeky way of saying "this is what the xtx/xt perform like" without explicitly stating that the results in the video aren't fake - clever of TPU honestly lol
 
if those numbers are true, it is embarrassing for Nvidia. They have to lower 4080 prices quite a lot...
 
Let's be honest, even if thr 7900XTX is faster AND cheaper than the 4080, everyone will still buy the 4080, like how it always goes. Its weird, everywhere on enthusiast sites, everyone always claims that they just buy "whatever offers rhe best performance at the price", but mysteriously, the sales figures never seem to demonstrate that, even when AMD puts forth a card that can be proven to be empirically better than its Nvidia competitor, so what's happening?

I'm just frustrated with the GPU market. Especially how everyone complains online how they want competition, but then never use their purchases to make that a reality. If you refer to the literature on consumer psychology, it all demonstrates that consumers DON'T make rational choices, that they're made on emotions and a weird desire to build a parasocial relationship with a brand.....it just seems like there's LITERALLY, no product that AMD can ever make, regardless of how much better it is, that will ever translate to an increase in marketshare. If I had the time and resources, I'd love to do some research based on personal interviews with GPU customers to determine the ACTUAL reasons behind their choice.

Look at one of the very first replies to this article, a user references some weird "hardware bug" with RDNA3 and uses that to completely dismiss the whole product line....first, the product isn't even released yet, how could they possibly have access to legitimate and accurate information demonstrating a "hardware bug" already? Secondly, it seems like this person has already made their mind up on buying Nvidia in perpetuity and this "hardware bug" is merely a pretext to create the air of rationality around their choice.
 
if those numbers are true, it is embarrassing for Nvidia. They have to lower 4080 prices quite a lot...
I hope they lower it to $1,100, then I'll be able to get the 4080 for $100 more than the 7900 XTX!
suck it, AMD fanboys! I'm paying more to get less! can't wait to play portal rtx with 20 fps...
 
If I had the time and resources, I'd love to do some research based on personal interviews with GPU customers to determine the ACTUAL reasons behind their choice.
here I'll give you the most common "answers"
"amd drivers are icky"
"nvidia just works"
"I mostly play games, but nvidia does 3d better, and I plan to do 3d one day... maybe. I'll get back to you on that in 5 more gpu generations"
"nvidia cards just don't have the same problems as amd"
"nvenc is better than amd's encoder, I don't stream tho"
"you pay more for premium - nvidia is premium. I love premium. I love luxury products."
"nvidia looks better in my pc case"

in 99% of cases when someone gives you a reason other than performance as to why they choose nvidia over AMD - it's a bullshit cope, people are just unwilling to try alternatives, afraid of the unknown/dark and so on
the only reasons why I chose the 6800 xt over the 3080 is because it has 16gb of vram (so I believe it'll be more futureproof because of that alone, there are examples of the 3080's 10 gigs of vram being inadequate in modern games, in nvidia's own benchmark games ironically...), draws less power and overclocks a little better
 
I hope they lower it to $1,100, then I'll be able to get the 4080 for $100 more than the 7900 XTX!
suck it, AMD fanboys! I'm paying more to get less! can't wait to play portal rtx with 20 fps...
RTX 4080 should be sold at $899...
 
Back
Top