Why? Just look at ampere where there was about 10% difference between the non ti and ti cards.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD RDNA3 unveiling event
- Thread starter CAT-THE-FIFTH
- Start date
Why? Just look at ampere where there was about 10% difference between the non ti and ti cards.
The specs between 3080 and 3090 were quite close. That is not the case for 4080 and 4090, the 4080 is significantly cut down compared to the 4090.
That's a poor comparison. The non ti to ti top GPUs had like 10% difference in core count. The 4090 has 60% more cores than the 4080!
In any game where the 4080 and 4090 are close, it's a failure for the 4090 to reach its potential for whatever reason
Last edited:
The cards still existed despite being close on performance to their previous iterations. And they mostly got panned in reviews for being so close in performance to the non ti variants. The 12 gig 4080 originally had clock speeds higher than the 16 gig variation, so maybe that makes up the deficit in certain games?
Also i love how this is cropping up in the rdna3 thread.
Last edited:
At least it's semi relevant
The rtx4000 thread has music videos and people talking about 3dfx
Lol the 4090 is only ~12% faster than a 6950XT at 1080p as both are CPU bound. Jump up to 4K and it is ~65% faster because the bottleneck switches to the GPU. So cherry picking some CPU bound game or resolution and saying the 4080 is only 10% slower than a 4090 makes you look desperate, because even a 6950 or 3090Ti is “only” 10% slower than a 4090.
So no, the 4090 is not bottlenecked at 4K and the 4080 will be about 35% slower given the massively cut down specs. (and based on Nvidia’s own marketing).
That leaves the 7900XTX plenty of room to be ~ 15-20% faster than a 4080 and 15-20% slower than a 4090 in raster at 4K.
At 4K the 4080 is only 25ish% faster than a 3090Ti.
I will caveat this with “based on the Nvidia and AMD marketing slides”.
I'd wait for proper benchmarks rather than some random guy on a Chinese site, how's he even have drivers when the cards not released yet and still under NDA for another week so not even the press drivers would be available?.
True - and with DLSS your card uses a lot less power and runs cooler. I play at 4k and DLSS is a godsend.
I wouldn't be surprised if Nvidia actually pays game developers to not put AMD FSR in their games.
Isn’t the game count pretty similar between the two
Also note the AMD slides stages it's upto 1.5x-1.7x and not on average. The Nvidia card faces more driver overhead so max fps is impacted as a result
Last edited:
Performance will be a bit all over the board with RDNA3 due to the organization of the ALUs/SPs. 1.5-1.7x is likely an average across AAA titles. There will be games that see a bigger, and others that smaller delta. It's going to be interesting to see direct game-for-game performance comparison, and similarity, how AMD marketing positions these comparisons. Also will be interesting, assuming they stick with this implementation, if software development evolves to optimize here -- if the cart will drag the horse, so to speak -- across generations. My guess is in console-originated games, it will.
Last edited:
Soldato
- Joined
- 21 Jul 2005
- Posts
- 20,701
- Location
- Officially least sunny location -Ronskistats
Yeah I mean statements like that were valid for debunker two years ago. Currently its just misinformation now.
Its seems to be sliding down this trap again. Anti AMD posters drag the RDNA comparison to the 4090, and just want to talk about Ada cards. The two cards being released are at the thousand pound segment, they wont be beating in very many situations but thereabouts. Where they do beat the competition so far is where it matters, on price! This is your option.
Last edited:
They also had charts with UP TO on RDNA2 which ended up being averages
Some in the Nvidia camp really don't want it to be true don't understand because we should all want them to do well it only will help prices in the long run
Likely, and the Xbox X Next will very likely follow suit. It's a prime architectural change for consoles: notable performance uplift for small cost.
With RDNA 2 we could see their confidence as they outright compared the 6900XT with the 3090 in a much wider variety of games. This time it's smokes and mirrors with some numbers showing FSR, some without it etc.
Not sure where the part of us wanting AMD to fail comes from. They need to catch up with Nvidia on the feature set atleast with the halo card
Last edited:
You can mod FSR 2.0 in most games having DLSS. The problem isn't the game count but rather it's worse than DLSS when the game is in motion.
In Cyberpunk for instance, DLSS starts having issues with shimmering on neon signs in the distance. You can see a faint white border around the signs which flickers and disappears the moment you get close to it. FSR has the same flaw except you also see more jaggy edges and more ghosting in motion.