Radeon RX 7900XT vs. GeForce RTX 4070 Ti, 50+ Game Benchmark @ 1440p & 4K
Video Statistics and Information
Channel: Hardware Unboxed
Views: 220,211
Rating: undefined out of 5
Keywords: hardware unboxed
Id: lSy9Qy7sw0U
Channel Id: undefined
Length: 21min 12sec (1272 seconds)
Published: Sat Mar 11 2023
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.
I mean, if the cost per frame doesn't go down significantly with each generation we have a clear problem
Anyone got any insight on why AMD is not taking much of a performance hit from hardware lumen? Most other benchmarks show AMD taking a notable performance hit from the hardware mode while Nvidia cards more or less stayed the same.
There are some serious methodological issues. Why would you test Nvidia using FSR? Why would you not use DLSS in games that supported it just because AMD lacks the feature?
Cards should be tested how the consumer would actually use them. Obviously that is a judgment call, but necessary if you want to provide consumers the best info. Nobody ever uses FSR when DLSS is an option. DLSS balanced is closest in quality to FSR UQ. Most users will use DLSS quality whenever it is an option, and nobody doesn't use DLSS because FSR isn't available for a card they don't own.
Recalibrating cost-per-frame against regional pricing instead of MSRP is also weird. Weirder is that you can buy $800 4070tis, there are multiple models. Several more in the $820 range.
TLDR: Basically identical performance at both 4K and 1440p. 4070ti clearly better in raytracing but has only 12GB of VRAM vs 20GB on the 7900XT which might become an issue down the line.
My personal opinion is that releasing a card at $800 with 12GB of VRAM is unnaceptable these days. Many games at 1440p and 4K are already starting to thread that 12GB line especialy with RT and some even go over it.
I disagree on the point where he says he doesn't need to test with DLSS2 because Radeon doesn't support it, so the solution is to test with FSR2. The solution is to test native.
Nvidia forces AMD to lower the price by 100 bucks. I guess it's Jebaited 2: Jebait never dies.
Frankly, NVIDIA's software advantages makes this a ridiculously easy decision.
both of these cards should have been priced 200-300 usd less. they really are midrange cards with absurd prices.
the rx7900xt still sucks though. nvidia offers a lot more features and better software than amd. if you're going to buy one of these cards, amd gives you zero reasons why you should consider their offering.
We've got some progress by Steve since it seems he didn't 'skew' the results this time. Which is scary since if he followed the same format as he did for the RTX 4080 vs 7900 XTX, the RTX 4070 Ti would've gotten beaten pretty handedly. Goes to show how easily you can frame the comparison however you'd like while still using valid data. Considering how many people use this channel for purchasing decisions, itโs annoying that whether or not he feels benevolent can make such an big impact.
For reference, below are the charts for these 2 reviews.
7900 XTX vs RTX 4080
7900 XT vs RTX 4070 Ti
Notable changes, MWII isn't counted twice and there is an inclusion of Cyberpunk. I'm sure there are other changes, but those are the ones that really tip the scales. Had he followed the same format as before, RTX 4070 Ti kind of gets blown out in this comparison.
A few notes on his editorial comments:
12gb is too low for that card