And before you start quoting "future-proof", lets see how 4070 does at the resolution it isn't even designed for, 4K (worst case scenario).
Wow, still better than most of those 16 GB cards that have so much "future proofing".
View attachment 303386
Now lets take RT into account (the standard that all new engines are using moving forwards, if you want to talk about "future proofing"). Oooof.
View attachment 303387
But go ahead, recommend the 6700 XT seriously, in supposed sincere good faith, the card that gets half the performance of the 4070.
Honestly.
There is more than your short sighted usage, upgrade path and/or spending habit in life. There is no comparison for the sweet spot RDNA2 is in right now, in raw perf/dollar. And its 12GB is much better matched to its core power. The 4070 has way more core oomph than its 12GB can carry, its a waste, and a shame. The 4070ti is worse. Both cards suffer from a major bandwidth deficit too. This will make itself known, except not for the coming two years. We have seen and had this discussion many times before. If your cards last longer than a couple of years, you will see them fall short with Nvidia's new approach to VRAM capacities. Especially in the segment at and above x70, for the lower ranges it doesn't matter much because you're already compromising straight out of the box. But with an x70 you really shouldn't have to - Pascal's x70 and x80 were perfectly balanced like that. They just ran out of everything all at once; 8GB was insufficient playing TW Warhammer 3 on my 1080 at 1440p for example. The game stutters, but at the same time, the GPU also just didn't want to push for more than 40 FPS, which is edge case playable. That's where the balance should be at.
Ada except for the 4080 and 4090 is nowhere near it not even ballpark.
Perf/dollar:
The simple gist of it all is, that Ada is just a shitty stack priced too high, even if an individual GPU scores nicely today.
Yeah, well not in TPU testing.
YouTubers hype up "problems" to get clicks, I'm not the only one seeing that either.
The fact that so many people are laser focused on the "VRAM problem" when the "RT problem" seems to be much more relevant moving forwards, is amazing to me. But then one of those "problems" allows people to favour the underdog, so...
We are in full agreement on the clicks issue. But that doesn't invalidate everything that comes out of there.
The RT problem is ALSO a VRAM problem
The real million dollar question is why Nvidia is so tight on VRAM when they are the ones pulling RT forward. We know the real answer. They want you to buy their Ada successor 2,5 years from now. The entire strategy is gearing up for gen-to-gen, minimal perf win upgrades. But Nvidia is there to help you, with MORE proprietary technologies like DLSS4, to reduce your VRAM usage. And at that point, you're stuck like a heroin addict on free frames.
Its not for me, that commercial clusterfuck, but you do you.