TECH TALK Nvidia RTX 4070 Ti Incoming
WHEN NVIDIA OFFICIALLY announced its Ada Lovelace architecture, it revealed plans for three graphics cards: RTX 4090090 as the king of the hill at $1,599; RTX 4080 16GB as the step-down princece for $1199;1,199; and the RTX 4080 12GB at $899. To say people weren’t pleased with the pricing and specs would be an understatement. Nvidia “unlaunched” the 4080 12GB, stating that “having two GPUs with the 4080 designation is confusing.”
To be clear, we weren’t offended by the name so much as other aspects of the card. The problem was in the specifications and potential performance, as well as the large jump in generational pricing. Nvidia’s own numbers show the 4080 16GB delivering 30 percent higher performance than the 12GB variant. That’s not a minor difference in performance; it’s a gaping chasm.
Let’s put things in perspective. By our own figures, the RTX 3080 12GB outperforms the original RTX 3080 10GB model by just 10 percent on average at 4K ultra—less at lower settings and resolutions. A 30 percent gap is what we’re used to seeing between different GPU families, like the 3080 Ti versus the 3070 Ti, where there’s a 30–45 percent gap. By that token, renaming the 4080 12GB as the 4070 Ti solves part of the problem.
What we don’t know yet is how much Nvidia plans to charge for the 4070 Ti. The 4090 launched at $100 more than the 3090, but the 4080 matches the 3080 Ti MSRP and sits $500 above the 3080. We’d like to think Nvidia will be sensible and launch the 4070 Ti at the same price ($599) as the outgoing 3070 Ti, but that’s probably misplaced hope. Even $699 would feel quite onerous for a formerly midrange model number, and Nvidia might even go higher than that—people will be rightly outraged if
By renaming the 4080 12GB as the 4070 Ti, Nvidia has only solved part of the problem.
Nvidia sticks to the original 4080 12GB price of $899.
We’ll find out next month exactly where the 4070 Ti lands, in price as well as performance. Nvidia has been pushing DLSS 3 performance data for the 40-series, but the more we play with it, the less enamored we become. Artifacting is only part of the problem; Frame Generation on its own simply doesn’t make a game feel more responsive and can actually make things worse.
PortalR TX is the perfect example of this. At native 4K, even the mighty RTX 4090 only manages an unplayable 20fps. Turning on DLSS 3’s Frame Generation bumps that up to 37fps, but the input still feels like 20fps (because it is). It’s not that Frame Generation makes no difference at all, but it’s mostly about smoothing out visuals at already reasonable framerates.
The RTX 4080 at 4K with DLSS Performance mode runs at 42fps, which is playable but not smooth. With Frame Generation, it gets 70fps. The input from your keyboard and mouse is still running at 42fps, but the screen updates more frequently and looks smoother, which I suppose ultimately feels a bit smoother. Just don’t expect Frame Generation to feel great if you’re not already getting well above 30fps.
One parting shot for Nvidia is that the melting 16-pin adapter issue has apparently been ‘solved’. Nvidia released an official statement, over a month after the launch, saying, “Our findings suggest a common issue is that connectors are not fully plugged into the graphics card. Nvidia is committed to ensuring an expedited RMA process, regardless of cable or card used.”
There were at least 50 reported cases of melting adapters, and it’s possible that the extra four sense pins on the 16-pin connector made it more difficult to fully connect the cable, resulting in this problem. Certainly, we didn’t hear about any 30-series cards with 12-pin or 16pin adapters melting. The bottom line is, make sure that you plug your extremely expensive graphics card in properly.