The 4070 Ti is pushed hard out of the box compared to other cards, when you apply a sensible power limit it becomes one of the most efficient cards with very little performance loss. Even if you don't limit the power manually, an in-game frame rate limiter largely achieves the same thing which most people will be using to prevent screen tearing on a VRR display.
The RTX 4070 Ti is a much maligned GPU by reviewers. However, after spending time with this GPU and understanding its power and efficiency, at the right pric...
www.youtube.com
Also it's not just about gaming power efficiency, my PC spends more time playing videos than gaming and just look at the atrocious power consumption on the 7900 XTX.
You can do the same with AMD cards too - their driver suite has had things like Radeon Chill for years from the Polaris days. It's been baked into their drivers for years:
A comprehensive review of the new AMD RX 500 series graphics cards.0:00 The 500 Series Cards5:10 Test Setup and Benchmarks8:52 Undervolting the RX 58012:19 W...
www.youtube.com
Nvidia basically copied it. That TPU review is still using the launch press driver for the RX7900XTX figures(22.40.00.57 Press Driver according to the test setup page). AMD already started fixing the idle power,and made Radeon Chill work better,a few weeks after launch:
AMD RDNA3 drivers with improvements for power consumption The updated GPU drivers for Radeon GPUs shows power related fixes. AMD RX 7900 series suffer from power efficiency issues in idle mode and while playing video. Energy efficiency of Radeon RX 7900 XT and RX 7900 XTX were disappointing from...
videocardz.com
Around 15W~25W worse than the RTX4070TI. Most likely to get improved over the next few months.
Also,as a person who has only built SFF systems since 2005,I have told people plenty of times,you need to look at the other parts in your system too(not just the CPU or GPU). For example:
1.)Motherboards can apply different voltage profiles,even at stock.
2.)Motherboards can also have inefficient VRM setups.
3.)A lot of the 80+ certifications don't cover lower loads,ie,under 20% effiency. So you really need to check measured curves to see you don't get caught out.
4.)Monitors can wary in power consumption! It can be as little as 10W to even more. It's something reviewers don't really measure.
So even if you just look at the CPU and GPU idle figures,not properly looking at other metric can easily add dozens of watts to idle and low load power consumption.
Nvidia inventory is increasing:
Nvidia's Q4 revenue decreased by 21%, significantly impacted by its gaming market platform which saw a YoY decrease in revenues. Read my thoughts on NVDA stock here.
seekingalpha.com
That can only be down to newer generation cards also staying unsold. The last time this happened,IIRC,with Turing Nvidia made a massive rejig of the whole range.
They basically relaunched the range by pushing the dGPUs one tier lower. This is why the RTX2070 which had a TU106 chip was replaced by the RTX2070 Super with a TU104 chip.
The RTX2060 6GB was then replaced by the RTX2060 Super 8GB,which was essentially a cut down RTX2070 8GB. So what I can see happening is that Nvidia pushes tiers down another level.
So the AD104 RTX4070TI 12GB gets replaced by an AD103 RTX4070 Super 16GB,which is a cut down RTX4080 16GB. The AD106 RTX4060TI 8GB gets replaced by an AD104 RTX4060 Super 12GB,which is a cut down RTX4070TI 12GB and so on. Also an AD102 RTX4080TI replaces the AD103 RTX4080. Might stay at 16GB at that tier though.
Personally both companies can do with a price cut and re-organisation of their ranges. They are basically chancing this current generation to see what they can get away with.