NVIDIA GeForce RTX 4060 Ti Shows Up in Geekbench Database

[HD64G]

8GB for $400 and 16GB for $500 are the latest rumors.

 

[RegaeRevaeb]

He never mentioned Nvidia. His statement is absolutely right, no matter what Nvidia does, their fans think.
Inferring he is a Nvidia fanboy just because he speaks about AMD, is itself a statement of true fanboyism

Of course absenting Team Green, and its similar fanboys/girls with same-vein critique, may also be considered an endorsement of the wink wink, nudge nudge type. It's like when political ads talk about bad things by bad people without explicitly naming the intended target. What

Anyway, I'm ultimately getting at is be kind to all sides of the issue(s), people. It's hard to enjoy the forums at times with the noise from either side.

 

[Pumper]

8GB for $400 and 16GB for $500 are the latest rumors.

Hard to believe that it would launch at same MSRP as 3060Ti, when all the other replacements received price hikes.

 

[ixi]

Ohh my, what a junk. No, but really. If its 9% faster... does it mean Nvidia will gimic performance on 3060 ti with drivers to get better comparison?

 

[neatfeatguy]

Hard to believe that it would launch at same MSRP as 3060Ti, when all the other replacements received price hikes.

Without knowing exactly how the card will do, just speculation right now, but if it's not really any faster than the 3060Ti it's not really a different card. If it can't handle 1440p at the same level as a 3070, then this 4060Ti is wasted. All previous cards show decent performance gains over their previous counterpart.

4070 roughly a 3080 (3080 is roughly 20% faster than the 3070)
4070Ti roughly a 3090Ti (3090Ti is roughly 25% faster than the 3070Ti)
4080 is roughly 12% faster than a 3090Ti (the 3090Ti, which in turn, is roughly 15% faster than a 3080)
(all based off rasterization and 1440p benchmark info from TPU)

Basically each card has seen at least a 20% uplift in performance over it's counter part in the 30xx. A 3060Ti is only about 10% behind a 3070 and 15% behind the 3070Ti. If the 4060Ti can't at least give us 3070Ti performance levels, then it's a failure on that aspect (not to mention the stagnant 8GB VRAM). Of course this is just performance information - if you add in cost differences it makes these Ada cards look worse simply based on performance value vs Ampere (based on MSRP, not the hyper-inflated prices).

I'd venture to say the 8GB model will be about 8-10% faster over a 3060Ti and the 16GB will be around that 15% faster.

But, we wait and see what all the hubbub is about.

 

[DemonicRyzen666]

It quite literally is, optical flow acceleration has been present since Turing. They claim Ada's is better, sure I believe them, but they're enabling DLSS 3 on hardware like AD107 (4050 mobile) while leaving the 3090 out, I refuse to believe that 4050 mobile is faster than the 3090 at anything including this

it's not faster, it's just more accurate according to nvidia.

 

[Dr. Dro]

it's not faster, it's just more accurate according to nvidia.

Accuracy can be improved by spending extra cycles in processing. It seems obvious to me that they're using DLSS 3 as an integral part of their performance uplift claims, both to mask the poor to nonexistent generational improvement with the configurations they've decided to ship and to induce fomo

 

[HD64G]

Hard to believe that it would launch at same MSRP as 3060Ti, when all the other replacements received price hikes.

Apart from 4090, the market didn't recieve well the pricing of this gen this far. They are forced to change their pricing strategy, especially for the low-to-mid tier GPUs. This part of the market seeks for bargains, not the 4K performance.

 

[Dr. Dro]

Yeah, AMD kicked the CPU market back into gear (thank god, finally), but GPUs are completely stagnant. Nvidia is still completely uncontested at the high end, as they have been for what, 7 years now? And AMD just can't seem to do for GPUs what they did for CPUs.

And Covid pricing gave everyone the excuse to cram 4-5 years worth of price inflation into the space of 12 months. It's never going back down, and will only continue to rise until there's some kind of catastrophic real-world political event (unlikely) or AMD gets its ass in gear (only moderately likely). Until then, we get to enjoy Nvidia mislabeling each successive generation with model numbers (and prices) one performance level too high, increasing more with each generation.

Despite driver bugs, RDNA 2 really was competitive with Ampere high end. Otherwise, the last time AMD was on equal footing (or ahead really) was in the Kepler days. With Ada, the bloodbath is back with a vengeance and it's only not worse because the Nvidia product stack is terrible and almost all shipping configurations of Ada have extreme unexplored potential.

The 4070 Ti is a third tier processor being sold as a high end product, the 4080 is realistically a non Ti 4070 and by simply enabling the disabled units and cache slices on 4090 they can ship a GPU that's 30%+ faster than it overnight.

upcoming AMD GPU, the RX 7600, coming in at €349

DOA at this price. It needs to be substantially faster than the 6650 XT, but even then it's still a 8 GB, 128bit low end card.

There's a sliver of hope that N33 overperforms in its weight class by regressing to a monolithic design (and doing away with the chiplet interconnect which is potentially the source of 7900 XTX's underwhelming performance), as I believe in the RDNA 3 design and genuinely think it's a high quality architecture, but the competition will be fierce, shame to the consumer that it's a fierce battle of overpriced weaklings.

 

[JimmyDoogs]

I'm not quite sure DLSS 3 is just simply just a flip of a switch. AMD doesn't have this in any iteration of their cards. I'm fully willing to be wrong on this one and I would love to see the addition of FG in a non 40 series card but there's been 0 leaked information pointing to this happening at the moment.