There are seven pages of RTX 3070 Ti analysis and data in this review, but the more I use the card, the more I realize this is massive overkill for the GPU, the gist of which can be summed up in a paragraph. The RTX 4060 Ti is essentially a hybrid RTX 3060 Ti/RTX 3070 with the added benefits of DLSS 3, good media blocks, and extra efficiency, but otherwise it’s a missed opportunity and a disappointment. There are some positives to highlight, but they’re overwhelmed by the feeling that Nvidia isn’t delivering good value for the segment of the market that needs it most.
It’s actually a shame because there yes Positive point of view for discussion. While generation-to-generation gains are variable and VRAM allocation is an issue compared to the RTX 3060 Ti, the benefits of the Ada Lovelace architecture can still be felt. DLSS 3 frame generation’s ability to overcome CPU limitations has obvious value in the mainstream, as you’ll see later.
Perhaps even more impressively, Cyberpunk 2077 RT Overdrive – a big statement about the future of gaming – not only runs fine on the RTX 4060 Ti at 1080p, I actually found that with the RT optimizations we’ve discussed in the past mode, it can actually run slightly faster than the RTX 4070 in DLSS balanced mode at 1440p in its unmodified state. Path tracing Cyberpunk 2077 running in its performance mode with better image quality and faster frame rates than the Series X and PS5 versions is well worth a look, and all from a $400 GPU.
Cyberpunk 2077 RT Overdrive, 1440p, Balanced Upgrade
Please enable JavaScript to use our comparison tool.
But elsewhere, it’s hard to get excited about what Nvidia has delivered here — and the sense of lack of momentum further down the stack we’re in is hard to shake off. Consider this: As a top-of-the-line lineup, the RTX 4090 offers features and performance we haven’t seen before for roughly the same price as its predecessor. RTX 4080? Another great GPU, but at least $300 more expensive than it should be. RTX 4070 and RTX 4070 Ti? These all make sense from a strategic standpoint by offering performance comparable to or better than the RTX 3080, but in each case the products support DLSS 3 and more memory.
Level 70 can make a reasonable argument for them – they’re “solid but unobtrusive” – but it’s much harder to do the same with the RTX 4060 Ti. No extra RAM this time around, unless you want to wait a few months and pay a hefty 25% premium for it. While RT performance was okay (typically in RTX 3070 territory), rasterization performance varied wildly – but even at the best of times that doesn’t represent any game-changing improvements over the existing RTX 3060 Ti. Push this card out of its comfort zone and it can actually be beaten by its predecessor. Yes, the use-case scenarios in which this happens don’t quite match the card, but any kind of generational degradation in performance is a red flag.
Below you’ll see how the RTX 40-series stack has fared in terms of specs so far – the 4060 Ti continues Nvidia’s trend of splitting its desktop graphics cards into a host of GPUs. For example, the RTX 3090 Ti through RTX 3060 Ti are only spread across their GA102 and G104 chips, while the RTX 4090 through 4060 Ti come with AD102, AD104 and AD106 GPUs — the 4060 Ti comes with the latter, with a modest CUDA core count and 128-bit memory bus. On the memory side, Nvidia has spent a lot of effort explaining how the larger L2 cache offsets the narrower bus, and as you’ll see, this works to an extent – although bandwidth-intensive games will suffer, which is We see scenes of why the older 3060 Ti can sometimes beat its successor.
Graphics card 4090 24GB | Graphics card 4080 16GB | RTX 4070 Ti 12GB | RTX 4070 12GB | RTX 4060 8GB | |
---|---|---|---|---|---|
processor | AD102 | AD103 | AD104 | AD106 | AD106 |
transistor | 76.3B | 45.9B | 35.8B | 35.6B | 22.9B |
Die size | 608 square millimeters | 379 square millimeters | 295 square millimeters | 295 square millimeters | 190 square millimeters |
CUDA color | 16384 | 9728 | 7680 | 5888 | 4352 |
boost clock | 2.52GHz | 2.51GHz | 2.61GHz | 2.475GHz | 2.535GHz |
memory interface | 384 bits | 256 bits | 192 bits | 192 bits | 128 bits |
memory bandwidth | 1018GB/sec | 742GB/sec | 557GB/sec | 504GB/sec | 288GB/sec |
Three Gorges Project | 450W | 320W | 285W | 200W | 160W |
power supply recommendation | 850W | 750W | 700W | 650W | 550W |
power cable | 4 x 8-pin | 3 x 8-pin | 2 x 8-pin | 2 x 8-pin | 2 x 8-pin |
basic price | $1499/£1649 | $1199/£1199 | $799/£799 | $599/£589 | $399/£389 |
release date | October 12, 2022 | November 16, 2022 | January 5, 2023 | April 13, 2023 | May 23, 2023 |
In terms of spec comparison with the upcoming RTX 3060 Ti, the CUDA core count is reduced from 4864 to 4352, the 128-bit memory interface is compared to the wider 256-bit equivalent memory interface on the older product (offset by a large L2 cache increase), and the The boost clock is ramped from 1665MHz to 2535MHz.Samsung-made 8nm die shrinks from 293mm2 to only 190 mm2 On the new TSMC 4nm node, the transistor count has increased from 17.4 billion to 22.9 billion. Overall, it’s a fairly modest increase, and while the new process Nvidia is using is expensive, we have to wonder if the chip will cost much less to produce than its predecessor due to its much smaller form factor.
All of this brings us to the big question of VRAM allocation, and a topic of intense discussion between us and Nvidia not only in the product briefing but also in a series of follow-up emails. Let us be very clear about where we stand on this. Developers should do their best to make their console ports work on 8GB GPUs, we shouldn’t see things like Forspoken, Hogwarts Legacy, and The Last of Us Part 1 where the 8GB experience sucks. But at the same time, if you’re buying a GPU in 2023 to match or exceed console visuals, or get AAA games running at optimal textures, you’ll obviously need at least 12GB because we’re finally leaving the intergenerational era behind back.
In short, the RTX 3060 Ti offers faster rasterization and RT performance than a gaming console, and of course the faster RTX 4060 Ti should follow suit – but it needs memory to fully do the job, not just accommodate higher A larger footprint for resolution textures, but also carrying the extra features we should expect from PC games. For example, ray tracing requires a lot of extra memory to store the BVH structures needed to make the feature work properly. Nvidia calls the RTX 4060 Ti a 1080p card – the size of the render target does decrease with lower resolution, but the textures don’t, and neither does the BVH structure. As it stands, I think if the RTX 4060 Ti can run Cyberpunk 2077 RT Overdrive as well as 1440p, I’m not sure if the “1080p card” label best reflects the capabilities of the GPU – but it does help justify the 8GB The rationality of memory allocation.
GeForce RTX 4060 Ti | GeForce RTX 3060 Ti | GeForce RTX 2060 Super | |
---|---|---|---|
Control, 1440p, High RT | 39.8fps/147W – 3.69 joules per frame | 38.7fps/198W – 5.11 joules per frame | 25.8fps/171W – 6.63 joules per frame |
Dying Light 2, 1440p, Ultra RT | 44.4fps/145W – 3.27 joules per frame | 41.6fps/197W – 4.74 joules per frame | 25.6fps/169W – 6.60 joules per frame |
Forza Horizon 5, 1440p, Extreme, RT Closed |
85.4fps/109W – 1.28 joules per frame | 82.9fps/189W – 2.28 joules per frame | 54.6fps/155W – 2.83 joules per frame |
Hitman 3, 1440p, Max, RT Off | 155.7fps/146W – 0.94 joules per frame | 138.4fps/198W – 1.43 joules per frame | 96.6fps/173W – 1.79 joules per frame |
The Ada Lovelace architecture has delivered some excellent power efficiency metrics in previous reviews, and it continues to do so when we look at how the 4060 Ti’s efficiency compares to similar products from the previous two generations. We calculate efficiency by benchmarking games fully unlocked at all framerates, then dividing power by performance, giving us a value defined by joules per frame. The lower the better.
With Control, power consumption per frame is reduced by 28% compared to the RTX 3060 Ti, rising to 44% compared to the RTX 2060 Super – a good deal for a game where the generational boost in performance isn’t astonishing. the beginning. Dying Light 2 has a similar performance, with a 31% reduction in power consumption per frame compared to the previous generation RTX 3060 Ti, and a 50.5% reduction in power consumption per frame compared to the RTX 2060 Super. These two headings cover our RT tests.
Turning to rasterization, Forza Horizon 5 performed very efficiently on all Ada cards, and in all cases, it never hit the power limit of the GPU – something that wasn’t possible with previous cards. The same thing happened here, with equally astonishing results. Power consumption per frame is reduced by 44% compared to RTX 3060 Ti and increased by 55% compared to RTX 2060 Super. When Hitman is running at maximum settings without RT, it consumes 34% less power per frame compared to its 4060 Ti predecessor, and when we compare it to the RT 2060 Super, it uses 34% less power per frame 47%.
That’s one of the biggest pieces of good news for this GPU, but it’s time to turn to the raw performance numbers, with all cards using a stock Core i9 13900K setup paired with 6000MHz DDR5 from the GSkill.
Nvidia GeForce RTX 4060 Ti Analysis
- Introduction, Hardware and Power Analysis [This Page]
- RT Benchmarks: Dying Light 2, Cyberpunk 2077, Control
- RT Benchmarks: Hitman 3, Metro Exodus Enhanced, F1 22
- RT/DLSS/FSR2/DLSS3 benchmarks: Cyberpunk 2077, Dying Light 2, Forza Horizon 5
- Gaming Benchmarks: Control, Cyberpunk 2077, F1 22, Forza Horizon 5
- Game benchmarks: Hitman 3, A Plague Tale: Requiem, Returnal
- Nvidia GeForce RTX 4060 Ti: The 8GB Question and the Digital Foundry Verdict
iGamesNews