Thursday, January 19th 2023
NVIDIA GeForce RTX 4060 Ti Possible Specs Surface—160 W Power, Debuts AD106 Silicon
NVIDIA's next GeForce RTX 40-series "Ada" graphics card launch is widely expected to be the GeForce RTX 4070 (non-Ti), and as we approach Spring 2023, the company is expected to ramp up to the meat of its new generation, with xx60-segment, beginning with the GeForce RTX 4060 Ti. This new performance-segment SKU debuts the 4 nm "AD106" silicon. A set of leaks by kopite7kimi, a reliable source with NVIDIA leaks, shed light on possible specifications.
The RTX 4060 Ti is based on the AD106 silicon, which is expected to be much smaller than the AD104 powering the RTX 4070 series. The reference board developed at NVIDIA, codenamed PG190, is reportedly tiny, and yet it features the 16-pin ATX 12VHPWR connector. This is probably set for 300 W at its signal pins, and adapters included with graphics cards could convert two 8-pin PCIe into one 300 W 16-pin connector. The RTX 4060 Ti is expected to come with a typical graphics power value of 160 W.
At this point we don't know whether the RTX 4060 Ti maxes out the AD106, but its rumored specs read as follows: 4,352 CUDA cores across 34 streaming multiprocessors (SM), 34 RT cores, 136 Tensor cores, 136 TMUs, and an unknown ROP count. The GPU is expected to feature a 128-bit wide GDDR6/X memory interface, and 8 GB could remain the standard memory size. NVIDIA is expected to use JEDEC-standard 18 Gbps GDDR6 memory, which should yield 288 GB/s of memory bandwidth. It will be very interesting to see how much faster the RTX 4060 Ti is over its predecessor, the RTX 3060 Ti, given that it has barely two-thirds the memory bandwidth. NVIDIA has made several architectural improvements to the memory sub-system with "Ada," and the AD106 is expected to get a large 32 MB L2 cache.
Sources:
kopite7kimi (Twitter), VideoCardz
The RTX 4060 Ti is based on the AD106 silicon, which is expected to be much smaller than the AD104 powering the RTX 4070 series. The reference board developed at NVIDIA, codenamed PG190, is reportedly tiny, and yet it features the 16-pin ATX 12VHPWR connector. This is probably set for 300 W at its signal pins, and adapters included with graphics cards could convert two 8-pin PCIe into one 300 W 16-pin connector. The RTX 4060 Ti is expected to come with a typical graphics power value of 160 W.
164 Comments on NVIDIA GeForce RTX 4060 Ti Possible Specs Surface—160 W Power, Debuts AD106 Silicon
That's what he heard!
Today there is a news that Apple which is the first and exclusive customer on the optical shrink TSMC N3 will begin to ship Apple M2 products as late as late 2023.
3nm M3 Chips To Launch With MacBook Air In 2H23, Will Bring Major Performance And Battery Life Gains (wccftech.com)
This means that AMD will move to N3 in 2025 or 2026.
Goodbye PC business !@ You won't be missed.
In normal times the 3070 Ti (not even that, because the TDP of 3070 Ti is 290W) would've been the top card (I mean that kind of performance, not the name) in the Ampere generation.
I think two things happened.
Firstly NV saw in 2016-17 the first bigger crypto mining craze (they started working on Ampere around that time) and realized people want more calculation powa and more cards.
Secondly since they were in contract with Samsung on a crappy node with high power consumption this new top card with 250W TDP would've been only maximum 10% faster than the previous flagship (2080 Super/Ti) which is a very bad generational uplift, worse than Pascal->Turing.
They were cocky by going with Samsung because AMD was nowhere to be found (in 2016 AMD had RX 480 and Fury X and that's it).
So they went allin not caring about TDP with Ampere so we got these 300W+ cards. They're much faster than Turing but they also eat much more than Turing. Or should I say they're faster only because they eat much more.
Look at the TPU database: the 3060 is barely faster than the 2060/Super or the 2070, the 3060 Ti barely faster than the 2070 Super/2080 and the same is true 1660TI vs 3050. These are same TDP tiers. At least they lowered the prices by moving down a tier these performance levels.
They got lucky with the second big crypto mining boom and the Ampere cards sold out easily despite the high TDPs.
Since people want these cards they gonna make them but I think without crypto mining the history would've completely different with an Ampere generation barely faster than Turing showing NV dominance and monopolistic behavior in the GPU space.
So many other things you can do on a pc besides game.....
Figures that after they've completely milked the masses with their overpriced power sukin crap, they would bring out this low-end thing that probably can't even run the "can it run crysis" utility without stuttering or overheating......
There is no consistency over the last decade with Ti models. I went back so many generations solely to give your argument the benefit of the doubt, because prior to Ampere, there was another xx60 Ti model using a 104 die, the 660 Ti - but that was 13 years ago!
The 4060 Ti isn't using a 104 die
Ampere 3060 Ti did use a 104 die. We've established that - but this ONE generation occurence isn't enough to call it a concrete pattern like you say it should.
Turing 2060 Ti didn't exist at all, and if you say "super=Ti" then it still didn't use a 104 die
Turing 1660 Ti didn't use 114 die
Pascal 1060 Ti didn't exist.
Maxwell 960 Ti didn't exist
Kepler 760 Ti didn't exist
Kepler 660 Ti (finally, we get there) used a 104 die.
Please, explain to me why you're so adamant that the 4060 Ti should be a 104 die again. I honestly don't understand the point you're trying to make.
this time around 4070 is a rebranded 60 Ti, just look at the 4070 ti to 4070, they cut more than 20%, this is precisely what 3060 Ti was compared to a 3070 Ti.
And the 60 Ti falls in the abyss of 128 bit bus.
Many limbo on names, endlessly compering 10 gens back of tiers according to them instead of just going price\pref at a given time.
Things change rapidly, adapt or be left behind. The neverending rant on memory bit, die size\name, product name is futile as dust.
No one disputing the fact that today you pay extra more and getting less for that extra but clinging to what was in the (not far) past and not accepting the economy global changes and NV hunger for more $$$ is somewhat naive.
The best thing you can do is to delay your GPU purchase as much as you can (I advise 5 years, no less) and if you really need\want to buy go for best cost\pref on the day of purchase. All else is just confusing spec info that might diviate you to less than optimal purchase decision.
I really don't think we ever got more performnace in the midrange than we do today. And since you mentioned the gtx 770,got that card almost day one for 370 euros if i remember, day one it was dropping easily to below 60 and even 30 in plenty of games. So sure we might think the prices have gone up, but fact of the matter is todays midrange gives you way more performance in modern games than the high end gave you 10 years ago.
What's next with the RTX 50 series? Higher prices still most likely.
Some people have asked me why I care. I care because I see PC gaming as a hobby and I always have. I see PC gamers as fellow hobbyists and I don't like seeing my fellow hobbyists being abused and ripped off by Nvidia. $550, if not more, for a lower end Ada RTX 4060 Ti which is really an entry level RTX 4050 Ti is ridiculous. Don't be silent and just take it like is being advised by a vocal minority here. Fight back.
But lets blame nvidia regardless
tpucdn.com/review/nvidia-geforce-rtx-2080-ti-founders-edition/images/power_average.pngBut Nvidia's gpus are CHEAPER than amds, wth are you talking about?? What nvidia pricing schemes, dont you see that amd is the one fleecing us? Like seriously what the heck is wrong with people and their nvidia hating
The xt is priced 14% higher across eu while being a worse product than the 4070ti. Lol
Vega was below 300W while the Ampere generation has 5 cards above 300W. AMD made power hungry cards earlier (HD 7970 GHz, R9 290/X, R9 390/X, R9 Fury/X all below 300W) but they had to in order to stay somewhat competitive. In the RDNA2 lineup only the 6950XT (an afterthought card) consumes more than 300W.
NV is the market leader so they should set an example.
You could indeed also get an RDNA2 midrange card to have the VRAM. But the 3060ti is a horrible example of midrange progress, its a card that effectively regressed in how long it'll last in the midrange compared to say a Pascal or Turing alternative with similar VRAM but a much less powerful core. So even in the midrange, on Nvidia, you do have adverse effects besides even price, which result in reduced usable lifespan before you're cutting down settings that noticeably hit IQ.
The bottom line remains: on Nvidia, you are paying the RT price on the entire stack. The last cards avoiding that issue are the 16 series - not exactly midrange by todays standards.
IMHO the midrange has definitely not improved in any way shape or form, past the 1060 6GB. It has regressed. Perf/$ gen to gen barely moves forward or is in fact worse; VRAM relative to core power got fár worse and cache doesn't alleviate that in most cases; and absolute pricing also just went up.
I'm honestly trying to find a nice purchase now in the midrange of last several generations. Its just not there. Every single Nvidia card is handicapped in some way or another at its given price point and stack position. Coming from a 1080, it seems the only cards I can look at are 6800 (-XT) or 3080 10GB, the latter being sub optimal just the same - and a whoppin 320W TDP. And then I'm still looking not at 450-500 (reasonable range imho) but 600+ EUR.
Seriously man, I need a full bucket of cognitive dissonance to see improvement here over the last two generations...
I also dare your GF to run Darktide @ 1440p on a 3060ti. Gonna be a fun sub 60 FPS experience... You're going to have to use FSR/DLSS as well to get decent frames. Please stop selling nonsense. And this isn't anything special either, its just a UE4 game.
Watchdogs 2 was another one, although it didnt look bad, setting shadows to the pcss option means your framerate will drop to the low 30ies. Its a game that came out with your 1080, try it, youll probably struggle to go above 20 at 1080p maxed out.
Some games have stupid max settings (gtav had 8x msaa lol). You cant blame the cards if they cant handle that.
But anyways, dlss exists for a reason, and its absolutely amazing. I know the usual crowd that's going to boo but in fact dlss is as good, sometimes even better than native (cod black ops for example looks way better with dlss q than native). So i dont see the problem with activating it. In fact i have it perma on even on my 4090,cause there is no reason not to.
Its easy to spot this trend here.
You can keep moving the goalposts to imply denial of an obvious trend, you do you. Let's see where a 160W 4060ti goes with 228GB/s... at a supposed 350-400 bucks like you say.