Monday, February 13th 2023
NVIDIA GeForce RTX 4060 Possibly Maxes Out AD107, NVIDIA's Smallest Ada Silicon
NVIDIA's mid-range, high-volume graphics card based on the GeForce "Ada" graphics architecture, the GeForce RTX 4060, a 60-class product, could feature specs that make it possible for NVIDIA to carve it out either with a maxed out 5 nm AD107 silicon, or a heavily cut-down AD106. Kopite7kimi, a reliable source with NVIDIA leaks, says that the RTX 4060 has specs which align with the full-spec AD107, with 3,072 CUDA cores across 24 streaming multiprocessors (SM), 96 Tensor cores, 96 TMUs, an unknown ROP count, and 8 GB of 18 Gbps GDDR6 memory. The memory bus width is hard to predict with this generation. The GPU's on-die L2 cache is 24 MB in size. The card has a typical graphics power (TGP) of 115 W, making it possible to build cards with just one 6-pin PCIe power connector.
Source:
kopite7kimi (Twitter)
95 Comments on NVIDIA GeForce RTX 4060 Possibly Maxes Out AD107, NVIDIA's Smallest Ada Silicon
www.techpowerup.com/forums/threads/nvidia-geforce-rtx-4060-possibly-maxes-out-ad107-nvidias-smallest-ada-silicon.304757/post-4951812
So the current plan is:
2022, RTX 4080 - $1200
2024, RTX 5080 - $2040
2026, RTX 6080 - $3468
2028, RTX 7080 - $5896
2030, RTX 8080 - $10022
2032, RTX 9080 - $17038
2034, RTX 1080 - $28965
RTX 4050 8GB, 128 bit, $249 w/ performance of 2060+10%
RTX 4060 8GB, 128 bit, $349 w/ performance of 3060+15%
RTX 4060 Ti 8GB/10GB, 128/160 bit, $399 w/ performance of 3070
RTX 4070 12GB, 192 bit, $599 w/ performance of 3080 12GB
For me, 4050 and 4070 are best of the 40 series in terms of VRAM, price, and performance. 4060, 4060 Ti, and 4080 are worst values. Maybe, if 4060 Ti has 10GB, it would be as good as 4070 in terms of price-performance.
Given that the rest of the Ada models have received core count increases, this is a significant cut down from the 3584 in the old 3060.
If it's priced closer to $300 perhaps the spec reduction compared to Ampere can be forgiven, but that's unlikely given how greedy Nvidia have been with pricing the last few years....
See, outside of Snowflake land, people don't say 'attacked', they just say there was a discussion. Try it! Very refreshing. Well to be fair, if both AMD and Nvidia persist and Intel remains trailing them, that IS where we will end up - price / perf won't move, so games won't move, consoles will catch up (they already sorta have, because ADA and RDNA3 have made further moves pretty much impossibru) and it'll all be one nice stagnant playground.
But that doesn't have to be a bad thing either, really does it. It just means that RT will remain a unicorn nobody wants to look for in earnest, the rest of gaming is in a pretty decent place. Diminishing returns happened already for over a decade, graphics are hardly the most important part of games by now - the baseline is more than good enough. Note... I've been predicting RT will probably meet a dead end as hardware price will inflate because of it...
A 3072-core part at about 2.7GHz is basically tied with a 3060Ti.
It's not quite that simple, it will be ROP-limited at higher resolutions and have less bandwidth than a 3060 but with much more cache, heavy RT titles will probably perform better on a 4060 than a 3060Ti, while raster-heavy titles leaning on ROPs and bandwidth will run better on the 3060Ti.
This is me guessing but it's an educated guess, extrapolating existing data and the pattern holds true within about 10% for the three Ada cards reviewed so far. It's not rocket science either as Ada's architecture is very similar to Ampere, just with scaled up cache - most of the gains are the 50% clock increases moving to TSMC over Samsung N8.
The entry-level market is ravenous.
Meanwhile I don't see this selling well, just because there's so much in the channel and moving down the price chart's.
Entry level GPU are not worth the design effort at this point, why because Intel and AMD are going to be putting a useful Igpu in there next CPU so who's buying, just a opinion no hate I await the hate though
@Why_Me "Probably why AMD is cleaning up in the gpu market" , yeah quality posting there, , keep it up, someone might bite yet to your trolling,, meanwhile see my last pm for exactly what I STILL think.
2 year olds bitching about price in every Nvidia thread... yet their 4000 series cards are fast and cheap enough that they are pushing down the price of existing cards by quite a bit. Competition.
Don't know how reliable this is; it looks like the 4060m spec (which will be out very soon). It would be weird if Nvidia used the same silicone for both mobile and desktop, since they aren't for the higher end cards. 115W is the base mobile spec too; they boost to 140W. Even the 4050m boosts to 140W. Surely the desktop will be higher?
4060m: www.techpowerup.com/gpu-specs/geforce-rtx-4060-mobile.c3946
4060: www.techpowerup.com/gpu-specs/geforce-rtx-4060.c3891