If we look at Steam stats, we will see that the most used graphics cards for gaming are the GTX 1060, RTX 2060, and RTX 3060, both in their laptop and mainstream graphics card versions. It is therefore easy to guess that for many users, RTX4060 it is the most anticipated graphics card, however, it would have undergone a last minute change and not for the better.
One of the things that has been bugging us is how NVIDIA compares its new RTX 40 with the RTX 30, the new generation of cards are based on units with fewer cores and units than the previous one, but it is offset by higher clock speeds. The problem comes when DLSS 3 appears in the equation of certain games and its interpolation of frames, which completely distorts comparative reality. However, in the lower ranges, this advantage will be lost and everything points to a boycott movement towards the mid-range.
NVIDIA cuts the benefits of its RTX 4060
If you are a user of an RTX 3050 or an RTX 3060, know that NVIDIA hates you with all its soul, because you are not the premium user they want, because these models offer much lower margins to those of other high-end ones. . . If we add to this that the cost of manufacturing the chips has increased over time, then in the end it is time to make cuts on models like the future RTX 4060.
And what does this extract consist of? As you read a few weeks ago, the RTX 4060 Ti will be based on the full version of the AD106 chip, within the Lovelace architecture. It was therefore logical to think that the RTX 4060 would use the same chip but with fewer active cores. However, NVIDIA made a last minute decision which is to keep the same PCB, but to use another chip of the same generation, but less powerful, more precisely the AD107, in this case the complete configuration.
So the NVIDIA trickster game would be based on turning the RTX 4050 Ti into the RTX 4060, the same way they did the same with the RTX 4070 Ti in the 12GB 4080, but trying to make it work this time .
These would be your specs
In the absence of an official announcement, we have that what has been leaked is as follows:
- 3072 units in a 32-bit float, which means 24 cores or 24 SMs.
- 24 MB of L2 cache.
- 128-bit GDDR6 bus at 18 Gbps
- 8 GB standard GDDR6 memory. Which is a considerable reduction from the 12 GB of the RTX 3060.
- 115W TDP.
We have to assume that the RTX 3060 has 28 cores or SM, so we have to expect a higher clock speed to compensate and overcome the difference. Its consumption of 115 W suggests to us that it will have good overclocking capabilities, especially since NVIDIA has used the 12VHPWR connector again, this time in 150 W mode. So there is room to increase its MHz. In another vein, we wonder if we will have shorter versions created for Mini-ITX cases, especially since this form factor is losing more and more popularity.
