A few weeks ago, Nvidia showed a mysterious advance of something called “SUPER”, and although it has not materialized in anything official, it is clear that the company wanted to take some prominence from AMD ahead of the E3 fair. Now, it seems that what they showed could end up being a complete revision of the company’s Turing line of cards. Nvidia would like to improve the performance of their graphics by installing the most powerful chips in the – until then – less powerful graphics; without increasing the price.
This would imply that the TU106 chip, which is currently housed in the RTX 2070, would be installed in the RTX 2060 SUPER (with an apparent count of 2176 CUDA cores). The TU104 chip, which is part of the RTX 2080, would do the same in the RTX 2070 SUPER with 2560 CUDA cores and so on. With this we would end up having an RTX 2080 Ti SUPER much more powerful than the current one, and that would arrive later. It is said that this graph would have an unlocked chip that could consume up to 300 W.
With this, Nvidia would completely reorganize its linear Turing bringing more power to the graphics of the lower range. According to several industry sources, the green company could make the official announcement next week. If so, AMD would be in a bind: the RX 5700 XT that has presented this week will cost $ 449 and seems to have a performance only slightly higher than the RTX 2070. If you get that SUPER version, the graph of the network would not have much what to do, given the best performance-price balance, in addition to the raytracing functionalities.
On the other hand, the TU104 chip of the RTX 2080 has an area of 545 mm2, while the RX 5700 XT achieves a lot with only 251 mm2. With this it is clear how cheap it is for AMD to manufacture the 7 nm chips compared to what Nvidia does. What does that imply? That red has at its disposal a very convenient price reduction to need it.