NVIDIA RTX 2080 and RTX 2070 graphics cards will no longer have two different silicons for each, they will now only have one high-performance silicon.
Last summer NVIDIA launched the GeForce RTX, graphics cards designed for high-end gaming. Some charts that implement Tensor Cores and RT Cores to support RayTracing and DLSS. New technologies designed for the high-performance gaming market. The company released two different silicons, one for the Founders Edition models and the other for the company's partners. Specifically the RTX 2080 and RTX 2070 had two variants of silicon. The RTX 2080, for example, has the TU104-400-A1 and TU104-400A-A1 silicon.
There will no longer be different silicon variants for the NVIDIA RTX
The two versions of this silicon had different characteristics, depending on their possible performance. Version 'A' was a high-performance silicon that supported overclocking without any problems. On the other hand, version not 'A' was a simpler silicon and could not be overclocked from the factory. The latter silicon supported user overclocking, but did not perform as well in this field. A time that seems to be over.
It will be from May when the RTX 2080 will only be based on the TU104-400 silicon and the RTX 2070 will only be based on the TU106-400. This data is interesting, since there will be no discrimination between the different graphs, at least at the silicon level. We can say that the Turing manufacturing process has been quite perfected. The limitations regarding overclocking will come from the PCB and the number of VRM phases.
Now the worst silicon for the simplest GPUs will be dropped. All will receive the same silicon, although there is no data on whether this silicon will receive improvements. Normally, it would be a little better, not in net performance, but in overclocking capacity.