Most likely based on the TU104 GPU, due to its number of CUDA cores, the Geforce RTX 2060 is a rather interesting card. It features the same number of CUDA cores as the RTX 2060 Super, 2,176, but packs more VRAM, 12GB, on a narrower memory interface, 192-bit. With such configuration, the RTX 2060 12GB has the same 336GB/s memory bandwidth as the original RTX 2060.
Bear in mind that the original RTX 2060 packed 1920 CUDA cores and came with 6GB of 14Gbps GDDR6 memory on a 192-bit memory interface, while the RTX 2060 Super has 2176 CUDA cores and 8GB of 14Gbps GDDR6 memory on a 256-bit memory interface.
The GPU works at a 1470MHz base and 1650MHz GPU boost clock, and the TBP (typical board power) was raised from 160W to 185W.
There will be plenty of custom versions from the usual suspects, and according to rumors, we might even see a Founders Edition.
Of course, it will lack all the Ampere-series features like the Resizable BAR, PCI Express 4.0 support or Ampere-based RT and Tensor cores.
While Nvidia did not give out any details about the price or the availability date, some rumors suggest December 7th as the launch date.
We also suspect it could end up at around $349, which was the official MSRP of the Geforce RTX 2060 6GB before it got cut down to $299, at least on paper, as retail prices are a whole different story these days. Of course, there is always a chance that Nvidia will have plenty of Founders Edition stock as these usually sell at MSRP.