Nvidia has introduced three new RTX GPUs using the ‘Super’ brand. Let’s start with the RTX 2060 Super, which uses the same TU106 Turing core found in the original RTX 2060 and 2070. However, it has four more streaming multiprocessors (SMs) enabled than the original RTX 2060, giving it 256 more stream processors, as well as more RT and Tensor cores. It also ups the memory from 6GB to 8GB, and it’s now attached to a 256-bit memory interface.
Meanwhile, the RTX 2070 Super now uses the TU104 Turing core first used in the RTX 2080, marking a step up from the original 2070. It only has 40 SMs enabled, compared with 46 on the 2080, but that still gives it 256 more stream processors than its predecessor, as well as more RT and Tensor cores.
Finally, the RTX 2080 Super offers a comparatively small upgrade over the first 2080, giving you 128 mores stream processors for a total of 3,072, as well as two more RT cores. The 2080 Super does have a notable tweak to memory, though, with its 8GB of GDDR6 RAM running at an effective frequency of 15.5GHz, rather than 14GHz.
GPU clock speeds have also seen some tweaking, with the RTX 2060 Super having a lower boost clock than its predecessor, but the other Super GPUs having higher boost clocks than their forerunners, including the original overclocked Founders Edition cards.
Prices for the top two Super GPUs are also lower than those for their older RTX brothers at launch, as you can see in the table below, while the RTX 2060 Super is priced slightly higher than the original 2060’s launch price. It’s possible that we’ll see a reduction in the price of first-gen RTX GPUs as a result of this launch too.
The RTX Super GPUs are available to pre-order now from Nvidia.