As rumored earlier, the Nvidia Geforce RTX 4070 SUPER is based on AD104 GPU, packing 56 SMs for a total of 7168 CUDA cores, 224 Tensor cores, and 56 RT cores. It also gets all 80 ROPs, which should make it significantly faster than the RTX 4070. According to Nvidia, this means 20 percent more cores and makes the card faster than the Geforce RTX 3090. The RTX 4070 SUPER will still come with 12GB of 21Gbps GDDR6X memory on a 192-bit memory interface, leaving it with 504GB/s of maximum bandwidth. The TDP is slightly up at 220W.

Performance-wise, the RTX 4070 SUPER should do quite well, pushing well over 60 FPS in most popular games at 1440p resolution, even without DLSS. At $599, it is a decent upgrade from the RTX 4070, which will now drop to $549.

The Nvidia Geforce RTX 4070 Ti SUPER is a bit strange as Nvidia had to cut it down enough not to thread on the Geforce RTX 4080 territory, but still give it an upgrade compared to the RTX 4070 Ti. It uses the AD103 GPU and packs 8448 CUDA cores and 16GB of 21Gbps GDDR6X memory on a 256-bit memory interface. It will draw close to the RTX 4080, but since it will be replacing the RTX 4070 Ti, it should fit well.

Nvidia is still targeting 1440p high frame rate gaming with the RTX 4070 Ti SUPER, and at $799, it could be a decent choice.

The Geforce RTX 4080 SUPER will replace the RTX 4080 on the market, and at $999, it will offer 10240 CUDA cores from AD103 GPU, and 16GB of 23Gbps GDDR6X memory on a 256-bit memory interface, giving it a slightly higher memory bandwidth compared to the RTX 4070 Ti SUPER. It targets 4K gaming, so we'll see where it fits once some reviews show up.

So far, it appears that the RTX 4070 SUPER and the RTX 4080 SUPER will get the reference version and Founders Edition cards, while the RTX 4070 Ti SUPER will only come from Nvidia AIC partners.