NVIDIA today gave its GeForce RTX 20-series graphics card lineup a mid-lifecycle update with the new "Super" RTX 20-series. It may not seem so, but the RTX 20-series graphics cards have been around for over 8 months, and AMD is finally giving the mid-segment some competition with its 7 nm PCIe Gen 4 Radeon RX 5700 "Navi" series with the company claiming competitive performance leads over the high-volume GeForce RTX 2060 and RTX 2070. NVIDIA decided to upscale the two SKUs at their price points with new models. We hence have the new GeForce RTX 2070 Super and GeForce RTX 2060 Super, which will be available starting on the 9th of July, priced at $399 and $499, respectively. The two will be joined by the high-end GeForce RTX 2080 Super later this month, on the 23rd.
The original RTX 2060 was carved out of the 12 nm "TU106" silicon by chopping off a quarter of its memory size/speed and enabling 1,920 out of the 2,304 CUDA cores physically present on the chip. The new RTX 2060 Super restores the memory subsystem to its full glory. The GPU now has 8 GB of GDDR6 memory across a 256-bit wide memory interface. At its rated speed of 14 Gbps, this works out to 448 GB/s of memory bandwidth. Interestingly, this is the same exact amount of memory and bandwidth as the $700 RTX 2080. NVIDIA enabled more CUDA cores present on the silicon. The CUDA core count is up at 2,176. The RT core count is proportionately increased to 34 from 30, and the Tensor core count is 272 from 240. NVIDIA also slightly pushed up the GPU clock speed, to 1470 MHz from 1365 MHz, although the GPU Boost frequency is trimmed to 1650 MHz from 1680 MHz.
NVIDIA is targeting the Radeon RX 5700 with the RTX 2060 Super, and unlike the RTX 2070 Super, it doesn't displace the original RTX 2060 from its price point of $349. Rather, it's priced at a $50 premium at $399. It's also $20 pricier than the $379 MSRP of the upcoming RX 5700, which will spice things up in the sub-$400 segment. The original RTX 2060 already demonstrated 1440p gaming credentials across most of our vast selection of games, and the RTX 2060 Super could only build on that. The extra 2 GB of memory will certainly help.
As we stated above, the RTX 2060 Super is carved out of the "TU106" silicon by disabling just one of the TPCs (two streaming multiprocessors), making it almost an RTX 2070 if you know how to overclock. This isn't the first time NVIDIA has done something of this nature. The GTX 1070 Ti is almost a GTX 1080. The resulting specifications are 2,176 CUDA cores, 136 TMUs, 64 ROPs, 34 RT cores, and 272 Tensor cores. The biggest change, though, is the memory: the full 256-bit bus width is enabled, and the GPU has 8 GB of GDDR6 memory at its disposal.