It appears that NVIDIA will indeed be announcing the GeForce RTX 20 series’ Super variants next week, with a launch in July (both the Founder Edition and board-partner cards). According to Igor Wallossek from TomsHardware Germany, the RTX 2080 Super will use the TU104 die instead of the previously predicted TU102.
With all the SMs enabled on the TU104-450 GPU, we are looking at 3,072 CUDA Cores and 8GB (NOT 11) of faster GDDR6 memory (16Gbps vs 14), making it one of the more faster Supers. The other GPUs will be trimmed down versions of higher-end models, with additional shaders and in case of the RTX 2060 Super more VRAM (8GB) and as well as a wider bus (256-bit).
These RTX Super cards will reportedly come with the Super label on the shroud, and Videocardz has come up with a mock-up of the design:
As the new Turing cards are announced at the existing price points of the 20-series GPUs, the older variants are going to get price cuts ranging from $50-100, prompting AMD to do the same with their upcoming Radeon RX 5700 graphics cards. NVIDIA’s