- Cuda Cores: 4,352
- Tensor Cores: 544
- RT Cores: 68
- Render Outputs: 88
- Texture Units: 272
- Boost Clock: 1,545MHz
- Base Clock: 1,350MHz
- Memory: 11GB GDDR6
- Memory Speed: 14Gbps
- Memory Interface: 352-bit
- Memory Bandwidth: 616GB/s
GeForce RTX 2080 Ti Super Versus Titan RTX
Going with a full Turing TU102 GPU and faster 16Gbps memory might be the entirety of the upgrade. However, if NVIDIA really wanted to, it could also go with more memory and faster faster GPU clocks. Doing so would leave the Titan RTX in a precarious spot, though. Should the GeForce RTX 2080 Ti Super come to fruition as outlined in the leak, the Titan RTX would still have an advantage in the amount of memory (24GB versus 11GB) and the boost clock (1,770MHz versus 1,545MHz for the reference model, and 1,635MHz for the Founders Edition).
To make things easier to visualize, here are the core specs of Titan RTX, GeForce RTX 2080 Ti Super (as leaked), and GeForce RTX 2080 Ti (both reference and Founders Edition models)...
- Titan RTX: 4,608 CUDA cores, 24GB GDDR6 (14Gbps), 1,350MHz base, 1,770MHz boost
- 2080 Ti Super: 4,608 CUDA cores, ?? GDDR6 (16Gbps), ???? base, ???? boost
- 2080 Ti: 4,352 CUDA cores, 11GB GDDR6 (14Gbps), 1,350MHz base, 1,545MHz boost
- 2080 Ti FE: 4,352 CUDA cores, 11GB GDDR6 (14Gbps), 1,350MHz base, 1,653MHz boost
Will there actually be a GeForce RTX 2080 Ti Super, though? I can't say for certain, but it's worth noting that kopite7kimi has been correct in the recent past. In September, the user leaked some accurate details about the GeForce GTX 1660 Super and GeForce GTX 1650 Super, neither of which had launched at the time.
If this new card does end up launching, it will be interesting to see how it's priced. Hopefully NVIDIA sticks with the same MSRP as the non-Super model. That could lead to some tantalizing deals on the non-Super model as vendors look to rid themselves of the old inventory to make room for the new part.