NVIDIA today refreshed the top-end of the GeForce RTX 30-series "Ampere" family of graphics cards with the new GeForce RTX 3080 Ti, which we are Releasing on 3rd Jun. The RTX 3080 Ti is being considered the next flagship gaming product, picking up the mantle from the RTX 3080. While the RTX 3090 is positioned higher in the stack, NVIDIA has been treating it as a TITAN-like halo product for not just gaming, but also quasi-professional use cases. The RTX 3080 Ti has the same mandate as the RTX 3080—to offer leadership gaming performance with real-time raytracing at 4K UHD resolution.
The GeForce RTX 3080 Ti is based on the same 8 nm GA102 silicon as the RTX 3080, but with more CUDA cores, while maxing out the 384-bit wide GDDR6X memory bus. It only has slightly fewer CUDA cores than the RTX 3090, the memory size is 12 GB as opposed to 24 GB, and the memory clock is slightly lower. NVIDIA has given the RTX 3080 Ti a grand 10,240 CUDA cores spread across 80 streaming multiprocessors, 320 3rd Gen Tensor cores that accelerate AI and DLSS, and 80 2nd Gen RT cores. It also has all 112 ROPs enabled, besides 320 TMUs. The 12 GB of memory maxes out the 384-bit memory bus, but the memory clock runs at 19 Gbps (compared to 19.5 Gbps on the RTX 3090). Memory bandwidth hence is 912.4 GB/s.
Typically replies within a few minutes.