NVIDIA’s GeForce RTX 2080 Ti is based on the “TU102” silicon. NVIDIA was referring to this 775 mm² chip when talking about the 18.5 billion-transistor count in its keynote. The block diagram has been revealed, giving a breakdown of its various “cores”.
While the GPU still has the same design as it predecessors, each of the 72 streaming multiprocessors (SMs) has RT cores and Tensor cores in addition to CUDA cores.
So you have six GPCs (graphics processing clusters) with 12 SMs each. Each SM has 64 CUDA cores, 8 Tensor cores, and 1 RT core. Each GPC has six geometry units. The GPU has 288 TMUs and 96 ROPs. The TU102 also supports a 384-bit wide GDDR6 memory bus, which supports 14 Gbps memory. There are also two NVLink channels, NVIDIA will launch NVLink a little later as next-generation multi-GPU technology.
