Maxing Out NVIDIA’s Tiniest Ada Silicon, AD107, in the NVIDIA GeForce RTX 4060

The GeForce RTX 4060, a 60-class product from NVIDIA, is a mid-range, high-volume graphics card built on the GeForce “Ada” graphics architecture. It may have specifications that allow NVIDIA to use either a fully optimized 5 nm AD107 silicon or a significantly downsized AD106. The RTX 4060 contains 3,072 CUDA cores spread among 24 streaming multiprocessors (SM), 96 Tensor cores, 96 TMUs, an unspecified number of ROPs, and 8 GB of 18 Gbps GDDR6 RAM, according to Kopite7kimi, a credible source with NVIDIA leaks. With this generation, it is difficult to anticipate the memory bus width. The on-die L2 cache of the GPU is 24 MB in capacity. Due to the card’s usual graphics power (TGP) of 115 W, cards can be constructed using just a single 6-pin PCIe power connector.