The GeForce RTX 4080 graphics card, NVIDIA’s second-fastest product from the GeForce “Ada Lovelace” generation, was officially unveiled today. The card is $400 less expensive than the RTX 4090 flagship, with a starting price of USD $1,200. It is technically intended to replace the RTX 3080 12 GB, whereas the RTX 3080 10 GB was intended to be replaced by an RTX 4080 12 GB model. The RTX 4080 12 GB, the only SKU with the name RTX 4080, was canceled by NVIDIA after receiving negative press for having specs that were significantly different from those of the RTX 4080 16 GB.

Based on the 4 nm “AD103” silicon, the GeForce RTX 4080 has 76 streaming multiprocessors and 9,728 CUDA cores. In addition to 112 ROPs, it receives 304 4th generation Tensor cores and 76 RT cores. Although it contains 16 GB of memory, which is a generational increase, its 256-bit GDDR6X memory bus is smaller. In an effort to counteract this, NVIDIA sped up the memory sub-system by using faster 22.4 Gbps-rated memory and architectural advancements like larger caches on the silicon. Not only is NVIDIA releasing the Founders Edition card, but also custom-designed boards from its partners. Every partner whose selection we’ve seen so far has at least one SKU with a starting price of $1,200. On November 16, 2022, the cards should be available.