Nvidia GeForce RTX 4080 offers two different configurations, 16GB and 12GB of VRAM respectively
It was previously reported that Nvidia has prepared two different configurations for the GeForce RTX 4070, although the memory speed is both 21Gbps, however, the capacity of GDDR6X is 10GB and 12GB respectively, and there is a big gap in performance between the two. Nvidia seems to be doing this not only with the GeForce RTX 4070 but with the GeForce RTX 4080 as well.
Recently, Twitter user @Zed__Wang said that Nvidia has provided two completely different public version designs for the GeForce RTX 4080, both equipped with GDDR6X memory, with a capacity of 12GB and 16GB respectively. The number of PCB layers of the two is also different. 12GB video memory is a 10-layer PCB, and 16GB video memory is a 12-layer PCB. The bit width of the former is 192 bits, and the bit width of the latter is 256 bits. Due to the reduction of video memory bit width, in order to make up for the lack of video memory bandwidth, higher-speed video memory is required. In addition, the number of CUDA cores or frequency may also be different.
Nvidia’s approach is not surprising, with the GeForce RTX 3080 series showing up in a number of different configurations. However, on the GeForce RTX 4090, Nvidia only provides a public version design solution to its partners, equipped with 24GB of GDDR6X memory and a 14-layer PCB. By comparison, the GeForce RTX 3080 and RTX 3090 series currently on sale are both 12-layer PCBs.
It’s unclear if both the 16GB and 12GB versions of the GeForce RTX 4080 will be available, or in what order. Regardless of the version of the GeForce RTX 4080, the overall power consumption of the graphics card appears to have been reduced compared to earlier rumors.
