NVIDIA doesn’t really pick it, they sell the chips (the GPU), marrying it to VRAM is something that the card manufacturer does. They’ll have min and max numbers for the GPU and possibly advertising rules (it’s called X when it has more than Y VRAM, and called Z if it has less).
As for why, it’s a balance with cost and performance and what people are willing to pay. Adding double the VRAM doesn’t usually double the performance, but it could nearly double the price.
Latest Answers