Published in Graphics

AMD mocks Nvidia’s memory size

by on13 April 2023

Have you been swimming?

AMD issued a blog post yesterday, right before the launch of the RTX 4070 mocking NVIDIA’s “lackluster” GDDR6 and GDDR6X memory sizes on its RTX 30/40 series graphics cards.

AMD said Nvidia memory sizes were so small that a Florida teacher was forced to resign after showing them to students [are you being silly? Ed]

The chipmaker said enthusiasts demand the best gaming experience possible, and video memory capacities below 12GB are not enough to run the latest titles at 4K ultra settings anymore.

AMD targeted Resident Evil 4, The Last of Us Part 1, and Hogwarts Legacy. AMD said the three games to use at least 11 GB of VRAM at 4K with rasterised graphics alone. (With ray-tracing, memory capacity jumps by an additional two gigabytes.).

AMD now officially recommends 16GB of video memory for 4K gaming, and 12GB for 1440P gaming. Budget gamers still using 1080P panels, AMD recommends 8GB GPUs, since 1080P resolutions put “less strain” on video memory.

These new recommendations by AMD are well beyond what most of NVIDIA’s RTX 30 series and even the RTX 4070 and 4070 Ti can handle. The RTX 3060 Ti, RTX 3070, RTX 3070 Ti and RTX 3080 10GB have memory capacities well under 12GB, and it's not until you reach the RTX 3090 series/RTX 4080 that AMD’s 4K guidelines can be met.

AMD’s RX 6000 series graphics cards start at 12GB of VRAM with the mid-range RX 6700 XT (that’s retailing now for just $350) and go up to 16GB with the RX 6800 series, RX 6900 XT, and RX 6950XT graphics cards. The RX 7900 series GPUs start at 20GB of capacity.

This is not just a case of AMD marketeers setting a standard that it knows Nvidia can’t beat. Video memory issues surrounding the RTX 3080 10GB and RTX 3070 series are becoming problematic. Some have been moaning that certain 2023 AAA titles must be tuned down to medium-quality settings in some instances to have a smooth and playable game.


Last modified on 13 April 2023
Rate this item
(2 votes)

Read more about: