When NVIDIA Might Launch An Entry-Level GeForce RTX 4050 And What To Expect

A new rumor was published on Twitter detailing the potential release date and memory specifications for NVIDIA’s upcoming RTX 4050 graphics card. The GPU could come as early as June of this year, and possibly launch before the RTX 4060. However, the rumor also suggests the RTX 4050 will come with less VRAM than the RTX 3050 8GB it's supposed to replace, regressing to just 6GB.

A reduction in memory capacity can be forgivable if the GPU has enough raw horsepower to make up the difference. However, we’re living in an era where memory capacity is almost as important as the GPU itself. Features like high-resolution texture and ray-tracing are now consuming a significant amount of video memory, to the point where 8GB and even 10GB graphics cards *RTX 3080 cough cough* are starting to struggle.

Once a GPU runs out of VRAM, performance tanks because the GPU is forced to swap out important data from the VRAM to the system RAM to keep the game running. But even this isn’t enough to prevent crashes in severe circumstances. Another side-effect is related to newer games, where the game will automatically prevent high-resolution textures from loading into the graphics card. This will save performance, and make the game run ok, but it will make objects in the game world look blurry and delete any fine-grain detail they should have.

A good example of this comes from a couple of videos recently made by Daniel Owen on YouTube, where he discovered the RTX 3070 Ti 8GB crashes in Resident Evil 4, while running the game at maxed-out settings at 1440P with ray tracing enabled. To get the game running, he had to lower texture quality to an almost unacceptable level. This wasn’t an issue with AMD’s competitor, the RX 6700 XT which has 12GB.

We can’t say how the RTX 4050 will perform without testing ourselves, but there is a good chance that gamers will have to sacrifice ray tracing graphics if they want to use higher rasterized graphical settings on the RTX 4050 or vice versa. Normally we would say this isn’t a problem for an entry-level GPU, but having to sacrifice RT features you paid for due to memory capacity issues alone is understandably frustrating. Full RTX 4050 GPU specifications are unknown at this point, but the latest rumors for the RTX 4060 suggest it will utilize NVIDIA’s AD107 die, with 3072 CUDA cores, and 8GB of GDDR6 memory on a 128-bit bus. As a result, we can probably expect the RTX 4050 to have a partially disabled GA107 die, with a memory bus reduction down to 96 bits due to the 6GB capacity.