Nvidia’s New DLSS SDK Update Reduces VRAM Usage by Up to 20 Percent
If you’ve been frustrated by how much video memory modern games demand, especially while using Nvidia’s DLSS upscaling tech, relief is on the way. The latest version of Nvidia’s Deep Learning Super Sampling (DLSS) SDK has been updated to significantly reduce VRAM usage, making it more efficient on supported graphics cards.
According to details shared by VideoCardz, the new update cuts VRAM demands by about 20 percent when DLSS is used. This change applies across multiple resolutions—from standard 1080p gameplay to ultra-high-end 8K setups. While the benefits will vary depending on your resolution, the update represents a notable optimization for gamers and developers alike.
This is particularly good news for RTX 30-series owners and even some users of RTX 20-series GPUs. DLSS has already become a vital feature for maintaining high frame rates in visually intensive games, and now it comes with the added perk of using less memory to do the job. That means more headroom for better textures, effects, or simply better system stability under load.
The memory savings scale with resolution: about 20MB less usage at 1080p, and more than 1GB at 8K. While the latter only impacts those running ultra-premium hardware, the overall efficiency gains are meaningful across the board. Games updated with this new SDK will benefit immediately, though developers will need to implement the changes on a case-by-case basis.
As modern games continue to push the limits of GPU memory, this update is a smart move by Nvidia—especially as more mid-range cards struggle to keep up with ballooning system requirements. Even if the savings aren’t dramatic for every user, improvements like this help ensure DLSS remains one of the most powerful tools in a PC gamer’s arsenal.