Nvidia Aims to Fix VRAM Issues with Neural Texture Tech

Recently on Twitter/X, @opinali examined the effects of Nvidia’s Neural Texture Compression technology (demo available here) when paired with DirectX Raytracing 1.2’s (DXR 1.2) Cooperative Vectors feature. 

Cooperative Vectors will allow developers to incorporate AI tasks into the real-time rendering workflow. This feature enables developers to leverage hardware-accelerated vector and matrix operations, encouraging innovative uses of AI within traditional gaming workloads.

By utilizing Neural Texture Compression (NTC), it's possible to decrease the size of textures by nearly 90%. This drastically cuts down the memory requirements for games, permitting more intricate visuals to be processed on hardware with less VRAM. With DXR 1.2's Cooperative Vectors feature, the AI capabilities of various GPU architectures can be harnessed to attain these outcomes. This signifies that Neural Texture Compression will not be exclusive to Nvidia, which is excellent news for all PC gamers.

According to @opinali, textures can represent 50-70% of a game’s VRAM consumption. This indicates that AI hardware could enhance high-end gaming experiences on graphics cards with limited VRAM. However, it’s important to note that this will not decrease the VRAM consumption of current games, and numerous modern titles are struggling to perform optimally on 8GB memory graphics cards.

In summary, Neural Texture Compression is an intriguing advancement. It holds significant potential to transform the gaming landscape. Additionally, it could enable developers to achieve more with less. Furthermore, this technology could also help reduce game download sizes, conserving valuable SSD space. Remember that this tech will be implemented in future PCs and consoles, meaning its widespread adoption is only a matter of time.

As always, for the latest news on hardware launches and industry developments, be sure to follow our dedicated hardware coverage.