Nvidia AI tech claims to slash gaming GPU memory usage by 85% with zero quality loss — Neural Texture Compression demo reveals stunning visual parity between 6.

Nvidia AI tech claims to slash gaming GPU memory usage by 85% with zero quality loss — Neural Texture Compression demo reveals stunning visual parity between 6.

(Image credit: Nvidia) (Image credit: Nvidia) Textures by far consume the most VRAM in any game, so any technique to keep them in check is a welcome addition. That said, it's important to note that this isn't exclusive to Nvidia, as Microsoft has standardized it as " Cooperative Vectors " in DirectX. Intel has previously shown off its own demo with noticeably better textures compared to block compression. AMD last talked about the tech in 2024 , but it's likely onboard the mission as well.

Currently, no game supports Cooperative Vectors or Nvidia's Neural Texture Compression, but we should start to see it implemented soon, given the industry's trajectory. AI has become the answer to seemingly every age-old problem, and corporations are inventing new ways to incorporate it where it doesn't belong . Innovations like NTC, however, show that it can be implemented tastefully to make an actual, meaningful difference.

Follow Tom's Hardware on Google News , or add us as a preferred source , to get our latest news, analysis, & reviews in your feeds.

Hassam Nasir is a die-hard hardware enthusiast with years of experience as a tech editor and writer, focusing on detailed CPU comparisons and general hardware news. When he\u2019s not working, you\u2019ll find him bending tubes for his ever-evolving custom water-loop gaming rig or benchmarking the latest CPUs and GPUs just for fun. ","collapsible":{"enabled":true,"maxHeight":250,"readMoreText":"Read more","readLessText":"Read less"}}), "https://slice.vanilla.futurecdn.net/13-4-19/js/authorBio.js"); } else { console.error('%c FTE ','background: #9306F9; color: #ffffff','no lazy slice hydration function available'); } Hassam Nasir Social Links Navigation Contributing Writer Hassam Nasir is a die-hard hardware enthusiast with years of experience as a tech editor and writer, focusing on detailed CPU comparisons and general hardware news. When he’s not working, you’ll find him bending tubes for his ever-evolving custom water-loop gaming rig or benchmarking the latest CPUs and GPUs just for fun.

beyondlogic Admin said: Nvidia has just demoed its Neural Texture Compression technique again at a GTC talk, where it showed VRAM usage dropping from 6.5 GB to just 970 MB in a scene. NTC uses a neural network to decompress textures instead of standard block-based compression, reducing texture size and VRAM usage while also improving final image quality. Nvidia AI tech claims to slash VRAM usage by 85% with zero quality loss — Neural Texture Compression demo reveals stunning visual parity between 6…. : Read more its decent but i get the feeling moving objects and scenes are going to breakdown. even the plate shown has a blurry outlook due to lack of ram. this is just going to give gpu vendors a excuse for lower memory cards and most likely see more horrid looking games. lol. is it good yes can it be used for bad yes lol. Reply

vanadiel007 So this is how they are planning to keep the profit margins up during this RAM shortage: a new compression technology that only needs 1 Gig instead of 8 Gig VRAM to provide you those same great graphics. Reply

usertests Some form of neural texture compression is already confirmed to be coming to Xbox Helix, and is highly likely to be used by the PS6. So if you have concerns about this, too bad, it's inevitable. Will it be used by Nvidia and AMD to skimp out on VRAM and offer more low VRAM cards? Possibly. But if the next-gen consoles include around 24-36 GB of memory, 8-12 GB cards won't cut it forever. Reply

warezme Will this work on video streaming? Imagine the boon to network providers if streaming services for apps like Netflix, Hulu, etc. requiring a fraction of the bandwidth to produce higher resolution streams. Reply

psyconz beyondlogic said: its decent but i get the feeling moving objects and scenes are going to breakdown. even the plate shown has a blurry outlook due to lack of ram. this is just going to give gpu vendors a excuse for lower memory cards and most likely see more horrid looking games. lol. is it good yes can it be used for bad yes lol. It's a lossless compression technique. Zero image quality loss. Not all AI is garbage trained on the internet… Reply

thestryker This is the technology that shows the most promise of the ai based things nvidia has put forth. As long as nvidia's implementation isn't proprietary to the point that developers need separate implementation for other vendors I could see this getting adoption. I'm also curious if it has to be used for everything being rendered or if it can work in conjunction with existing techniques. Reply

timsSOFTWARE beyondlogic said: its decent but i get the feeling moving objects and scenes are going to breakdown. even the plate shown has a blurry outlook due to lack of ram. this is just going to give gpu vendors a excuse for lower memory cards and most likely see more horrid looking games. lol. is it good yes can it be used for bad yes lol. I doubt it will actually be "zero quality loss", but it may very well be a lot better than using a lower mip/lower resolution. People with lower-end cards will benefit the most, if it works on their hardware. Reply

usertests timsSOFTWARE said: I doubt it will actually be "zero quality loss", but it may very well be a lot better than using a lower mip/lower resolution. People with lower-end cards will benefit the most, if it works on their hardware. It will be interesting to see this play out, and how old you can go. I think it would likely work on at least the RTX 5050 and 9060 XT 8GB. But AMD's version coming to next-gen consoles would be running on RDNA5. AMD users have been slapped in the face when it comes to FSR4, and it can happen again with this. Reply

QuarterSwede psyconz said: It's a lossless compression technique. Zero image quality loss. Not all AI is garbage trained on the internet… Then the examples they gave aren’t indicative of this technique since you can clearly see quality loss … Reply

-Fran- You only need 640KB of memory. We all know that. Regards. Reply

Key considerations

  • Investor positioning can change fast
  • Volatility remains possible near catalysts
  • Macro rates and liquidity can dominate flows

Reference reading

More on this site

Informational only. No financial advice. Do your own research.

Leave a Comment