Thursday, August 6th 2026
NVIDIA Neural Texture Compression Now Runs on RTX Spark
NVIDIA has officially brought its RTX Neural Texture Compression (NTC) technology to Windows-on-Arm, gearing up for a general release on its RTX Spark PC platform. First demonstrated at GTC 2026 in May, NVIDIA showcased how NTC can significantly reduce GPU VRAM usage by up to seven times. In a technology demo at one of the GTC 2026 sessions, NVIDIA revealed that NTC can lower VRAM usage from 6.5 GB to just 970 MB in certain scenes. NTC technology is an AI-driven texture output that enables games to feature high-quality, complex materials without sacrificing performance. Games can take advantage of the substantial VRAM reduction that NTC offers while maintaining the same texture quality.
For the upcoming RTX Spark, NVIDIA is enhancing the software ecosystem so gamers and future buyers receive a mature software that works seamlessly out of the box, eliminating the need for workarounds. Even Japan's legendary SEGA has released AAA titles optimized for RTX Spark and is dedicating more resources to support the ecosystem. NVIDIA has already started shipping its first preview build of the CUDA toolkit for the RTX Spark platform, specifically as a native Windows-on-Arm software toolkit. Now, emerging technologies like NTC will run on RTX Spark, which includes configurations such as one with 6,144 CUDA cores and another with 5,120 CUDA cores. Both configurations are based on the "Blackwell" GPU microarchitecture and are paired with a 20-core Arm CPU and up to 128 GB of LPDDR5X memory.Traditionally, games use block-compressed formats like BCn, such as BC5, BC6, or BC7, which are commonly applied in 4x4 pixel formats and rendered by the GPU. However, NVIDIA has trained small neural networks that can produce the desired pixel format and texture appearance at a fraction of the size of traditional texture compression formats. Instead of using gigabytes of VRAM for textures, NTC drastically reduces VRAM usage by emulating textures, allowing for either much lower VRAM consumption or significantly enhanced material appearance.
Sources:
NVIDIA on GitHub, via VideoCardz
For the upcoming RTX Spark, NVIDIA is enhancing the software ecosystem so gamers and future buyers receive a mature software that works seamlessly out of the box, eliminating the need for workarounds. Even Japan's legendary SEGA has released AAA titles optimized for RTX Spark and is dedicating more resources to support the ecosystem. NVIDIA has already started shipping its first preview build of the CUDA toolkit for the RTX Spark platform, specifically as a native Windows-on-Arm software toolkit. Now, emerging technologies like NTC will run on RTX Spark, which includes configurations such as one with 6,144 CUDA cores and another with 5,120 CUDA cores. Both configurations are based on the "Blackwell" GPU microarchitecture and are paired with a 20-core Arm CPU and up to 128 GB of LPDDR5X memory.Traditionally, games use block-compressed formats like BCn, such as BC5, BC6, or BC7, which are commonly applied in 4x4 pixel formats and rendered by the GPU. However, NVIDIA has trained small neural networks that can produce the desired pixel format and texture appearance at a fraction of the size of traditional texture compression formats. Instead of using gigabytes of VRAM for textures, NTC drastically reduces VRAM usage by emulating textures, allowing for either much lower VRAM consumption or significantly enhanced material appearance.


17 Comments on NVIDIA Neural Texture Compression Now Runs on RTX Spark
If it's only on the DX Spark now, then this may take a while to get to us.
Ideally it would be an open source API tech, embedded in DX12, so that us AMD folks could use it as well... Alas, gotta wait and wait longer.
If compression can be done on-the-fly, this would enable Nvidia to dramatically reduce vram on their cards (e.g. selling 4gb cards again) thus reducing cost and undercutting competitors all while providing the same level of visual fidelity.
Bonus points if compression can be precompiled then cached on disk to avoid stuttering when new scenes are loaded.
You almost word it in a way that implies AMD could have gone proprietary only a decade back and saw increased success when in fact open source was probably the only option that was financial viable for them and that didn't eat up what little engineering resources they had. Realistically AMD didn't have the choice of going proprietary until after it saw success with a few generations of Zen and had money coming in. Even then, AMD going prorietary isn't a good thing for market. It's AMD trying to play Nvidia's game, they are doing so at a significant resource and implementation disadvantage as disucessed earlier and it's bad for customers. Either way you slice it, Nvidia is dictating the market.
DLSS wasn't what started Nvidia being percieved as higher value. That started far far before DLSS ever came around. Even Fermi outsold it's AMD counterparts 3 to 1 and that was ages before DLSS was even a twinkle in Jensen's eye.
NTC being proprietary is certainly not a good thing either but one can hope that any impelementations will go through an API middle layer that will allow game devs to route that to vendor appropriate technologies. That said, if path tracing implementations are anything to go by and their lack of support in some games for AMD hardware despite the API and hardware supporting it, I cannot say I'm hopeful.
Aside from further perpetuating the surveillance datacenter grid, making every electronics more expensive, this will also ruin games.
Open implementations mean that someone is going to spend millions in R&D for something their competitors can use.
Get ready for the RTX 6090 8GB! for only 2999 USD (MSRP, real market price may vary).
That is a Price reduction by more than 100% compared to the real market price of the 5090 today!
100% cheaper, price compression complete.
NTC will also require devs to compress an additional set of their textures in a specialized format, so it'll add to a game's size. Devs would basically be compressing textures just for newer Nvidia cards without a standard API implementation that works with other vendors. Nevermind that I'm not certain it'd run on lower end cards that must also be running DLSS and other tech that uses the tensor cores like the AI denoiser for RT / PT.
I can imagine some games might implement it, simply because Nvidia can sponser them to do so. They have a ton of money to throw around.
I've been building my own PCs foe like 3 decades and I don't recall the PC ever being in the bad position we see today 😭