Modern PC and console users have been familiar with AI-powered graphics upscaling technology for many years. Now, ARM Holdings – one of the biggest names in the mobile chip industry – is bringing similar technology to smartphones and tablets, opening a new era of graphics for mobile devices.
Neural Super Sampling: Better, lighter graphics for mobile GPUs
ARM has just introduced a new technology called Neural Super Sampling (NSS) – an AI-based real-time image upscaling method that doubles the resolution with only about 4 milliseconds of additional processing per frame. NSS allows game developers to render at low resolutions like 540p but output images nearly equivalent to 1080p, significantly reducing the workload on the GPU.
According to ARM, using NSS can reduce GPU load by up to 50%, thereby saving power or allowing for a significant increase in frame rates (FPS) on mobile devices. This is a major step forward in optimizing graphics performance on smartphones, especially in visually demanding games.
Compared to the PC market, Nvidia pioneered this with Deep Learning Super Sampling (DLSS) technology since 2019 on the GeForce RTX series. ARM’s new technology appears to be comparable to DLSS 1.0 or 2.0, but specifically designed for the mobile device ecosystem.
https://www.youtube.com/watch?v=XmdsWErzwC0
NSS was also developed based on Accuracy Super Resolution – ARM’s previous solution based on AMD’s FidelityFX Super Resolution 2 (FSR2) technology. FSR2 has already been integrated into several prominent games like Fortnite or Infinity Nikki.
Development tools and ARM’s expansion roadmap
To help developers easily integrate NSS, ARM has announced the Neural Graphics Development Kit – a specialized development kit for mobile. This package includes:
- Plugins for Unreal Engine,
- Vulkan emulator on PC,
- Performance optimization profiling tools,
- ML extensions for the Vulkan graphics API.
Furthermore, ARM has decided to open-source the AI models related to their graphics upscaling technology on GitHub and Hugging Face. This helps the development community gain easier access and accelerates the research application process.
In addition to NSS, ARM is also planning to deploy other AI technologies such as:
- Neural Frame Rate Upscaling,
- Denoising for path tracing (advanced image noise removal),
- And other variants of Neural Super Sampling.
These technologies are expected to be available to developers starting from 2026, before full hardware support is officially released.
ARM’s heavy investment in AI graphics for mobile demonstrates the shifting trend in the gaming and visual technology industry, as users increasingly demand premium experiences directly on their smartphones. If successful, ARM could help bridge the gap between PC and mobile graphics, creating a new generation of mobile games that are sharper and smoother than ever before.
