At its Arm Everywhere China event in Shanghai, Arm announced a new AI-native compute platform and GPU designed to bring desktop-class graphics to mobile devices while reducing power consumption and heat. The British semiconductor and software design company said the approach is intended to offer an open alternative to Nvidia’s DLSS 5 technology and to support a full AI-native computing ecosystem.

Arm’s announcement includes the AI-native compute platform CSS for Mobile 2 and the Arm Mali G2-Ultra NX GPU, a chip built around neural computing. Arm claims the GPU can improve graphics performance at a much lower cost in terms of heat and power consumption.

The company did not confirm which devices will use the Mali G2-Ultra NX GPU. The earlier Mali G1-Ultra chip is found in devices from Xiaomi, Vivo, and OnePlus.

Why mobile ray tracing needs a different approach

Deyan Lazarov, a member of Arm’s GPU product team in the Edge AI business unit, said Arm has been working on the Mali G2-Ultra NX for the past two and a half years. The chip is intended to allow mobile developers to use compute-intensive technology such as ray tracing.

“There was actually growing adoption of ray tracing in mobile titles, ” Lazarov said. “The technology has been a bit slow to get going, but now we’re seeing more and more titles actually interested in enabling ray tracing.”

Arm said the new GPU allows mobile developers to take advantage of Unreal Engine technologies such as MegaLights. Arm worked with Sumo Digital to create Neural Dawn, a game used to showcase the Mali G2-Ultra NX.

Lazarov said mobile devices face constraints that consoles and PCs do not. “They cannot just dissipate heat as much as a console can, as much as a PC can, and they are operating off of a battery all the time.”

Arm’s solution integrates neural processing with the traditional rendering pipeline. He said this can reduce GPU workload and power consumption, which has system-level effects on device efficiency.

How the AI reconstruction technology works

Arm’s key technologies include Neural Super Sampling (NSS), Neural Super Sampling and Denoising (NSSD), and Neural Frame Rate Upscaling (NFRU). The technology can upscale images, cast additional rays, and generate further frames using AI reconstruction.

Lazarov gave an example of two frames. In the first frame, the GPU renders one quarter of the image and the remaining three quarters are reconstructed on Arm’s neural accelerators through NSS. In the second frame, no traditional GPU rendering occurs; all pixels are AI reconstructed.

Games expected to use the new chip include Arena Breakout Infinite from Tencent, Where Winds Meet from NetEase, and Infinity Nikki from Infold Games.

Arm said its networks are convolution type networks, not generative models. “Our networks are not generative, so they’re not from the same nature as the latest DLSS model, ” Lazarov said. “Our networks are convolution type networks, which means that they don’t imagine anything, they don’t create anything, they just kind of know what things need to look like.”

Arm has made its resolution and frame rate generation networks public on GitHub and Hugging Face. Developers can capture datasets from their own games and retrain Arm’s networks to match a specific art style.

Open approach compared with DLSS 5

Lazarov contrasted Arm’s approach with Nvidia’s DLSS 5. He described DLSS as closed and proprietary.

“DLSS is not open. It’s not like our network. It’s proprietary, it’s closed, it’s a black box, you don’t know what’s happening, you don’t know what it’s trained on. You cannot retrain it on your own [data]. You integrate it, you’re happy with the results, or you’re not. It’s a bit of a take it or leave it, whereas [ours] is take it and make it so that you like it, ” Lazarov said.

Arm said its technology was developed with a mobile-first approach. The networks are designed to operate within mobile power, bandwidth, and other budgets. Lazarov said a technology like DLSS would not run on a phone even with software support because it would exceed those budgets.

The Mali G2-Ultra NX also targets reduced memory traffic. Arm reported a 70 percent reduction in traffic out to DRAM, which is a major contributor to mobile power consumption. Arm said the new GPU achieves 4X performance and 4X efficiency improvements.

Lazarov said that when last year’s GPU was pushed to its maximum limit, it could run Neural Dawn at 40 fps, and only 15 fps under normal conditions. The new GPU can run the game at 60 fps.

Developer tools and next steps

Arm currently provides plug-ins for Unreal Engine and has a roadmap to expand to other game engines. Lazarov said Unity is being considered, but Arm prioritized Unreal for now. Arm recently announced a partnership with Unity China.

For developers using