Arm Details Its Mobile‑First AI Reconstruction Approach, Distinct from the "Black Box" Style of DLSS 5

Arm, the British semiconductor and software‑design powerhouse, hosted its Arm Everywhere China showcase in Shanghai to announce a suite of fresh AI‑computing solutions aimed squarely at mobile devices. Central to the reveal was the company’s vision of an AI‑native ecosystem that includes the new CSS for Mobile 2 compute platform and the Mali G2‑Ultra NX graphics processor. This GPU is built around neural‑accelerator technology, promising higher graphics throughput while keeping heat and power draw to a minimum. Arm positions its on‑device AI‑driven graphics pipeline as a more transparent alternative to Nvidia’s contentious DLSS 5, aligning more closely with the original intent of game developers.

While Arm did not disclose which specific products will ship with the Mali G2‑Ultra NX, the predecessor Mali G1‑Ultra already powers smartphones from brands such as Xiaomi, Vivo and OnePlus, suggesting a similar market trajectory for the new chip. Deyan Lazarov, a member of Arm’s GPU team within the Edge AI division, explained that the G2‑Ultra NX has been under development for roughly two and a half years.

Its purpose is to give mobile developers access to heavyweight rendering techniques—most notably ray tracing—without sacrificing the power‑efficiency constraints of handheld devices. "There’s a growing appetite for ray tracing on phones," Lazarov noted. "It was a slow burn at first, but now more titles are actively exploring how to incorporate it." Recent reports from GamesIndustry.biz highlighted that the new GPU can tap into Unreal Engine features like MegaLights.

Arm even partnered with Sumo Digital to produce a demo title, *Neural Dawn*, which showcases the Mali G2‑Ultra NX’s capabilities. According to Lazarov, the chip targets three categories of games: those that are "mobile‑first" yet aspire to desktop‑level visual fidelity; cross‑platform titles that aim for a seamless experience across consoles, PCs and phones; and legacy games that developers want to refresh with modern graphics without a full rebuild. He cited upcoming titles such as Tencent’s *Arena Breakout Infinite*, NetEase’s *Where Winds Meet* and Infold Games’ *Infinity Nikki* as early adopters.

Mobile hardware, unlike consoles or PCs, operates under strict thermal and power budgets. "A phone can’t dump heat the way a console can, and it’s always running off a battery," Lazarov explained. Arm’s answer is to fuse neural processing directly into the traditional rendering pipeline.

By offloading portions of the workload to dedicated neuro‑accelerators, the GPU’s power consumption drops, which in turn lifts system‑level efficiency. This reduction in GPU demand means developers can allocate saved resources to higher visual quality, smoother frame rates, longer battery life—or any combination thereof. To make the AI‑native graphics stack appealing to developers, Arm is embracing an open‑software philosophy.

Over the past two years the company has built a toolbox that includes plug‑ins for Unreal Engine and a roadmap that eventually covers Unity and other engines. "We wanted an AI approach that is developer‑friendly, giving them tangible efficiency gains they can reinvest where they see fit," Lazarov said. Arm’s suite of technologies comprises Neural Super Sampling (NSS), Neural Super Sampling and Denoising (NSSD) and Neural Frame‑Rate Upscaling (NFRU).

In practice, these tools can upscale a low‑resolution render, add extra rays for better lighting, and even synthesize new frames using AI reconstruction. Lazarov illustrated the process with a two‑frame example: in the first frame the GPU renders only 25 % of the pixels, while the remaining 75 % are reconstructed by the neuro‑accelerators via NSS. In the second frame, the GPU does no traditional rendering at all—the entire image is generated by the AI pipeline.

Nvidia’s DLSS 5 also uses AI to enhance image detail, but it has drawn criticism for being a "black box" that can diverge from a developer’s artistic intent. Arm draws a clear line between its method and Nvidia’s. "Our networks are not generative; they’re convolutional models that don’t imagine new content, they simply know how existing content should look," Lazarov explained.

The networks are trained on real‑world data and then released publicly on platforms like GitHub and Hugging Face. This openness allows developers to fine‑tune the models with their own art assets, ensuring the upscaled output respects the unique visual style of each game and preserving intellectual‑property control.

By contrast, DLSS is a proprietary, closed system. Developers cannot inspect the training data or retrain the model; they must accept whatever results Nvidia delivers. "It’s a take‑it‑or‑leave‑it proposition," Lazarov said.

"Our approach lets you take the model and adapt it until it fits your vision." Another fundamental distinction is the mobile‑first design of Arm’s solution. The neural networks are engineered to operate within the tight power, bandwidth and memory constraints of smartphones.

"If you tried to run DLSS on a phone, you’d instantly blow past the power budget," Lazarov warned. Arm’s architecture also dramatically cuts memory traffic—by roughly 70 %—which is a major source of power draw on mobile SoCs.

Less DRAM access translates directly into longer battery life and cooler operation. Performance numbers underscore the advantage. When pushed to its limits, the previous‑generation GPU could only sustain *Neural Dawn* at about 40 fps under ideal conditions and dropped to 15 fps in typical use.

The Mali G2‑Ultra NX, however, comfortably hits 60 fps with the same title, delivering a four‑fold boost in both performance and efficiency. On the tooling side, Arm currently offers Unreal Engine plug‑ins and is actively working on Unity support, having recently announced a partnership with Unity China. For studios that rely on custom engines, Arm provides an SDK that streamlines integration without requiring a dedicated plug‑in. Lazarov emphasized that developers can retrofit existing games with Arm’s AI graphics in as little as three months, enabling rapid upgrades to visual fidelity or performance.

The *Neural Dawn* demo itself was built by a lean team of fewer than 20 people in under 18 months, illustrating how Arm’s technology can accelerate development cycles. "Our tech lets developers quickly and easily enhance existing games, make them more efficient, or push them in any direction they choose," Lazarov concluded. "It also empowers them to ship groundbreaking new titles at unprecedented speed."