Arm Unveils Mali G2-Ultra NX GPU With Neural Accelerators for Mobile Gaming

Arm on September 8, 2026 introduced the Mali G2-Ultra NX, its first AI-native mobile GPU, integrating dedicated neural accelerators directly into the graphics pipeline as part of the new Arm CSS for Mobile 2 compute platform. The company detailed the hardware and its supporting developer tools in a pair of technical posts timed to its Arm Everywhere China event in Shanghai.

The platform pairs the Mali G2-Ultra NX with a new Arm C2 CPU cluster built around the C2-Ultra and C2-Pro cores and two Arm Scalable Matrix Extension 2 (SME2) units. Stefan Rosinger, Head of Product for Edge AI at Arm, wrote that the design brings CPU, GPU, system IP, software and developer tools together as one system, targeting the power and thermal limits of mobile devices.

Neural Graphics On The GPU

The Mali G2-Ultra NX places neural accelerators inside the shader cores so that neural graphics workloads run alongside conventional graphics and compute work, reusing the GPU memory system, coherent caches and control structures. Arm said the integration cuts data movement across the pipeline.

Three neural technologies ship with the GPU. Neural Super Sampling (NSS) reconstructs higher-resolution images from lower-resolution renders. Neural Frame Rate Upscaling (NFRU) generates intermediate frames to raise frame rates. Neural Super Sampling and Denoising (NSSD) combines upscaling with denoising for ray-traced scenes with complex lighting.

Deyan Lazarov, Senior Product Manager at Arm, wrote in the Mali G2-Ultra NX deep-dive that Neural Dawn, the game Arm co-developed with Sumo Digital, integrates NFRU and NSSD in a production pipeline. Arm reported the techniques delivered up to 4x higher performance efficiency and up to 70 percent lower external memory traffic compared with native rendering, gains it said can make Unreal Engine MegaLights practical on mobile.

New Execution Engine And Ray Tracing

Alongside neural acceleration, the GPU introduces a new execution engine, which Arm described as the biggest Mali GPU instruction set architecture upgrade in seven generations, with up to 2x more registers per warp. Arm said the architectural changes yield up to 24 percent higher benchmark performance and 14 percent higher non-AI gaming performance over the previous generation, and that pairing them with NFRU supports gameplay at up to 120 FPS.

The third-generation ray tracing unit delivers up to 13 percent lower DRAM traffic on leading ray tracing benchmarks, according to Arm. Hardware support for Opacity Micromaps handles transparent geometry more efficiently; in one Arm demo, the company said the feature raised frame rates by 30 percent and cut the ray tracing workload by up to 70 percent.

Developer Tools And Ecosystem

Arm said it began preparing the software ecosystem two years before the hardware through Arm Neural Technology and an open neural graphics development kit. The Arm Neural Graphics Development Kit now bundles NSS and NFRU with machine learning extensions for Vulkan and plug-ins for engines including Unreal Engine, while an SDK covers custom-engine integration alongside profiling and model-optimization tools.

Keli Zhou, Engine Lead for Where Winds Meet, said the studio is integrating Arm Neural Technology into its Messiah Engine and that the game will be among the first to bring the technology to players. Arm is also working with Tencent Games and Unity China on neural graphics integration.

The CSS For Mobile 2 Platform

On the CPU side, Arm said the C2-Ultra delivers up to 1.7x performance improvement across recent AI models, up to 15 percent higher single-thread performance, 15 percent faster web browsing, 12 percent faster application launch, and 12 percent higher multi-thread performance at the cluster level. Across an end-to-end agentic workflow spanning speech processing, memory retrieval, reasoning, application execution and web browsing, Arm said the C2-Ultra with two SME2 units completed the work 24 percent faster than the prior generation.

CSS for Mobile 2 connects the compute engines through the SI L2 system interconnect, which Arm said provides lower-latency access and higher bandwidth across the platform. The platform is supported by Arm KleidiAI, framework integrations and the new Arm AI Portal, giving the more than 22 million developers building on Arm access to optimized models and deployment resources.

Arm said more than 14 billion Mali GPUs have shipped to date. The Mali G2-Ultra NX, C2 CPU cluster and CSS for Mobile 2 platform are positioned for the next generation of mobile devices, with silicon partners able to configure the components independently or within a broader CSS configuration.