Developer

In this session, we will demonstrate how KleidiAI enables optimal performance for some of the world’s most advanced language models on Arm CPUs, and dive deeper into Arm Scalable Matrix Extension (SME)  and how developers can leverage SME capabilities to accelerate AI applications.

This session focuses on solving the challenges of deploying large language models (LLMs) on edge devices, using Tencent’s Hunyuan model as a core case study. We will demonstrate how to efficiently deploy it on Arm-based edge devices with the lightweight inference framework llama.cpp. Key techniques such as inference performance tuning and hardware resource adaptation will be dissected in detail, providing developers with a proven, practical solution for bringing AI models to Arm edge devices.

Arm and Stability AI have collaborated to bring cutting-edge AI audio generation directly onto Arm CPUs. In this session, we’ll show how the Stable Audio Open Small model has been fully optimized to run on smartphones with the LiteRT runtime, powered by XNNPack and KleidiAI. We’ll also share our latest advancements leveraging SME2 to achieve even greater performance, making high-quality audio generation accessible to all developers.

We are relentless in our support of innovation in real-time graphics. We will build upon the story begun with Arm Accuracy Super Resolution as a modern day best in class solution for upscaling to then share our vision for neural graphics on Arm GPUs. Attendees will take away an understanding of how Arm will deliver practical, performant neural graphics technologies for mobile graphics and spur a new wave of innovation.

This session explores the growing Windows on Arm ecosystem with a focus on performance analysis and optimization. We’ll introduce the open-source toolkits such as WindowsPerf and demonstrate how they can be used to profile, analyze, and fine-tune applications for Arm-based Windows devices. Attendees will gain insights into leveraging these tools effectively, discover best practices for debugging and performance tuning, and leave with practical knowledge to enhance application efficiency and reliability on the Windows on Arm platform.

Migrate-ease is an innovative project designed to analyze codebases specifically for x86_64 architectures and offers tailored suggestions aimed at facilitating the migration process to aarch64. This tool streamlines the transition, ensuring a smooth and efficient evolution of your software to leverage the benefits of aarch64 architecture. In this talk, we will demonstrate how developers can use migrate-ease to simplify their migration to Arm based servers.

In this session, we will introduce SOAFEE Blueprint and showcase how developers in Auto industry can leverage SOAFEE to build advanced and secure auto application.

The age of high-performance multi-core Arm SoCs is transforming industries from autonomous driving and robotics to embodied intelligence. Yet real-time performance and functional safety cannot be solved by hardware power alone. Join Wind River’s Jingrui Li as he reveals proven strategies to unlock the full potential of multi-core Arm platforms. Through practical examples with VxWorks RTOS, Wind River Hypervisor, and Wind River Linux, this session will showcase how to balance real-time responsiveness with safety requirements to accelerate innovation with confidence.