Blogs

Arm enables robotics training at scale with Isaac Sim, supporting full AI workflows from build to deployment on a single DGX Spark system

How Arm is unleashing rack-scale performance and efficiency for the agentic AI data center

This release delivers a major upgrade to Vulkan capture, introducing a new capture pipeline built on GFXReconstruct

Nanite demonstrates that GPU-driven rendering works on mobile. Learn how Mali GPUs handle complex geometry and modern pipelines

Migrating to Arm on Google Cloud helped SoldOutAfrica improve performance, reduce costs, and increase event revenue

The latest collaboration between Arm and Anglia Ruskin University Cambridge challenged students to develop industry-relevant projects – and left room for innovations of their own

Graviton and SoC Migration Powers bring Arm development expertise inside Kiro

深度解析基于 CPU 的向量化、统一内存与本地检索工作流如何协同发力，依托 Arm 及 DGX Spark，在桌面级 AI 平台上构建响应迅速、隐私可控的 RAG 管线

Arm showcases Neural Frame Rate Upscaling (NFRU) at GDC, a neural graphics technique that generates frames to improve smoothness in mobile games

Learn how to migrate from Picolibc to LLVM-libc in ATfE, update linker scripts, fix semihosting, and prepare for the new default C library

Arm and Ferrous Systems bring Rust to Cortex-R82AE, enabling memory-safe, high-performance real-time systems for physical AI

Imagine typing “a blue ceramic mug” into your phone, and a few seconds later your newly generated 3D mug spawns in your game

How Arm CPUs and open source enable real-time, privacy-preserving voice conversations with faster-whisper and vLLM

Deploy HIPAA-compliant AI fully offline on Android with Armv9 SME2, achieving fast, accurate clinical summaries in under 10 seconds

Learn how Vulkan barriers map to Mali GPUs, how CSF streams interact, and how to optimize synchronization for better GPU utilization.

Zephyr RTOS gains Armv9-A enablement from Arm and BayLibre, unlocking Cortex-A320 and SVE support for efficient, real-time Edge AI deployments.

Learn how ExecuTorch and Arm Scalable Matrix Extension 2 are unlocking faster, more responsive on-device AI for mobile experiences like interactive image segmentation.

Cross-NUMA memory access can limit llama.cpp scaling on Neoverse. This blog post shows two NUMA-aware optimizations delivering up to 55% faster text generation.

End-to-end guide to training, quantizing, and deploying an int8 Conformer speech model on Arm, using ExecuTorch and Ethos-U85 for efficient edge AI today.

Learn how LLVM sanitizers in the Arm Toolchain help detect undefined behavior, improve code quality, and expose hidden bugs in embedded C and C++ projects.

Partner Technical Training is accelerating to meet the increasing on demand training expectations of partners and the growing size of the training audience. AI is playing a key role in making this happen