Native ARM64 launcher for Tracker physics video analysis on Snapdragon X
A Windows-on-ARM optimized version of the Tracker physics video analysis tool, created using Claude, that runs natively on ARM64 processors like Qualcomm Snapdragon X. It replaces x64 emulation with a native ARM Java runtime, improving automatic tracking efficiency roughly threefold and making the tool fully usable on ARM-based laptops.
Visit project ↗Entities
Related
Local image-to-3D model for Apple Silicon and iPhone
A port of Hunyuan3D-Paint and Hunyuan3D-Shape to Swift MLX and Python MLX, enabling image-to-3D generation on Apple Silicon Macs and iPhones with low memory usage. It solves the problem of running 3D generation locally without cloud dependencies or high-end GPUs.
Reame: CPU-first LLM inference server built on llama.cpp released
Reame is a new LLM inference server designed to run efficiently on cheap CPU hardware, including shared vCPUs, free tiers, and 2-core ARM boxes. Built on llama.cpp, it features disk KV cache, self-regulating speculation, generation archive, and interleaved multi-user support. The project emphasizes treating CPU hardware as a first-class citizen rather than a fallback.
On-device Android AI agent with MCP server for UI control
Aura is an on-device AI agent for Android that can interact with the phone's UI — tapping, typing, scrolling, launching apps, and navigating — using its own on-device MCP server. It uses LLMs to interpret commands and execute actions via MCP tools. The MCP server is also accessible to external clients like Claude Code, enabling remote control of the Android device. This solves the problem of automating phone tasks without cloud dependency.
User tests LLMs on Intel 285HX CPU-only mini-PC, finds Llama-Swap incompatible with SYCL
A user set up a homelab server on an MS-02 mini-PC with an Intel Core Ultra 285HX and 64GB RAM, testing Qwen3, Qwen3.6, and Gemma4 via Llama.cpp. They found Llama-Swap's quick swapping helpful but incompatible with SYCL, complicating testing. The post shares early impressions of running LLMs on a CPU-only system.
LLM hardware recipe database with filters and community usage tracking
A community-driven database that lists which LLM models run on which hardware, with performance details. Users can filter by hardware, submit new recipes, and mark which recipes they actively use to show popularity. It solves the problem of finding compatible model-hardware combinations for LLM deployment.