I use My Phone as External GPU for Local AI

Can you turn your smartphone's LPDDR5X RAM into extra eGPU for local LLM inference? In this video, I set up Galaxy S23 as a compute device for local AI inference. Install Termux, compile llama.cpp with DGGML_RPC enabled, set up ADB port forwarding over a USB 3.2 connection, and offload layers of Gemma 4 12B to see if it outperforms CPU or SSD offloading. Follow me on X: https://x.com/redStapler_twit *** My Portable AI PC Setup *** CPU: https://link.amazon/B02zcl7NO GPU1: https://link.amazon/B0be75r6Q RAM: https://link.amazon/B059ALsex Board: https://link.amazon/B00Q66vTM PSU: https://link.amaz