J’ai débridé ma RTX 3090 : le verdict avec Ollama
Summary
The article documents unlocking RTX 3090 capabilities on Linux using the gpu_unlock module to speed up local AI inference with Ollama and Qwen3.8 27B. It covers setup steps (initramfs, GRUB, PCI address targeting) and presents an 18-test benchmark comparing performance with and without unlock, plus verification notes and open-source references.