I have spent unreasonable amounts of time building llama.cpp for my hardware setup (AMD GPU) on both Windows and Linux. That was one of the main reasons of writing that blog post for me. Lmao.