gfx900

Run LLMs on AMD Vega8 / Vega10 APUs and GPUs (Ryzen 5700G / gfx90c and Vega 56/64): ROCm 7 gfx900 backport + Vulkan/RADV launchers, Docker images, benchmarks for llama.cpp and LM Studio

pytorch2.3.1 whl gfx802, gfx803 ,gfx900 (AMD RX580, FirePro s7150 x2, Vega64) GPU

Tuning llama.cpp's Vulkan backend for AMD Vega 56/64 (gfx900/GCN5): 1-bit and MoE inference kernels, multi-GPU layer-split findings, and the dead ends