User benchmarks Qwen 3.6 27B on dual Radeon R9700 with llama.cpp

User shares multi-GPU setup testing Qwen 3.6 27B Q8 MTP on llama.cpp with 2× Radeon R9700. Build uses ThinkStation P7, Xeon w7-3455, and 128 GB RAM, with detailed server log results.