Stop wasting electricity
Run on my rtx4090 llama.cpp params: llama-server -m ~/Projects/llm/models/Qwen3.6-27B-UD-Q4_K_XL.gguf –flash-attn on -ngl all -ctk q4_0 -ctv q4_0 -t 32 -c 262144 Power limit was set using sudo nvidia-smi -pl N On my observation, GPU constantly…