Clustering Nvidia DGX Spark and M3 Ultra Mac Studio for 4x Faster LLM Inference | Not Hacker News!