Three months wrong about why my 4-node AMD cluster was slow

Three months on a 4-node Minisforum MS-S1 Max Strix Halo cluster, and the one assumption that was quietly killing my inference speed. Try out ChatLLM - http://chatllm.abacus.ai/ltf and Abacus AI DeepAgent - http://deepagent.abacus.ai/ltf 🛒 Gear Links 🛒 👀 2 400Gbps switch: https://bhpho.to/4r9vJi1 👀👀 4 400Gbps switch: https://bhpho.to/4qQqOlI 💻☕ Thunderbolt 5 external SSD: https://amzn.to/3XqetZO 💻☕ Favorite 15" display with magnet: https://amzn.to/3zD1DhQ 🎧⚡ Great 40Gbps T4 enclosure: https://amzn.to/3JNwBGW 🛠️🚀 My nvme ssd: https://amzn.to/3YLEySo 📦🎮 My gear: https://www.amazon.com/shop/alexziskind 🎥 Related Videos 🎥 🏆 Skip M3 Ultra & RTX 5090 for LLMs | NEW 96GB KING -    • Skip M3 Ultra & RTX 5090 for LLMs | NEW 96...   💻 Smallest RTX Pro 6000 rig | OVERKILL -    • Going up against Apple's Ultra   🔧 Cheap mini runs a 70B LLM 🤯 -    • Cheap mini runs a 70B LLM 🤯   🌙 RAM torture test on Mac -    • TRUTH about RAM vs SSD UPGRADES for MacBook   🚀 FREE Local LLMs on Apple Silicon | FAST! -    • FREE Local LLMs on Apple Silicon | FAST!   🪞 REALITY vs Apple’s Memory Claims | vs RTX4090m -    • REALITY vs Apple’s Memory Claims | vs RTX4...   📦 Set up Conda -    • python environment setup on Apple Silicon ...   🤖 INSANE Machine Learning on Neural Engine -    • INSANE Machine Learning on Neural Engine |...   🛠️ Developer productivity Playlist -    • Developer Productivity   🔗 AI for Coding Playlist: 📚 -    • AI   — — — — — — — — — ❤️ SUBSCRIBE TO MY YOUTUBE CHANNEL 📺 Click here to subscribe:    / @azisk   — — — — — — — — — Join this channel to get access to perks:    / @azisk   — — — — — — — — — 📱 ALEX ON X:   / digitalix   Donato's channel:    / @donatocapitella   #macmini #llm #nvidia