Linus and Alex unpack a high-end liquid-cooled workstation, the Comino Grando RM-M, which arrived in massive wooden crates requiring power tools and a crowbar to crack open around 01:10. Inside, they find a densely packed 4U system designed for scientific research and AI workloads. Once they strip the panels off at 06:15, the engineering inside really shines: a completely custom liquid cooling loop feeding dual Intel Xeon Platinum CPUs and four liquid-cooled NVIDIA GPUs via quick-disconnect hoses connected to a central Delrin distribution manifold and a massive front radiator.
Naturally, it wouldn't be an LTT video without some technical hiccup—Linus manages to blue-screen the rig around 09:42 just by standing near it, prompting Jake to joke about his curse. Once they get it running, the benchmark numbers are outrageous. The dual Xeon Platinum 8368Q processors (76 cores and 152 threads total) pull over 71,000 points in Cinebench R23 while barely breaking a sweat at 55°C. The GPU power is even crazier: rendering the Blender BMW scene at 14:10 takes under eight seconds, and firing up GPU stress tests drives the total power consumption past 2,000 Watts across multiple wall meters.
What's most impressive about this build isn't just the sheer computing horsepower, but how quietly and efficiently the liquid cooling manages thermal dissipation in a chassis that can sit on a desktop rather than locked in a noisy server rack. The pre-filled, quick-disconnect GPU water blocks are a brilliant touch for enterprise hardware maintenance, even if modern GPU drivers still struggle to keep 4-way NVLink setups completely stable during extended synthetic testing.