Since this article is very light on details-

It’s using Huawei LX2 ARMv9 cores, 304 cores @1.5ghz 690W (with 1x? SVE2 unit and an SME unit, nice) with two seats per node, originally it was to have 20,480 nodes in 90 cabinets for 40,960 CPUs (12,451,840 cores…) but the TOP500 listing has more cores (13,789,440 - 4,400 more) than some of the original documentation, so they must have added a few cabinets.

Pretty neat, brute-forcing it with CPUs makes for an infinitely more useful machine than GPU compute, despite the higher power cost. (42MW vs. 29MW for El Capitan)

I’ve heard it said that when you have the fastest supercomputer, you have the advantage of running computations and simulations that are only accessible by future hardware. The country that has the fastest supercomputer is therefore the most advanced. They have a machine that can reach into the future and do what everyone else cannot.

Maybe Orange pedophile trump should cutaway more from science, inbred man followed by all inbreeding morons, is your family tree a stick if so you’re a perfect candidate to be a maga gop