NASA engineer Craig Hunter has recently benchmarked Apple's newest system and newest CPU for CFD workloads. The M1 Ultra made headlines as an innovative system-on-a-chip promising unprecedented levels of performance and capabilities.
M1 Ultra can be configured with up to 128GB of high-bandwidth, low-latency unified memory that can be accessed by the 20-core CPU, 64-core GPU, and 32-core Neural Engine.
Apple was kind enough to lend me a Mac Studio with 20-core M1 Ultra CPU (16 performance cores / 4 efficiency cores), 64-core GPU, 128GB unified memory, and an 8TB SSD. As spec’d, this is a $7999 system with all the bells and whistles, but a more modest $3999 system would lead to the same performance conclusions in this review.
Benchmarking involved the parallel performance of the NASA USM3D CFD solver as it computes flow over a classic NACA 0012 airfoil section at low speed conditions. Typically machines get less and less efficient as more and more cores join the computation, but the outcome was drastically different for the M1 Ultra.