One of the big take aways from my work benchmarking the new generation of hardware, designed to run machine learning models at vastly…