High-performance non-ONTAP workhorse targets AI use cases and aims to ensure GPUs get fed optimally, with claimed 2.5x boost in performance over previous models