Model and evaluate complex system-on-chip subsystems, weighing technical options and supporting silicon bring-up for AI acceleration hardware.