In a traditional always-on edge system, data relevance can only be determined after digitization. Since system power consumption is dominated by the ADC and digital processors, a
digitize-first architecture is grossly inefficient and wastes significant power analyzing data that will simply be thrown away.
An analogML core eliminates this inefficiency by bringing near-zero-power inferencing into the analog domain in an
analyze-first architecture. Data relevance is determined prior to digitization, allowing the higher-power digital system to remain off unless important data are detected.