Prototype low-power alternative to energy-guzzling GPU chips widely used in machine learning

https://www.technologyreview.com/2024/06/05/1093250/how-a-simple-circuit-could-offer-an-alternative-to-energy-intensive-gpus/

"32 variable resistors perform simple machine learning classification tasks... classified 120 items with 95% accuracy... also performs nonlinear XOR... training uses 2nd identical circuit to instruct, both start with same resistance for each... type of analog computer encoding along continuum instead of discrete... analog better energy efficiency... type of neuromorphic computer, resistors adjust in bottom-up approach, no dedicated component for memory, these represent additional energy efficiencies"

Related: Computing scheme accelerates machine learning while improving energy efficiency of traditional data operations
https://techxplore.com/news/2024-09-scheme-machine-energy-efficiency-traditional.html