Ultrasonic computation hardware for convolutional neural network computing and other computation applications

An apparatus comprising:

• at least one processing device comprising a processor coupled to a memory;
• and
• one or more ultrasonic computation engines each coupled to the at least one processing device;
• the at least one processing device being configured to implement a machine learning system;
• the one or more ultrasonic computation engines each being configured to perform computations for the machine learning system;
• wherein each of the one or more ultrasonic computation engines comprises:
• first and second ultrasonic transducer arrays arranged on opposing ends of the ultrasonic computation engine and adapted to receive respective first and second signals;
• first and second ultrasonic propagation regions arranged between the first and second ultrasonic transducer arrays and proximate respective ones of the first and second ultrasonic transducer arrays;
• and
• an intermediate computational structure arranged between the first and second ultrasonic propagation regions;
• wherein the intermediate computational structure is configured to perform at least a portion of the computations for the machine learning system, including generating an additional signal that is a function of the first and second signals.

A method comprising:

• applying first and second signals to respective first and second ultrasonic transducer arrays arranged on opposing ends of an ultrasonic computation engine, the ultrasonic computation engine further comprising first and second ultrasonic propagation regions arranged between the first and second ultrasonic transducer arrays and proximate respective ones of the first and second ultrasonic transducer arrays, and an intermediate computational structure arranged between the first and second ultrasonic propagation regions;
• propagating ultrasonic waves corresponding to respective ones of the first and second signals from respective ones of the first and second ultrasonic transducer arrays through the respective first and second ultrasonic propagation regions towards the intermediate computational structure;
• receiving in the intermediate computational structure the propagating ultrasonic waves from the respective first and second ultrasonic propagation regions;
• and
• generating in the intermediate computational structure, based at least in part on the received propagating ultrasonic waves, an additional signal that is a function of the first and second signals.