Encode inputs to reduce energy usages in analog computation acceleration

A device, comprising:

• memory cells configured to store a set of first parameters as an input for an operation of multiplication and accumulation;
• digital to analog converters;
• an analog section of the operation of multiplication and accumulation;
• analog to digital converters;
• and
• a controller configured to:
• analyze the set of first parameters to identify an encoding parameter;
• generate, using the digital to analog converters and according to the set of first parameters, analog inputs to the analog section of the operation of multiplication and accumulation, wherein the analog section is configured to generate analog outputs responsive to the analog inputs;
• and
• determine, using the analog to digital converters and according to the encoding parameter and the analog outputs, a set of second parameters as an output responsive to the input for the operation of multiplication and accumulation;
• wherein the analog section has a characteristic of consuming energy at varying levels for processing input parameters having varying magnitudes.

A non-transitory computer storage medium storing instructions which, when executed in a computing device, cause the computing device to perform a method, comprising:

• receiving a set of first parameters as an input for an operation of multiplication and accumulation;
• analyzing the set of first parameters to identify an encoding parameter;
• generating, according to the encoding parameter and the set of inputs, a set of third parameters;
• providing the set of third parameters to an analog accelerator having an analog section of the operation of multiplication and accumulation;
• receiving, from the analog accelerator, a set of fourth parameters generated by the analog accelerator in response to the set of third parameters;
• and
• generating, according to the encoding parameter and the set of fourth parameters, a set of second parameters as an output responsive to the input for the operation of multiplication and accumulation;
• wherein the analog section has a characteristic of consuming energy at varying levels for processing input parameters having varying magnitudes.