Custom mass multiplication circuits

1. A mass multiplier operating as a part of an integrated circuit implementing a convoluted neural network (CNN), comprising:

a port receiving a stream of discrete data values column by column and line by line from a matrix of data values;

first hardware circuitry multiplying each data value as received by a plurality of weight values of a specific aperture function simultaneously, creating a set of products;

second hardware circuitry performing a subfunction of the specific aperture function for each data value as received, using individual ones of the set of products, producing partial output values for the aperture function; and

third hardware circuitry retaining the partial output values and providing the partial output values back to the first hardware circuitry as needed to produce complete output values;

wherein input values are received in a continuous stream and output values are produced in a continuous stream.

2. The mass multiplier operating as a part of an integrated circuit implementing a convoluted neural network (CNN) of claim 1 wherein the discrete values received are unsigned binary of fixed width, the weight values are unsigned binary of fixed width of two or more bits, and each multiple is composed as a summation of bit-shifted duplicates of the input.

3. The mass multiplier operating as a part of an integrated circuit implementing a convoluted neural network (CNN) of claim 2 , wherein the set of shifted duplicates is increased to allow the use of subtraction operations to reduce or otherwise optimize the circuit.

4. The mass multiplier operating as a part of an integrated circuit implementing a convoluted neural network (CNN) of claim 3 wherein unused outputs of the set of shifted duplicates may be omitted either explicitly or implicitly.

5. The mass multiplier operating as a part of an integrated circuit implementing a convoluted neural network (CNN) of claim 1 wherein the set of products is produced by combinatorial logic.

6. The mass multiplier operating as a part of an integrated circuit implementing a convoluted neural network (CNN) of claim 1 wherein the set of products is produced by a single stage pipeline, using single or multiple clock cycles.

7. The mass multiplier operating as a part of an integrated circuit implementing a convoluted neural network (CNN) of claim 1 wherein the set of products is produced by a multi-stage pipeline by combining no more than two addends per stage.

8. The mass multiplier operating as a part of an integrated circuit implementing a convoluted neural network (CNN) of claim 7 wherein unused elements of intermediate sub-compositions are eliminated from the circuit either explicitly or implicitly.

9. The mass multiplier operating as a part of an integrated circuit implementing a convoluted neural network (CNN) of claim 1 wherein the set of products is produced by a multi-stage pipeline by combining three or more addends per stage and the sub-compositions are adjusted accordingly.