Performing XNOR equivalent operations by adjusting column thresholds of a compute-in-memory array

1. An apparatus comprising:

a compute-in-memory array comprising rows and columns, the compute-in-memory array configured:

to adjust an activation threshold generated for each column of the compute-in- memory array based on a function of a weight value and an activation value; and

to calculate a conversion bias current reference based on an input value from an input vector to the compute-in-memory array, the compute-in-memory array programmed with a set of weight values, in which the adjusted activation threshold and the conversion bias current reference are used as a threshold for determining output values of the compute-in-memory array.

2. The apparatus of claim 1 , further comprising a comparator configured to compare a bit line population count to a sum of the conversion bias current reference and the adjusted activation threshold in order to determine an output of a bit line.

3. The apparatus of claim 1 , in which an artificial neural network including the compute-in-memory array comprises a binary neural network.

4. The apparatus of claim 1 , in which the activation threshold is less than half of a number of rows of the compute-in-memory array, the number of rows corresponding to a size of the input vector.

5. The apparatus of claim 1 , in which the conversion bias current reference is less than half of a number of rows of the compute-in-memory array, the number of rows corresponding to a size of the input vector.

6. A processor-implemented method comprising:

adjusting an activation threshold generated for each column of the compute-in-memory array having rows and columns based on a function of a weight value and an activation value;

calculating a conversion bias current reference based on an input value from an input vector to the compute-in-memory array, the compute-in-memory array being programmed with a set of weight values, in which the adjusted activation threshold and the conversion bias current reference are used as a threshold for determining output values of the compute-in-memory array.

7. The processor-implemented method of claim 6 , further comprising comparing a bit line population count to a sum of the conversion bias current reference and the adjusted activation threshold in order to determine an output of a bit line.

8. The processor-implemented method of claim 6 , in which an artificial neural network including the compute-in-memory array comprises a binary neural network.

9. The processor-implemented method of claim 6 , in which the activation threshold is less than half of a number of rows of the compute-in-memory array, the number of rows corresponding to a size of the input vector.

10. The processor-implemented method of claim 6 , in which the conversion bias current reference is less than half of a number of rows of the compute-in-memory array, the number of rows corresponding to a size of the input vector.

11. A non-transitory computer-readable medium having program code recorded thereon, the program code being executed by a processor and comprising:

program code to adjust an activation threshold generated for each column of a compute-in-memory array having rows and columns based on a function of a weight value and an activation value; and

program code to calculate a conversion bias current reference based on an input value from an input vector to the compute-in-memory array, the compute-in-memory array being programmed with a set of weight values, in which the adjusted activation threshold and the conversion bias current reference are used as a threshold for determining output values of the compute-in-memory array.

12. The non-transitory computer-readable medium of claim 11 , further comprising program code to compare a bit line population count to a sum of the conversion bias current reference and the adjusted activation threshold in order to determine an output of a bit line.

13. The non-transitory computer-readable medium of claim 11 , in which an artificial neural network subject to the adjusting and the calculating comprises a binary neural network.

14. The non-transitory computer-readable medium of claim 11 , in which the activation threshold is less than half of a number of rows of the compute-in-memory array, the number of rows corresponding to a size of the input vector.

15. The non-transitory computer-readable medium of claim 11 , in which the conversion bias current reference is less than half of a number of rows of the compute-in-memory array, the number of rows corresponding to a size of the input vector.

16. An apparatus comprising:

means for adjusting an activation threshold generated for each column of the compute-in- memory array having rows and columns based on a function of a weight value and an activation value; and

means for calculating a conversion bias current reference based on an input value from an input vector to the compute-in-memory array, the compute-in-memory array being programmed with a set of weight values, in which the adjusted activation threshold and the conversion bias current reference are used as a threshold for determining output values of the compute-in-memory array.

17. The apparatus of claim 16 , further comprising means for comparing a bit line population count to a sum of the conversion bias current reference and the adjusted activation threshold in order to determine an output of a bit line.

18. The apparatus of claim 16 , in which an artificial neural network including the compute-in-memory array comprises a binary neural network.

19. The apparatus of claim 16 , in which the activation threshold is less than half of a number of rows of the compute-in-memory array, the number of rows corresponding to a size of the input vector.

20. The apparatus of claim 16 , in which the conversion bias current reference is less than half of a number of rows of the compute-in-memory array, the number of rows corresponding to a size of the input vector.