Adaptive high voltage circuitry and methods for programming operations in an analog neural memory array in a deep learning artificial neural network
Numerous embodiments are disclosed for a high voltage generation algorithm and system for generating high voltages necessary for a particular programming operation in analog neural memory used in a deep learning artificial neural network. Different calibration algorithms and systems are also disclosed. The system can modify a high voltage signal applied to an array of cells during a programming operation as the number of cells being programmed changes.
1. A method for programming selected non-volatile memory cells in an array of non-volatile memory cells, the method comprising:
performing mass programming of all memory cells in the array;
performing a soft erase of all memory cells in the array;
performing a hard program of unselected memory cells in the array;
performing a soft program of selected memory cells in the array, the soft program operation comprising a coarse program operation and a precision program operation, wherein the precision program operation comprises a compensation scheme that modifies a high voltage signal applied to the selected memory cells based on the number of cells that are selected memory cells.
2. The method of claim 1 , wherein the compensation scheme modifies the magnitude of the high voltage signal based on the number of cells that are selected memory cells.
3. The method of claim 1 , wherein the high voltage signal comprises a high voltage pulse and the compensation scheme modifies the duration of the high voltage pulse based on the number of cells that are selected memory cells.
4. The method of claim 1 , wherein the high voltage signal comprises multiple high voltage pulses and the compensation scheme modifies the total duration of the pulses based on the number of cells that are selected memory cells.
5. The method of claim 1 , wherein the soft erase step erases all memory cells in the array to an erased level of 1-5 μA.
6. The method of claim 1 , wherein the coarse program operation programs some or all of the selected cells to a programmed level of 0.1-1.5 μA.
7. The method of claim 1 , wherein the precision program operation programs some or all of the selected cells to a programmed level of 1 pA-20 nA.
8. The method of claim 1 , wherein the memory cells are split-gate flash memory cells.
9. The method of claim 1 , wherein the array is a vector-by-matrix multiplication array.
10. The method of claim 2 , wherein the array is a vector-by-matrix multiplication array.
11. The method of claim 3 , wherein the array is a vector-by-matrix multiplication array.
12. The method of claim 4 , wherein the array is a vector-by-matrix multiplication array.
13. The method of claim 5 , wherein the array is a vector-by-matrix multiplication array.
14. The method of claim 6 , wherein the array is a vector-by-matrix multiplication array.
15. The method of claim 7 , wherein the array is a vector-by-matrix multiplication array.
16. The method of claim 8 , wherein the array is a vector-by-matrix multiplication array.