Neural network array comprising one or more coarse cells and one or more fine cells
In one example, a system comprises a neural network array of non-volatile memory cells arranged in rows and columns; and a logical cell comprising a first plurality of non-volatile memory cells in a first row of the array and a second plurality of non-volatile memory cells in a second row adjacent to the first row; wherein the first plurality of non-volatile memory cells and the second plurality of non-volatile memory cells are configured as one or more coarse cells and one or more fine cells.
1. A system, comprising:
a neural network array comprising non-volatile memory cells arranged in rows and columns, wherein the neural network array is part of a neural network; and
a logical cell comprising a first plurality of non-volatile memory cells in a first row of the neural network array and a second plurality of non-volatile memory cells in a second row adjacent to the first row;
wherein the first plurality of non-volatile memory cells and the second plurality of non-volatile memory cells are configured as one or more coarse cells and one or more fine cells.
2. The system of claim 1 , wherein during a programming operation of the logical cell, a coarse programming method is used to program the coarse cells and fine programming method is used to program the fine cells.
3. The system of claim 1 , further comprising a tuning cell.
4. The system of claim 3 , wherein the tuning cell is used to tune an adjacent cell.
5. The system of claim 1 , wherein one or more of the first plurality and the second plurality comprise one or more tuning cells, and wherein during a programming operation of the logical cell, a tuning method is used to program the one or more tuning cells.
6. The system of claim 1 , wherein the non-volatile memory cells are split-gate flash memory cells.
7. The system of claim 1 , wherein the non-volatile memory cells are stacked-gate flash memory cells.
8. The system of claim 1 , wherein the neural network is an analog neural network.
9. A method comprising:
programming a logical cell comprising a first plurality of non-volatile memory cells in a first row of an array of non-volatile memory cells and a second plurality of non-volatile memory cells in a second row adjacent to the first row, where the first plurality and the second plurality are configured as one or more coarse cells and one or more fine cells, wherein the array of non-volatile memory cells is part of a neural network, the programming comprising:
programming the one or more coarse cells using a coarse programming method;
programming the one or more fine cells using a fine programming method.
10. The method of claim 9 , further comprising:
verifying a value programmed in the logical cell.
11. The method of claim 9 , wherein the logical cell further comprises one or more tuning cells in a same row, and the method further comprises programming the one or more tuning cells using a tuning method.
12. The method of claim 11 , further comprising:
verifying a value programmed in the logical cell.
13. The method of claim 9 , wherein the non-volatile memory cells are split-gate flash memory cells.
14. The method of claim 9 , wherein the non-volatile memory cells are stacked-gate flash memory cells.
15. The method of claim 9 , wherein the neural network is an analog neural network.