IP Library Granted Patent US 11,086,569
Granted Patent B2
US 11,086,569 · App. 16/559,230 · Granted Aug 10, 2021

Memory system and method

Inventors: Youyang Ng (Yokohama, JP); Gibeom Park (Yokohama, JP); Kazuhisa Horiuchi (Yokohama, JP); Ryo Yamaki (Yokohama, JP); Koji Horisaki (Yokohama, JP)
Assignee: TOSHIBA MEMORY CORPORATION
G06F3/0659G06F3/0604G06F3/0679G06N3/04G06N3/08G11C16/26G11C16/0483H01L27/115
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 11,086,569
App. No.
16/559,230
Granted
Aug 10, 2021
Kind
B2
Abstract

According to an embodiment, a memory system includes a plurality of memory cells and a memory controller. The memory controller executes a read operation on the plurality of memory cells using one or more reference read voltages to acquire a histogram representing the number of memory cells with respect to a threshold voltage. The memory controller inputs the histogram to a trained neural network model including an output layer for outputting one or more actual read voltages to read data from the memory cells. The memory controller executes a read operation on the memory cells using the one or more actual read voltages output from the output layer.

Claims (48)

1. A memory system comprising:

a plurality of memory cells; and

a memory controller that:

executes a read operation on a plurality of memory cells using one or more reference read voltages to acquire a histogram representing a number of memory cells with respect to a threshold voltage,

inputs the histogram to a trained neural network model, the trained neural network model comprising an output layer for outputting one or more actual read voltages to read data from the plurality of memory cells, and

executes a read operation on the plurality of memory cells using the one or more actual read voltages output from the output layer.

2. The memory system according to claim 1 , wherein

the neural network model comprising an input layer including a first node and a second node, and

the memory controller inputs the histogram to the first node and inputs property information to the second node.

3. The memory system according to claim 2 , wherein

the neural network model further comprises one or more hidden layers, and

the one or more hidden layers include a third node to be activated by a non-linear activation function.

4. The memory system according to claim 2 , wherein

the property information includes the number of executed program/erase cycles to the plurality of memory cells.

5. The memory system according to claim 2 , wherein

the property information includes the number of executed read operations to the plurality of memory cells.

6. The memory system according to claim 2 , wherein

the property information includes location information of the plurality of memory cells.

7. The memory system according to claim 2 , further comprising

a temperature sensor, wherein

the property information includes an output value of the temperature sensor.

8. The memory system according to claim 2 , wherein

the property information includes an elapsed time after a program operation is executed on the plurality of memory cells.

9. The memory system according to claim 1 , wherein

the number of the reference read voltages is smaller than the number of the actual read voltages.

10. A method for controlling a non-volatile memory comprising a plurality of memory cells, the method comprising:

executing a read operation on the plurality of memory cells using one or more reference read voltages;

acquiring, by the read operation, a histogram representing a number of memory cells with respect to a threshold voltage;

inputting the histogram to a trained neural network model, the trained neural network model comprising an output layer for outputting one or more actual read voltages to read data from the plurality of memory cells; and

executing a read operation on the plurality of memory cells using the one or more actual read voltages output from the output layer.

11. The method according to claim 10 , further comprising:

acquiring property information; and

inputting the properly information to the trained neural network model.

12. The method according to claim 11 , wherein

the neural network model includes one or more hidden layers, and

the one or more hidden layers include a node to be activated by a non-linear activation function.

13. The method according to claim 11 , wherein

the property information includes the number of executed program/erase cycles to the plurality of memory cells.

14. The method according to claim 11 , wherein

the property information includes the number of executed read operations to the plurality of memory cells.

15. The method according to claim 11 , wherein

the property information includes location information of the plurality of memory cells.

16. The method according to claim 11 , wherein

the property information includes an output value of a temperature sensor.

17. The method according to claim 11 , wherein

the property information includes an elapsed time after a program operation is executed on the plurality of memory cells.

18. The method according to claim 10 , wherein

the number of the reference read voltages is smaller than the number of the actual read voltages.

Assignments (2)
CHANGE OF NAME AND ADDRESS Recorded Jan 31, 2022
From: TOSHIBA MEMORY CORPORATION
To: KIOXIA CORPORATION
Reel/Frame 058905/0582 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 3, 2019
From: NG, YOUYANG; PARK, GIBEOM; HORIUCHI, KAZUHISA; YAMAKI, RYO; HORISAKI, KOJI
To: TOSHIBA MEMORY CORPORATION
Reel/Frame 050252/0181 →
Cited By (1)
US 12,656,974