Data storage device and method for predicting future read thresholds
Before a read threshold is needed to read a wordline in memory, a data storage device can infer a plurality of read thresholds based on possible conditions of the memory that may exist when the read threshold is eventually needed. When the read threshold is needed, it is selected from the previously-inferred read thresholds based on the current conditions of the memory. This can improve latency and throughput, improve quality of service, reduce power consumption, and reduce errors.
1. A data storage device comprising:
a memory; and
a processor coupled with the memory and configured to:
before receiving a command to read a wordline in the memory, infer a plurality of read thresholds from a respective plurality of sets of parameter values and conditions of the memory that may exist when the command to read the wordline in the memory is received; and
in response to receiving the command to read the wordline in the memory:
determine a present set of parameter values and conditions of the memory; and
select a read threshold from the plurality of read thresholds based on the present set of parameter values and conditions of the memory.
2. The data storage device of claim 1 , wherein the processor is further configured to infer the plurality of read thresholds during an idle time.
3. The data storage device of claim 1 , wherein the processor is further configured to infer the plurality of read thresholds in parallel with a memory operation.
4. The data storage device of claim 1 , wherein the processor is further configured to infer the plurality of read thresholds using an inference engine with a large machine-learning model.
5. The data storage device of claim 4 , wherein the inference engine is implemented as firmware in a controller that comprises the processor.
6. The data storage device of claim 1 , wherein at least one parameter value in the plurality of sets of parameter values is fixed when inferring the plurality of read thresholds.
7. The data storage device of claim 6 , wherein the at least one parameter value comprises program temperature.
8. The data storage device of claim 6 , wherein the at least one parameter value comprises wordline number.
9. The data storage device of claim 1 , wherein the parameter values of the memory comprise: program/read temperature, program-erase-cycle count, and/or failed bit count.
10. The data storage device of claim 1 , wherein the parameter values of the memory comprise time and temperature group information, temperature information, bit error rate information, program-erase count (PEC) information, physical page location information, a read threshold on a representative wordline, a data retention level of a block, and/or a state-by-state cell-voltage-distribution measurement.
11. The data storage device of claim 1 , wherein the memory comprises a three-dimensional memory.
12. In a data storage device comprising a memory, a method comprising:
generating a plurality of inference exploration grids;
selecting an inference exploration grid from the plurality of inference exploration grids;
inferring a plurality of read thresholds from the selected inference exploration grid based on a plurality of possible memory conditions; and
prior to reading a wordline in the memory, selecting a set of read thresholds from the plurality of read thresholds based on a current memory condition.
13. The method of claim 12 , wherein the generating the plurality of inference exploration grids is scheduled after a time and temperature group information update is conducted.
14. The method of claim 12 , wherein a plurality of granular operations are used to infer the plurality of read thresholds.
15. The method of claim 12 , wherein a single atomic operation is used to infer the plurality of read thresholds.
16. The method of claim 12 , further comprising:
applying interpolation and/or extrapolation between grid of possible memory conditions and/or between recorded read thresholds.
17. The method of claim 12 , further comprising:
gathering a plurality of tuples of possible memory conditions; and
unifying the plurality of tuples.
18. The method of claim 12 , wherein the unifying comprises clustering possible memory conditions to a limited number of groups, wherein each group is related to a read threshold.
19. The method of claim 12 , wherein the memory conditions comprise one or more of the following: time and temperature group information, temperature information, bit error rate information, program-erase count (PEC) information, physical page location information, a read threshold on a representative wordline, a data retention level of a block, and a state-by-state cell-voltage-distribution measurement.
20. A storage system comprising:
a memory; and
means for:
before receiving a command to read a wordline in the memory, inferring a plurality of read thresholds from a respective plurality of sets of parameter values and conditions of the memory that may exist when the command to read the wordline in the memory is received; and
in response to receiving the command to read the wordline in the memory:
determining a present set of parameter values and conditions of the memory; and
selecting a read threshold from the plurality of read thresholds based on the present set of parameter values and conditions of the memory.