Group history read method to improve quality of service of solid state drives

1 . A method for reading data from a memory, comprising:

reading data from a memory group of the memory;

calculating a group divergence metric for read biases used to read the data from the memory group;

determining that the group divergence metric meets a threshold;

when it is determined that the group divergence metric meets the threshold, partitioning the memory group into a first sub-group and a second sub-group, wherein the first sub-group has a lower error rate than the second sub-group;

reading data from the first and second sub-groups;

calculating respective sub-group divergence metrics for the read biases used to read the data from the first and second sub-groups; and

determining that one of the sub-group divergence metrics meets the threshold;

when it is determined that one of the sub-group divergence metrics meets the threshold, partitioning the one sub-group of the first and second sub-groups which met the threshold into a third sub-group and a fourth sub-group.

2 . The method of claim 1 , wherein the divergence metric is function of at least one of a maximum failed bit count (maxFBC) and a minimum failed bit count (minFBC).

3 . The method of claim 2 , wherein the divergence metric is calculated from 2(maxFBC−minFBC)/(maxFBC+minFBC)).

4 . The method of claim 2 , wherein the partitioning the memory is based on the maxFBC being less than an error correction capability.

5 . The method of claim 4 , wherein

the first sub-group includes a first page of the data read from the memory group that is associated with the minFBC, and

the second sub-group includes a second page of the data read from the memory group that is associated with the maxFBC.

6 . The method of claim 4 , further comprising:

providing first and second queues, wherein

the first queue indexes all pages of data that have a failed bit count (FBC) less than 30% of the error correction capability, for partitioning into the first sub-group, and

the second queue indexes all pages of data that have the FBC larger than 60% of the error correction capability, for partitioning into the second sub-group.

7 . The method of claim 1 , further comprising merging together any sub-groups including at least the first and second sub-groups to form a set of two or more multiple groups which all have similar optimal read thresholds and which all have maxFBCs less than 70% of an error correction capability.

8 . The method of claim 1 , wherein

the memory group comprises one of a word line group, a page group, a block, a super block, and a memory die, and

the first and second sub-groups comprise partitions from one of the word line group, the page group, the block, the super block, and the memory die.

9 . The method of claim 1 , further comprising placing cold data blocks into a fourth sub-group.

10 . The method of claim 1 , further comprising storing optimal read thresholds for the first and second sub-groups into a single level cell page.

11 . A memory system comprising:

a memory device; and

a controller in communication with and configured to control the memory device, wherein the controller is configured to:

read data from a memory group of the memory device;

calculate a group divergence metric for read biases used to read the data from the memory group;

when the group divergence metric meets a threshold, partition the memory group into a first sub-group and a second sub-group, wherein the first sub-group has a lower error rate than the second sub-group;

read data from the first and second sub-groups;

calculate respective sub-group divergence metrics for the read biases used to read the data from the first and second sub-groups; and

when one of the sub-group divergence metrics meets the threshold, partition the one sub-group of the first and second sub-groups which met the threshold into a third sub-group and a fourth sub-group.

12 . The system of claim 11 , wherein the divergence metric is function of at least one of a maximum failed bit count (maxFBC) and a minimum failed bit count (minFBC).

13 . The system of claim 12 , wherein the divergence metric is calculated from 2(maxFBC−minFBC)/(maxFBC+minFBC)).

14 . The system of claim 12 , wherein the controller is configured to partition the memory based on the maxFBC being less than an error correction capability.

15 . The system of claim 14 , wherein

the first sub-group includes a first page of the data read from the memory group that is associated with the minFBC, and

the second sub-group includes a second page of the data read from the memory group that is associated with the maxFBC.

16 . The system of claim 15 , wherein the controller is configured to provide first and second queues, wherein

the first queue indexes all pages of data that have a failed bit count (FBC) less than 30% of the error correction capability, for partitioning into the first sub-group, and

the second queue indexes all pages of data that have the FBC larger than 60% of the error correction capability, for partitioning into the second sub-group.

17 . The system of claim 15 , wherein the controller is configured to merge together any sub-groups including at least the first and second sub-groups to form a set of two or more multiple groups which all have similar optimal read thresholds and which all have maxFBCs less than 70% of an error correction capability.

18 . The system of claim 11 , wherein

the memory group comprises one of a word line group, a page group, a block, a super block, and a memory die, and

the first and second sub-groups comprise partitions from one of the word line group, the page group, the block, the super block, and the memory die.

19 . The system of claim 18 , wherein the controller is configured to place cold data blocks into a fourth sub-group.

20 . The system of claim 19 , wherein the controller is configured to store optimal read thresholds for the first and second sub-groups into a single level cell page.