Appratus and method for changing a read voltage applied for reading data from a non-volatile memory cell

1 . A read retry table (RRT) apparatus coupled to a plurality of memory dies via a data path, wherein the RRT apparatus is configured to:

collect data from a plurality of memory cells coupled to a plurality of word lines in the plurality of memory dies via the data path;

perform a first clustering on the plurality of word lines based on an error correction capability of error correction circuitry for collected data;

perform a second clustering on an outlier of the first clustering; and

generate or update a read retry table based on values obtained from the first clustering and the second clustering,

wherein the first clustering comprises:

searching K cluster centroids based on a number of clusters corresponding to a size of the RRT, where K is a positive integer;

adding at least one of the plurality of word lines to at least one of K clusters established based on the K cluster centroids; and

recalculating and updating the K cluster centroids based on the added word line.

2 . The RRT apparatus of claim 1 , wherein the plurality of memory dies is arranged in a same row or a same column of a same wafer.

3 . The RRT apparatus of claim 2 , wherein the plurality of word lines corresponds to at least one same row address sampled from a memory block or a word line group included in each of the plurality of memory dies.

4 . The RRT apparatus of claim 3 , wherein the RRT apparatus is coupled to the plurality of memory dies via a same channel of the data path.

5 . The RRT apparatus of claim 1 , wherein the RRT apparatus selects the K cluster centroids by:

selecting, as a first centroid among the K cluster centroids, a random word line among the plurality of word lines;

searching, as a second centroid among the K cluster centroid, a word line having a greatest Voronoi distance from the first centroid among the plurality of word lines;

searching, as a third centroid among the K cluster centroid, a word line having a greatest Voronoi distance from the first and second centroids among the plurality of word lines; and

sequentially searching, as a K-th centroid among the K cluster centroid, a word line having a greatest Voronoi distance from the previously selected centroids among the plurality of word lines.

6 . The RRT apparatus of claim 5 ,

wherein the RRT apparatus is configured to:

establish a Voronoi distance range from the K cluster centroids for each of the K clusters, the Voronoi distance range corresponding to the error correction capability of the error correction circuitry; and

classifying a word line which does not belong to the K clusters as the outlier of the first clustering.

7 . The RRT apparatus of claim 1 , wherein the RRT apparatus is configured to calculate a fail bit count (FBC) for each of the plurality of word lines based on preset levels of a read voltage, the preset levels corresponding to selected values among multi-bit data stored in the plurality of memory cells.

8 . The RRT apparatus of claim 7 ,

wherein the RRT apparatus is configured to:

establish a Voronoi distance range from the K cluster centroids for each of the K clusters, the Voronoi distance range determined based on the fail bit count and the error correction capability of the error correction circuitry; and

classify a word line which does not belong to the K clusters as the outlier of the first clustering.

9 . The RRT apparatus of claim 1 , wherein the second clustering comprises:

adding the outlier of the first clustering to an additional cluster having a Voronoi distance range calculated based on the error correction capability of the error correction circuitry and a fail bit count which is obtained by applying values stored or predetermined in the read retry table before the RRT apparatus generates or updates the read retry table; and

calculating the values for the read retry table to add all of the outlier of the first clustering to the additional cluster.

10 . A method for operating a data processing apparatus, the method comprising:

collecting data from a plurality of memory cells coupled to a plurality of word lines in a plurality of memory dies via a data path;

performing a first clustering on the plurality of word lines based on an error correction capability of error correction circuitry for collected data;

performing a second clustering on an outlier of the first clustering; and

generating or updating a read retry table (RRT) based on values obtained from the first clustering and the second clustering,

wherein the method further comprises transferring the collected data to a server, which is configured to perform the first clustering and the second clustering, via a network.

11 . The method of claim 10 , wherein the generating or updating the RRT comprises:

receiving the values obtained from the first clustering and the second clustering via the network;

generating or updating the RRT based on the values; and

storing the generated or updated RRT in a non-volatile memory storage.

12 . The method of claim 10 , wherein the plurality of word lines corresponds to at least one same row address sampled from a memory block or a word line group included in each of the plurality of memory dies.

13 . A method for operating a data processing apparatus, the method comprising:

collecting data from a plurality of memory cells coupled to a plurality of word lines in a plurality of memory dies via a data path;

performing a first clustering on the plurality of word lines based on an error correction capability of error correction circuitry for collected data;

performing a second clustering on an outlier of the first clustering; and

generating or updating a read retry table (RRT) based on values obtained from the first clustering and the second clustering,

wherein the first clustering comprises:

searching K cluster centroids based on a number of clusters corresponding to a size of the RRT, where K is a positive integer;

adding at least one of the plurality of word lines to at least one of K clusters established based on the K cluster centroids; and

recalculating and updating the K cluster centroids based on the added word line.

14 . The method of claim 13 , wherein the selecting comprises:

selecting, as a first centroid among the K cluster centroids, a random word line among the plurality of word lines;

searching, as a second centroid among the K cluster centroid, a word line having a greatest Voronoi distance from the first centroid among the plurality of word lines;

searching, as a third centroid among the K cluster centroid, a word line having a greatest Voronoi distance from the first and second centroids among the plurality of word lines; and

sequentially searching, as a K-th centroid among the K cluster centroid, a word line having a greatest Voronoi distance from the previously selected centroids among the plurality of word lines.

15 . The method of claim 13 ,

wherein the first clustering comprises:

establishing a Voronoi distance range from the K cluster centroids for each of the K clusters, the Voronoi distance range corresponding to the error correction capability of the error correction circuitry; and

classifying a word line which does not belong to the K clusters as the outlier of the first clustering.

16 . The method of claim 13 , wherein the first clustering comprises calculating a fail bit count (FBC) for each of the plurality of word lines based on preset levels of a read voltage, the preset levels corresponding to selected values among multi-bit data stored in the plurality of memory cells.

17 . The method of claim 13 ,

wherein the first clustering comprises:

establishing a Voronoi distance range from the K cluster centroids for each of the K clusters, the Voronoi distance range determined based on the fail bit count and the error correction capability of the error correction circuitry; and

classifying a word line which does not belong to the K clusters as the outlier of the first clustering.

18 . The method of claim 15 , wherein the second clustering comprises:

adding the outlier of the first clustering to an additional cluster having a Voronoi distance range calculated based on the error correction capability of the error correction circuitry and a fail bit count which is obtained by applying values stored or predetermined in the RRT before the generating or updating the RRT; and

calculating the values for the RRT to add all of the outlier of the first clustering to the additional cluster.