One or more of multiple metrics for multiple logical page types of the memory device are determined. Each of the metrics is indicative of a number of bit errors associated with a particular logical page type of the multiple logical page types. A current page margin associated with a first logical page type of the multiple logical page types is modified to determine a modified page margin based at least in part on a ratio using one or more of the multiple metrics. The current page margin associated with the first logical page type is adjusted in accordance with the modified page margin.

A memory sub-system configured to execute a read command of a first type using a combine process to read soft bit data and hard bit data from memory cells. For example, a memory device is to: measure signal and noise characteristics of memory cells for the read command; calculate, based on the characteristics, an optimized voltage and two adjacent voltages that have offsets of a same amount from the optimized voltage; read the memory cells for hard bit data using the optimized voltage and for soft bit data using the two adjacent voltages; and transmit, to the processing device, a response including the hard bit data. The soft bit data can be selectively transmitted based on a classification determined from the characteristics. When a read command of a second type is executed, soft bit data is not read; and/or the signal and noise characteristics are not measured.

A processing device performs operations including receiving a request to locate one or more distribution edges of one or more programming distributions of a memory cell, the request specifying a target error rate for the one or more programming distributions, measuring at least one error rate sample of a first programming distribution selected from the one or more programming distributions, and determining a location of a first distribution edge of the first programming distribution at the target error rate based on a comparison of the at least one error rate sample of the first programming distribution against the target error rate.

A memory system includes a memory chip and a memory controller. The memory chip has a first plane and a second plane. A threshold voltage corresponding to multiple bit data is set for each of the memory cells. The memory controller causes the memory chip to execute a first read process on the first plane and the second plane in parallel by using a plurality of first read voltages different from each other for the first plane and the second plane. The first read process being a process of reading a data group of one bit among the multiple bits by using the first read voltages. The memory controller subsequently adjusts the voltage levels of the first read voltages on the basis of the data group read from the memory cells of the first plane and the data group read from the memory cells of the second plane.

The present disclosure generally relates to data storage devices, such as solid state drives (SSDs). A read threshold calibration operation is utilized to generate a calibrated read threshold for one or more voltage states of a cell of a MLC memory. A single-level cell (SLC) read is then executed to sense the ratio of bit values at the read thresholds of the voltage states, where SLC read refers to reading at a single read threshold, rather than to the cell type. The sensing results in a binary page with certain statistics of 1's and 0's. The ratio of 1's (or 0's) in the binary page is used to determine a deviation from the expected ratio, where the deviation is used to adjust the calibrated read threshold to match the voltage states of the MLC memory.

A memory sub-system configured to execute a read command of a first type using a combine process to read soft bit data and hard bit data from memory cells. For example, a memory device is to: measure signal and noise characteristics of memory cells for the read command; calculate, based on the characteristics, an optimized voltage and two adjacent voltages that have offsets of a same amount from the optimized voltage; read the memory cells for hard bit data using the optimized voltage and for soft bit data using the two adjacent voltages; and transmit, to the processing device, a response including the hard bit data. The soft bit data can be selectively transmitted based on a classification determined from the characteristics. When a read command of a second type is executed, soft bit data is not read; and/or the signal and noise characteristics are not measured.

Data storage devices, such as solid state drives (SSDs), are disclosed. A read threshold calibration operation is utilized to generate a calibrated read threshold for one or more voltage states of a cell of a MLC memory. A single-level cell (SLC) read is then executed to sense the ratio of bit values at the read thresholds of the voltage states, where SLC read refers to reading at a single read threshold, rather than to the cell type. The sensing results in a binary page with certain statistics of 1's and 0's. The ratio of 1's (or 0's) in the binary page is used to determine a deviation from the expected ratio, where the deviation is used to adjust the calibrated read threshold to match the voltage states of the MLC memory.

Methods, systems, and devices for techniques for determining memory cell read offsets are described to support determining voltage offsets and corresponding read voltage levels for one or more memory cell levels using a relationship between read voltage levels and voltage offsets. A memory device may estimate first voltage offsets using a first procedure and may perform a read operation using the first voltage offsets. If a first voltage offset results in a read error for a corresponding memory cell level, the memory device may determine an updated voltage offset using the relationship. The relationship may predict a voltage offset for a given read voltage level, such that the memory device may use the relationship to predict an updated voltage offset for a memory cell level. The memory device may use the updated voltage offset(s) to perform a second read operation for the one or more memory cells.

A controller estimates optimal read threshold values for a memory device using deep learning. The memory device includes multiple pages coupled to select word lines in a memory region. The controller performs multiple read operations on a select type of page for each word line using multiple read threshold sets, obtains fail bit count (FBC) information associated with each read operation, and determines an optimal read threshold set for each word line based on the FBC information. When optimal read threshold sets for the select word lines are different each other, the controller predicts a best read threshold set using the optimal read threshold sets.

A memory sub-system configured to iterative calibrate read voltages, where higher read voltages are calibrated based on the calibration results of lower read voltages. For example, a memory device initially determines first read voltages of a group of memory cells. The memory device calculates a second read voltage optimized to read the group of memory cells according to first signal and noise characteristics measured based on at least one of the first read voltages. A third read voltage is estimated based on an offset of the second read voltage from a corresponding voltage among the first read voltages. Second signal and noise characteristics of the group of memory cells are measured based on the third read voltage. The memory device then calculates a fourth read voltage optimized to read the group of memory cells according to the second signal and noise characteristics.