阅读说明：本技术 一种应用于损耗均衡的存储单元质量度量方法 (Memory cell quality measurement method applied to wear leveling ) 是由 刘政林 潘玉茜 陈卓 文思诚 于 2019-08-29 设计创作，主要内容包括：本发明公开了一种应用于损耗均衡的存储单元质量度量方法,包括以下步骤：S1、根据实际存储系统的测量能力,从存储单元的特征集合中选取多个特征值作为存储单元质量度量特征；S2、根据存储单元质量度量特征和预训练的存储单元质量度量模型,计算得到存储单元质量度量结果。本发明考虑到存储单元不同时期的损耗程度差异,采用操作时间、阈值电压分布等多个能够更加准确反映存储单元当前可靠性状态的度量特征,并建立了一种准确的应用于损耗均衡的存储单元质量度量模型,能够大幅提高质量度量结果的准确度,从而提高存储系统损耗均衡执行效率并延长存储系统的使用寿命。(The invention discloses a memory cell quality measurement method applied to wear leveling, which comprises the following steps: s1, selecting a plurality of characteristic values from the characteristic set of the storage unit as the quality measurement characteristics of the storage unit according to the measurement capability of the actual storage system; and S2, calculating to obtain a memory cell quality measurement result according to the memory cell quality measurement characteristics and the pre-trained memory cell quality measurement model. The invention considers the loss degree difference of the storage units in different periods, adopts a plurality of measurement characteristics such as operation time, threshold voltage distribution and the like which can more accurately reflect the current reliability state of the storage units, establishes an accurate storage unit quality measurement model applied to loss balance, and can greatly improve the accuracy of the quality measurement result, thereby improving the loss balance execution efficiency of the storage system and prolonging the service life of the storage system.)

1. A method for memory cell quality measurement for wear leveling, comprising the steps of:

S1, selecting a plurality of characteristic values from the characteristic set of the storage unit as the quality measurement characteristics of the storage unit according to the measurement capability of the actual storage system;

And S2, calculating to obtain a memory cell quality measurement result according to the memory cell quality measurement characteristics and the pre-trained memory cell quality measurement model.

2. the method of claim 1, wherein the set of memory cell characteristics comprises an operating time of a memory cell, a current across a memory cell, a power consumption of a memory cell, a threshold voltage distribution, a memory block cell location flag, a number of program/erase cycles a memory cell has currently undergone, a number of conditionally erroneous memory cells, a number of error bits and an error rate for any subset of memory cells in a particular data pattern.

3. The method of claim 2, wherein the special data patterns comprise data patterns for enhancing wear of memory cells and data patterns for enhancing interference between memory cells.

4. the method of claim 1, wherein the storage unit quality metric features are a subset of the set of storage unit features, and wherein the number of features included in the subset is 2 or more.

5. the memory cell quality metric method of claim 4, characterized in that the memory cell quality metric characteristic comprises an operating time of a memory cell.

6. the method of claim 5, wherein the measuring the operating time of the memory cell comprises: and detecting the R/B signal state of the chip to which the storage unit belongs, if the R/B signal state is low, recording the operation time and the operation type of the storage unit, and when the programming time, the reading time and the erasing time of the storage unit are not 0, obtaining the current programming time, the reading time and the erasing time as the operation time of the storage unit.

7. The method of claim 1, wherein the memory cell quality metric features corresponding to the values of the remaining available program/erase cycles of the memory cells within the closed interval from the maximum value to the minimum value are sequentially measured, and the normalized remaining available program/erase cycles and the corresponding memory cell quality metric features are used to train and obtain the memory cell quality metric model.

8. the method of claim 7, wherein in the process of measuring the memory cell by using the memory cell quality metric model, when the number of remaining available program/erase cycles of the memory cell reaches a minimum value, the corresponding memory cell quality metric feature is calculated, the normalized number of remaining available program/erase cycles and the corresponding memory cell quality metric feature are recorded as a set of samples, and when the number of recorded samples reaches a preset sample threshold, the memory cell quality metric model is trained and updated by using the recorded samples.

9. The memory cell quality metric method of claim 1, wherein the memory cell quality metric method is applied to a quality metric of wear-leveled memory cells.

10. The memory cell quality measurement method according to claim 1, wherein the detected memory cell quality measurement result is stored in a quality metric table of the memory system, and when the memory system performs a wear leveling operation, the memory system performs wear leveling address mapping change and data migration operation according to the memory cell quality metric value stored in the quality metric table.

Technical Field

The invention belongs to the field of quality measurement of storage units, and particularly relates to a quality measurement method of a storage unit applied to wear leveling.

Background

With the development of information technology, memories have become more and more important as carriers for storing data in electronic devices in the fields of communication, consumption, computers, industrial control, military and the like, but low reliability has been one of the main problems of memories.

Wear leveling (also referred to as wear leveling) is a technique to extend the useful life of erasable computer storage media such as solid state drives SSDs, USB flash drives, phase change memories, and the like. Without wear leveling, the underlying flash controller must permanently assign the logical address of the operating system to the physical address of the flash. This means that each write to a previously written block must first read, erase, modify, and rewrite to the same location. This method is time consuming and often written locations wear out quickly, and once a frequently used memory block reaches the end of life, the memory system will not work properly. The service life of the storage system without using wear leveling is greatly reduced. The quality measurement of the memory unit is a core step of wear leveling operation of the memory system, and has high research value.

Currently, the quality of memory cells is typically measured in erase counts in wear leveling operations. The measuring method does not consider the loss degree difference of the storage unit in different periods, the actual reliability degree of the storage unit is difficult to accurately reflect, and the reliability is low.

Therefore, it is an urgent need to provide a method for measuring the quality of a memory cell with high accuracy.

Disclosure of Invention

Aiming at the defects of the prior art, the invention aims to provide a memory cell quality measuring method applied to wear leveling, and aims to solve the problem that the memory cell quality measuring method in the prior art is low in accuracy because the loss degree difference of memory cells in different periods is not considered.

In order to achieve the above object, the present invention provides a method for measuring the quality of a memory cell applied to wear leveling, comprising the following steps:

S1, selecting a plurality of characteristic values from the storage unit characteristic set as storage unit quality measurement characteristics according to the measurement capability of the actual storage system;

And S2, calculating to obtain a memory cell quality measurement result according to the memory cell quality measurement characteristics and the pre-trained memory cell quality measurement model.

Further preferably, the characteristic set of memory cells includes an operation time of the memory cells, a current on the memory cells, a power consumption of the memory cells, a threshold voltage distribution, a memory block cell location flag, a number of program/erase cycles currently experienced by the memory cells, a number of conditionally erroneous memory cells, a number of error bits and an error rate of any subset of memory cells, and a number of error bits and an error rate of any subset of memory cells under a special data pattern.

Further preferably, the special data pattern includes: data patterns for enhancing wear of memory cells, and data patterns for enhancing interference between memory cells.

more preferably, the memory cell quality metric feature is a subset of a memory cell feature set, and the number of features included in the subset is 2 or more.

further preferably, the memory cell quality metric characteristic comprises an operating time of the memory cell.

Further preferably, the method for measuring the operation time of the memory cell includes: and detecting the R/B signal state of the chip to which the storage unit belongs, if the R/B signal state is low, recording the operation time and the operation type of the storage unit, and when the programming time, the reading time and the erasing time of the storage unit are not 0, obtaining the current programming time, the reading time and the erasing time as the operation time of the storage unit.

Further preferably, the quality metric characteristics of the memory cells corresponding to the values of the remaining available program/erase cycles of the memory cells in the closed interval range from the maximum value to the minimum value are sequentially measured, and the normalized remaining available program/erase cycles and the quality metric characteristics of the memory cells corresponding to the remaining available program/erase cycles are used for training to obtain the quality metric model of the memory cells.

Further preferably, in the process of measuring the memory cell by using the memory cell quality metric model, when the remaining available program/erase cycles of the memory cell reach the minimum value, the corresponding memory cell quality metric feature is calculated, the normalized remaining available program/erase cycles and the corresponding memory cell quality metric feature are used as a group of samples and recorded, and when the number of recorded samples reaches a preset sample threshold, the recorded samples are used to train and update the memory cell quality metric model.

Further preferably, the memory cell quality measurement method provided by the invention is applied to the quality measurement of wear-leveling memory cells.

Further preferably, the detected quality measurement result of the storage unit is stored in a quality metric table of the storage system, and when the storage system performs wear leveling operation, the storage system performs wear leveling address mapping change and data migration operation according to the quality metric value of the storage unit stored in the quality metric table, so as to prolong the service life of the storage system.

Through the technical scheme, compared with the prior art, the invention can obtain the following beneficial effects:

1. The invention provides a memory cell quality measuring method applied to wear leveling, which considers the wear degree difference of memory cells in different periods, adopts various measuring characteristics such as operation time, threshold voltage distribution and the like, and can more accurately reflect the current reliability state of the memory cells, calculates the quality measurement value through a mathematical model, solves the problem of low accuracy of the memory cell quality measuring method caused by not considering the wear degree difference of the memory cells in different periods in the prior art, and has higher measuring accuracy.

2. The invention provides a method for measuring the operation time of a storage unit, which measures the operation time through a mark signal of the storage unit, does not need to occupy more computing resources of a storage system, can realize the measurement of the operation time under the condition of not influencing the operation efficiency of the storage system, and has higher measurement efficiency.

3. the storage unit quality measurement method applied to wear leveling provided by the invention can select specific storage unit quality measurement characteristics through the measurement capability of an actual storage system, so that the storage unit quality measurement characteristics are measured under the condition of not sacrificing larger equipment resources, and the measurement cost is lower.

4. the memory unit quality measurement model obtained by training in the memory unit quality measurement method applied to wear leveling provided by the invention can be optimized and updated in real time in the operation process of the memory system, and is more intelligent compared with the method for measuring units in different reliability stages by adopting the same judgment standard in the prior art.

5. The storage unit quality measurement method applied to the loss balance can update the storage unit quality measurement model under the condition that the loss balance algorithm of the storage system is not changed, so that the quality measurement accuracy is improved, and the loss balance algorithm is assisted to further prolong the service life of the storage system.

drawings

FIG. 1 is a flow chart of a method for measuring the quality of a memory cell for wear leveling according to the present invention;

FIG. 2 is a flowchart of a method for calculating an operating time of a memory cell according to an embodiment of the present invention;

FIG. 3 is a flow chart of a method for training a memory cell quality metric model according to an embodiment of the present invention;

FIG. 4 is a block diagram of an artificial neural network employed by embodiments of the present invention.

Detailed Description

in order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

In order to achieve the above object, the present invention provides a method for measuring quality of a memory cell applied to wear leveling, as shown in fig. 1, including the following steps:

S1, selecting a plurality of characteristic values from the storage unit characteristic set as storage unit quality measurement characteristics according to the measurement capability of the actual storage system;

Specifically, the memory cell feature set includes an operation time of the memory cell, a current on the memory cell, a power consumption of the memory cell, a threshold voltage distribution, a memory block cell location flag, a number of program/erase cycles currently experienced by the memory cell, a number of conditionally erroneous memory cells, a number of error bits and an error rate of any subset of memory cells, and a number of error bits and an error rate of any subset of memory cells under a special data pattern. Wherein the special data pattern includes: data patterns for enhancing wear of memory cells, and data patterns for enhancing interference between memory cells.

specifically, the memory cell quality metric features are subsets of a memory cell feature set, and the number of features included in the subsets is 2 or more. Further, the memory cell quality metric characteristic includes an operating time of the memory cell. Specifically, the method for measuring the operating time of the memory cell includes: and detecting the R/B signal state of the chip to which the storage unit belongs, if the R/B signal state is low, recording the operation time and the operation type of the storage unit, and when the programming time, the reading time and the erasing time of the storage unit are not 0, obtaining the current programming time, the reading time and the erasing time as the operation time of the storage unit.

And S2, calculating to obtain a memory cell quality measurement result according to the memory cell quality measurement characteristics and the pre-trained memory cell quality measurement model.

Specifically, the quality measurement characteristics of the memory cells corresponding to the values of the remaining available program/erase cycles of the memory cells in the closed interval range from the maximum value to the minimum value are sequentially measured, and the normalized remaining available program/erase cycles and the quality measurement characteristics of the memory cells corresponding to the remaining available program/erase cycles are adopted for training to obtain the quality measurement model of the memory cells. Further, in the process of measuring the memory cell by using the memory cell quality measurement model, when the remaining available program/erase cycles of the memory cell reach the minimum value, the corresponding memory cell quality measurement feature is calculated, the normalized remaining available program/erase cycles and the corresponding memory cell quality measurement feature are used as a group of samples and recorded, and when the recorded samples reach a preset sample threshold, the recorded samples are used for training and updating the memory cell quality measurement model.

To further illustrate the method for measuring the quality of a memory cell applied to wear leveling proposed by the present invention, the following detailed description is made with reference to the following embodiments: