JP7358031B2 - Lifespan prediction device, lifespan prediction method, and lifespan prediction program for nonvolatile semiconductor storage devices - Google Patents

Description

The present invention relates to a lifespan prediction device, a lifespan prediction method, and a lifespan prediction program for predicting the lifespan of a nonvolatile semiconductor storage device such as a memory card.

In recent years, nonvolatile semiconductor storage devices such as memory cards (SD cards, etc.), USB memories, and solid state drives (SSD) have become widespread. A nonvolatile semiconductor memory device includes a nonvolatile semiconductor memory such as a NAND flash memory, and a controller that controls access to the nonvolatile semiconductor memory. As a method for predicting the lifespan of a nonvolatile semiconductor memory, a method has been proposed in which the read voltage of the nonvolatile semiconductor memory is internally monitored to predict the lifespan and notify the outside (see, for example, Patent Document 1).

There are a wide variety of types of nonvolatile memories and controllers built into nonvolatile semiconductor storage devices, and there are also a wide variety of combinations thereof. Many nonvolatile semiconductor memory devices that are not equipped with the above-mentioned life prediction function are also on sale.

In addition, a method has been proposed in which data is written in a nonvolatile semiconductor memory device using a predetermined access pattern, the access time for each unit storage area is measured, and the threshold value for product use of the nonvolatile semiconductor memory device is determined based on the measured access time. (For example, see Patent Document 2).

Japanese Patent Application Publication No. 2013-122793 Japanese Patent Application Publication No. 2016-157227

Nonvolatile semiconductor storage devices such as memory cards are often sold bundled with electronic devices such as digital cameras and smartphones. Manufacturers of electronic equipment have a strong need to predict the lifespan of nonvolatile semiconductor memory devices sold in bundles.

The present invention has been made in view of these circumstances, and its purpose is to provide a technique for predicting the lifespan of a nonvolatile semiconductor memory device from the outside.

A lifespan prediction device according to an aspect of the present invention is a lifespan prediction device that predicts the lifespan of a nonvolatile semiconductor storage device including a nonvolatile semiconductor memory and a controller, and writes data to the nonvolatile semiconductor storage device in a predetermined access pattern. an access time calculation unit that calculates an average or most frequent access time based on the response time of a write process; and whether the average or most frequent access time calculated by the access time calculation unit is less than or equal to a lifetime safety threshold. and a lifespan determination part that determines the lifespan of the nonvolatile semiconductor memory device based on the determination result by the comparison processing part.

Another aspect of the present invention is a lifespan prediction method. This lifespan prediction method is a lifespan prediction method for predicting the lifespan of a nonvolatile semiconductor storage device including a nonvolatile semiconductor memory and a controller. An access time calculation step of calculating the average or most frequent access time based on time, and a comparison of determining whether the average or most frequent access time calculated by the access time calculation step is less than or equal to a life safety threshold. and a lifespan determination step of determining the lifespan of the nonvolatile semiconductor memory device based on the determination result from the comparison processing step.

According to the present invention, it is possible to provide a technique for externally predicting the lifespan of a nonvolatile semiconductor memory device.

It is a graph showing the distribution of access times. FIG. 1 is a block diagram showing the configuration of a life prediction device according to an embodiment. FIG. 3 is a schematic diagram showing an access pattern during writing. It is a flowchart which shows the procedure of preliminary processing before life prediction. It is a flowchart which shows the procedure of life prediction processing. It is a flowchart which shows the procedure of life prediction processing. FIG. 2 is a schematic diagram showing the internal structure of a threshold counter.

Hereinafter, the present invention will be explained based on a preferred embodiment with reference to FIGS. 1 to 7. Identical or equivalent components and members shown in each drawing are designated by the same reference numerals, and redundant explanations will be omitted as appropriate. Further, the dimensions of members in each drawing are shown enlarged or reduced as appropriate to facilitate understanding. Further, in each drawing, some members that are not important for explaining the embodiments are omitted.

The nonvolatile semiconductor device life evaluation device according to the embodiment of the present invention accesses the nonvolatile semiconductor memory device from an external interface based on a standard, and measures the access time. The lifespan of the nonvolatile semiconductor memory device is predicted based on the access time.

One conventional method for predicting the lifespan of nonvolatile semiconductor memory devices is one in which an internal control chip directly monitors the write voltage of a nonvolatile semiconductor memory (hereinafter referred to as nonvolatile memory). be. The control chip predicts the lifetime of the nonvolatile memory by monitoring the write voltage in real time, and determines a threshold at which a failure will occur in the nonvolatile memory. Based on this threshold, writing to non-volatile memory that is presumed to be faulty is suppressed to prevent problems in system operation. As described above, there already exists a method of directly predicting the lifespan of a nonvolatile memory by directly monitoring the voltage during writing to the nonvolatile memory using a control chip.

Manufacturers of nonvolatile semiconductor memory devices perform wear leveling on accesses and tune to avoid accessing the same nonvolatile memory as much as possible in order to extend the life of internal nonvolatile memory. Therefore, from outside the nonvolatile semiconductor storage device, it is difficult to accurately determine which nonvolatile memory has been accessed and how many times, and it is basically impossible to determine which nonvolatile memory has reached the end of its lifespan. .

Furthermore, many of the nonvolatile memories and control chips built into nonvolatile semiconductor memory devices on the market are not designed to return any kind of response to the outside when their lifetimes are reached. Therefore, it is extremely difficult to predict the life span of nonvolatile semiconductor memory devices that are generally available on the market from the outside.

The lifetime prediction device according to the present embodiment predicts the lifetime of the entire nonvolatile semiconductor memory device, which is generally available on the market, and includes a nonvolatile memory and a control chip. The life prediction device measures the access time for access from an external interface based on the standard, and predicts the product usage life of the nonvolatile semiconductor memory device based on the measured access time. The lifespan evaluation device also makes it possible to predict the lifespan of a nonvolatile semiconductor memory device whose usage time is unknown, which was difficult to predict in the past.

Hereinafter, an overview of life expectancy of the nonvolatile semiconductor memory device according to this embodiment will be explained. In this embodiment, the access time when writing a specific sector amount of data to a nonvolatile semiconductor memory device is measured. Predict the lifetime from the measured change in write access time. The access time during data writing refers to the time from transmitting a write command and write data from an external interface to a nonvolatile semiconductor memory device to receiving a write completion response command from the nonvolatile semiconductor memory device. Hereinafter, this may be simply referred to as "writing time."

It is difficult to predict the number of times a nonvolatile semiconductor memory device will last until it reaches the end of its lifespan, unless you know how many times it has already been written when you start counting the number of times it has been written. Ta. However, it has been found that the change in access time for writing a fixed specific sector amount gradually becomes faster from the initial state, and eventually becomes a constant access time when the service life is reached.

As the number of writes increases, the access time for writing a fixed specific sector amount becomes faster. This is because the oxide film of the NAND memory deteriorates due to writing, making it easier for electrons accumulated in the NAND memory to pass through. This is due to the fact that it becomes faster. In addition, as the number of NAND memories that cannot be written to increases, the number of processes that cannot be written to increases, and the write processing by the control chip becomes shorter than the originally time-consuming process of writing to NAND memory, and the internal processing time itself becomes shorter. by.

Here, when a nonvolatile semiconductor memory device reaches the end of its lifespan, it means that although it accepts write commands sent from the host controller, written data cannot be normally read from the host controller. Even if an attempt is made to read the data in the written area from the host controller again, for example, a read error occurs and the host controller is unable to read the data normally. When the nonvolatile semiconductor memory device reaches the end of its life, a control chip within the nonvolatile semiconductor memory device performs write processing to the NAND memory in response to a write command sent from the host controller. However, after the control chip performs a process for every NAND memory in the nonvolatile semiconductor storage device when the same NAND memory reaches the end of its lifespan, it returns a response to the host controller indicating the end of the write process. Therefore, after the nonvolatile semiconductor memory device reaches the end of its life, the access time for write processing of a fixed specific sector amount is always constant, and the average write access time of a fixed specific sector amount is also constant. Therefore, since the access time for writing a fixed specific sector amount becomes a certain amount of time when the lifetime is reached, the host controller calculates backward from the access time that has reached its lifetime and calculates the access time for writing a fixed specific sector amount at the current time. Based on the access time, it is possible to determine how much life is left.

As shown in Patent Document 2, nonvolatile semiconductor memory devices with the same internal configuration have a characteristic that they exhibit access time patterns with the same characteristics in response to a fixed specific pattern of write commands. For this reason, it can be said that nonvolatile semiconductor memory devices that obtain an access time pattern as an initial value and have the same characteristics reach the end of their lifespan at the same access time.

The write access time as an initial value is measured based on the following (1) to (3).
(1) A nonvolatile semiconductor memory device is accessed continuously in fixed specific sector units (for example, 256 sector units), and the access time is recorded on the order of milliseconds.
(2) The access pattern is a pattern in which the address is accessed in the forward direction from the minimum LBA (Logical Block Address) to the maximum LBA, or in the reverse direction from the maximum LBA to the minimum LBA, in units of specified LBA. The minimum LBA is typically 0, and the maximum LBA varies by system. Thereby, the access time can be derived for each nonvolatile semiconductor memory device.
(3) In this case, a feature of the recording method is that data for each access time is plotted in detail.

Hitherto, a benchmark test using application software has been used as a method for evaluating the performance of HDDs. This is used as an index that shows a relatively rough performance evaluation of HDDs, and is used to compare the relative performance of HDDs through partial access, and does not take into account lifespan evaluation. There wasn't. Therefore, although it is possible to detect rough differences in the type of access time that each measured individual has, it is not possible to finely classify the type of each measured individual based on the access time. .

Therefore, in order to clarify the time when an LBA is accessed, an access pattern is used that increases or decreases the access start position without overlapping LBAs in fixed LBA units. For example, the publicly available IDEMA (The International Disk Drive Equipment and Materials Association) HDD evaluation pattern (Command completion time measurement tool White Paper) can be used.

Using such an access pattern, the direction of access, the LBA to be accessed, the data to be transferred, and the access method are strictly determined, and data is written to the entire area of the nonvolatile memory instead of a partial area. From this, accurate characteristics of the access performance of the individual to be evaluated can be extracted. Note that the data to be transferred is preferably data that imposes the greatest load on the nonvolatile semiconductor memory device. The data that imposes the most load is normally 256 sectors of 0xFF in the maximum amount of transfer commands prepared in the 24-bit LBA method, and is data in which all bits are "1".

The access pattern shall be measured against fixed transfer data for each LBA of non-volatile memory, which should be strictly evaluated.The plot of the measured data for each LBA shall be given meaning, and what should be calculated for each access time. It is characterized by being able to clarify what is being done. FIG. 1 is a graph showing the distribution of access times. In FIG. 1, the vertical axis represents access time and the horizontal axis represents LBA.

As a device for plotting access times, it is desirable to use a dedicated jig that does not have inappropriate time overhead caused by an operating system (OS), etc., but it is possible to omit the dedicated jig and measure using software. is also possible.

This makes it possible to accurately grasp the access performance of the entire nonvolatile semiconductor memory device including the nonvolatile memory and control chip. It is known that access performance exhibits unique performance for each non-volatile semiconductor storage device with a different internal configuration as described above, and as shown in Figure 1, many access times are I know to concentrate on. Therefore, by averaging the write access time for a fixed specific sector amount of data, the specific time in which the access time is concentrated can be expressed as a component unique to the nonvolatile semiconductor memory device.

In order to first evaluate the access performance of the non-volatile semiconductor memory device, the access time strictly evaluated by writing in fixed specific sector units is averaged and is tentatively defined as the access time at the start of evaluation.

To determine the access time at the end of the service life, first, in order to evaluate the access performance of a non-volatile semiconductor memory device, we executed an access pattern that continuously writes the entire area in fixed specific sector units. Indicates the access time when the device reaches the end of its life. As described above, since the access time in fixed specific sector units when a failure occurs is always the same, it is equal to the access time averaged in fixed specific sector units. This averaged access time is equivalent to the processing time when the NAND memory reaches the end of its life when the control chip of the nonvolatile semiconductor memory device writes to the entire area of the nonvolatile memory in fixed specific sector units. Expressed as the average of the totals. Therefore, in addition to the method of writing data until the end of its life, it is possible to use the average time or the most frequent access time to compare the access time when the nonvolatile semiconductor memory device reaches its end of life. It is possible to evaluate whether there is

For example, if access to a certain capacity is executed in advance and the response time of the control chip when the NAND memory of the nonvolatile semiconductor storage device reaches the end of its life is known, then the access time per capacity at that time can be found. When all the NAND memories in the nonvolatile semiconductor memory device finally reach the end of their lifetime, it is possible to calculate the average access time when accessing in fixed specific sector units. This access time is tentatively defined as the access time at the end of the lifespan. As a result, by executing the access pattern, the host controller can obtain the values of the access time at the start of the evaluation and the access time at the end of the life span, which are averaged access times when writing in fixed specific sector units. can. The present invention is effective in determining the lifespan of non-volatile semiconductor devices that have the same internal structure for which the access time at the start of evaluation is clarified. It is assumed that

As a method for predicting the lifespan, the host controller performs write processing in units of specific sectors fixed in free areas of the nonvolatile semiconductor memory device between access processing required by normal processing. The host controller compares the average time or the most frequent access time per fixed specific sector unit due to write processing with the access time at the start of the evaluation and the access time at the end of the life span. The host controller can know how far the life has progressed from the current fixed average time or mode of access time per specific sector unit.

The access unit in the write process may be, for example, n times the fixed specific sector unit when measuring the access time at the start of the evaluation, so that large data can be written in a small number of times. When the interval between successive command outputs is short, the smallest fixed specific sector unit may be written frequently. Furthermore, these write processes may be combined.

At this time, the measured access time is stored for a specific number of times, the average value or mode is calculated for each fixed specific sector unit, and this average value is compared with the access time at the end of the life span recorded in the initial value registration database. Make a comparison. Non-volatile semiconductor memory devices have the characteristic that the access time for a fixed specific sector unit converges to a specific time, and as the number of writes increases, access times that deviate from the convergence time may appear, but in general the access time converges. It shows the tendency to go. Therefore, by determining the average value or the mode of access times in fixed specific sector units, the time at which most of the access times converge may be used as the time to be compared with the access time at the end of the life span.

Note that the address of an empty area in which no data is written in the nonvolatile semiconductor memory device as seen from the host controller is not an absolute address inside the nonvolatile semiconductor memory device due to the wear leveling control of the control chip. Wear leveling control of the control chip is always working to equalize the number of writes to the NAND memory throughout the nonvolatile semiconductor memory device. Therefore, the number of writes to the NAND memory in the free area of the nonvolatile semiconductor storage device as seen from the host controller is not significantly different from the number of writes to the NAND memory to which other data is written. Therefore, from the perspective of the host controller, by measuring the command response time of an empty area to which no data has been written, it is possible to predict the progress of the number of writes in the entire nonvolatile semiconductor memory device.

As described above, the host controller, for example, writes data in a fixed specific sector unit to an empty area to which no data is written between access operations based on normal access instructions, and obtains the access time. The host controller can predict the lifespan of a nonvolatile semiconductor memory device by comparing changes in the average or most frequent access time per fixed specific sector unit with the access time at the end of the lifespan.

For nonvolatile semiconductor memory devices with the same internal configuration and access time pattern with the same characteristics, even if the number of writes is unknown before the evaluation starts, the evaluation is based on the access time at the start of evaluation and the access time at end of life. It is possible to predict the remaining lifespan. By measuring the current average or most frequent access time and calculating backward from the access time at the end of life, it is possible to determine how long the life is and predict the remaining life.

Hereinafter, a life prediction device according to an embodiment of the present invention will be specifically described.
FIG. 2 is a block diagram showing the configuration of the life prediction device 10 according to the embodiment. The nonvolatile semiconductor memory device 30 to be evaluated includes a nonvolatile semiconductor memory 31 and a controller 32. Upon receiving the write command and write data from the command execution unit 40, the controller 32 writes the data into a designated storage area of the nonvolatile semiconductor memory 31. When writing of the data is completed, a response signal is returned to the command execution unit 40. When the controller 32 receives a read command from the command execution unit 40 , it reads data from the designated storage area of the nonvolatile semiconductor memory 31 and transmits it to the command execution unit 40 .

The command execution unit 40 is connected to the life prediction device 10 and the host controller 50, receives commands from the life prediction device 10 and the host controller 50, and executes write processing and read processing. In the following description, a memory card is assumed as the nonvolatile semiconductor storage device 30.

The host controller 50 stores in advance the access time at the start of evaluation and the access time at the end of the life of the nonvolatile semiconductor memory device 30 in the initial value registration database 52 through external input from a keyboard or the like. The access time at the start of the evaluation and the access time at the end of the lifespan to be stored are related to the nonvolatile semiconductor memory device 30 whose lifespan is predicted, and are also changed when predicting the lifespan of a nonvolatile semiconductor memory device with the same internal structure and the same access pattern. It can be used without. When the operation is started, the host controller 50 performs normal write and read operations on the nonvolatile semiconductor memory device 30 through the command execution unit 40.

The host controller 50 outputs a predetermined command instruction to the command execution unit 40 during normal operation. After outputting a predetermined command to the command execution unit 40, the host controller 50 instructs the life prediction device 10 to execute life prediction if the next command is not issued consecutively.

The lifespan prediction device 10 includes a write processing section 11, a storage processing section 12, an access time calculation section 13, a comparison processing section 14, and a lifespan determination section 15. The life prediction device 10 is configured by, for example, a CPU, and executes processing by each of the above-mentioned parts according to a computer program stored in a memory (not shown) and various information used in the computer program. The life prediction device 10 may be configured as a part of the host controller 50, and functions as a life prediction processing section based on a partial program in the host controller 50, which is configured with an arithmetic processing circuit such as a CPU, for example. There may be.

The write processing unit 11 executes a write process in fixed specific sector units to the free area of the nonvolatile semiconductor memory device 30, and measures the access time due to the write process. The host controller 50 performs normal operations to read and write to the nonvolatile semiconductor memory device 30 via the command execution unit 40, and the write processing unit 11 writes to the nonvolatile semiconductor memory device 30 via the command execution unit 40. shall not be executed simultaneously.

FIG. 3 is a schematic diagram showing an access pattern during writing. The vertical axis indicates LBA, and the horizontal axis indicates command issue number (number indicating command issue order). Each command is issued in order of decreasing command issue number. The write pattern shown in FIG. 3 indicates access in the forward direction, and writes are performed on the entire surface of the medium in units of 256 sectors from 0LBA to the maximum LBA. The data to be written can be arbitrary, but if you write continuous data from 0, 1, 2, to F that always adds 1, if there is a problem with the written data, you will not know which data caused the problem. It's convenient and makes it easy to discover what's warm.

The storage processing unit 12 performs a process of storing the access time measured a specific number of times in a storage unit (not shown). The access time calculation unit 13 calculates the average or most frequent access time per fixed specific sector unit based on the stored access time. The following description will be made using an average access time obtained by averaging access times per fixed specific sector unit, but processing may also be performed using a most frequently occurring access time that is calculated as the most frequently occurring access time.

The comparison processing unit 14 compares the average access time calculated by the access time calculation unit 13 and the access time at the end of life recorded in the initial value registration database 52. Note that the average access time may be output from the life prediction device 10 to the host controller 50, and the host controller 50 may compare the average access time and the access time at the end of the life.

The comparison processing unit 14 compares the average access time with a threshold value that can be said to be on the safe side with respect to the access time at the end of the lifespan (hereinafter referred to as the safety threshold at the end of the lifespan). The safety threshold at the end of the lifespan is set, for example, to a value obtained by multiplying the access time at the end of the lifespan by a safety factor. The lifetime reaching safety threshold is set, for example, to a value of 12 milliseconds, which is obtained by multiplying the access time at the lifetime of 10 milliseconds by a safety factor of 1.2. The lifespan judgment unit 15 determines that the lifespan has been reached when the average access time is often equal to or less than the lifespan reaching safety threshold, and sends a message indicating that read/write operations to the nonvolatile semiconductor storage device 30 should be stopped. Outputs a lifespan reaching alarm. The lifespan determining unit 15 notifies that the nonvolatile semiconductor storage device 30 has reached the end of its lifespan by outputting a lifespan reaching alarm. The host controller 50 instructs replacement of the non-volatile semiconductor storage device 30, data backup, etc. based on the end-of-life alarm output by the life-span determination unit 15.

Next, the operation of the life prediction device 10 will be explained based on the pre-processing before life prediction and the life prediction process. FIG. 4 is a flowchart showing the procedure of pre-processing before life prediction. This pre-processing can be performed by an evaluation device other than the life prediction device 10. The evaluation device outputs a write command to the nonvolatile semiconductor storage device via an external interface corresponding to the command execution unit 40. The evaluation device obtains the time required to receive the write completion response command from the nonvolatile semiconductor memory device as the access time.

The evaluation device measures the response time when writing according to the access pattern shown in FIG. 3 (S1). The response time of each nonvolatile semiconductor memory device evaluated by the evaluation device is registered in the initial value registration database 52 as the access time at the start of evaluation (S2). It is determined whether the response time when the controller in the nonvolatile semiconductor storage device accesses the broken NAND memory can be calculated (S3). The determination in step S3 is made based on, for example, whether the individual nonvolatile semiconductor memory device is known by the manufacturer's model number, etc., and whether it is known or can be calculated based on the database corresponding to the manufacturer's model number, etc.

If the determination result in step S3 is that calculation is possible (S3: YES), the command response time at the end of life is calculated from the processing time of the controller, and the access time at end of life is obtained (S4). If the determination result in step S3 is that calculation is not possible (S3: NO), the evaluation device repeats the writing operation to the nonvolatile semiconductor memory device, and the response time at the end of the life is set as the access time at the end of the life. Acquire (S5).

After steps S4 and S5, the obtained access time at the end of life for each nonvolatile semiconductor memory device is registered in the initial value registration database 52 (S6). Through this preprocessing, the access time at the start of evaluation and the access time at end of life for each nonvolatile semiconductor memory device are registered in the initial value registration database 52.

FIG. 5 and FIG. 6 are flowcharts showing the procedure of life prediction processing. The host controller 50 accesses the initial value registration database 52 and determines whether the nonvolatile semiconductor memory device 30 is a nonvolatile semiconductor memory device registered in the database (S11). If the nonvolatile semiconductor storage device 30 is registered in the initial value registration database 52 (S11: YES), the access time at the start of evaluation and the access time at the end of life are read from the initial value registration database 52 (S13).

If the nonvolatile semiconductor storage device 30 is not registered in the initial value registration database 52 (S11: NO), the name of the nonvolatile semiconductor storage device 30, the access time at the start of evaluation, and the access at the end of the life span are determined based on the above-mentioned preprocessing. The time is registered in the initial value registration database 52 (S12). After step S12, the host controller 50 reads the access time at the start of evaluation and the access time at end of life from the initial value registration database 52 (S13). The host controller 50 outputs the read access time at the start of evaluation and the access time at end of life to the life prediction device 10.

The host controller 50 starts reading and writing to the nonvolatile semiconductor memory device 30 through normal operation (S14). The host controller 50 determines whether or not continuous commands are being output to the nonvolatile semiconductor storage device 30 (S15), and if the commands are being output (S15: YES), the determination in step S15 is repeated.

If the determination in step S15 is negative (S15: NO), based on the notification from the host controller 50, the write processing unit 11 of the life prediction device 10 is fixed to the free space of the nonvolatile semiconductor storage device 30. Writing is performed in units of specific sectors (S16). The write processing unit 11 writes data that is n times the fixed specific sector amount at one time, or data in the minimum fixed specific sector unit multiple times.

The amount of data can be changed by the pattern of the next consecutive command output. If the interval until the next consecutive command output is long, you can write large data that is n times the data of a fixed specific sector unit in a small number of times, or the interval until the next consecutive command output. When the data is small, one fixed specific sector unit may be written frequently. Alternatively, a combination of these may be used.

The storage processing unit 12 of the life prediction device 10 performs a process of storing the measured access time a specific number of times, and the access time calculation unit 13 calculates the average access time per fixed specific sector unit (S17). Note that the most frequent access time can be calculated and used instead of the average access time. The comparison processing unit 14 determines whether the average access time has become equal to or less than the life safety threshold (S18). In the judgment in step S18, in order to increase the accuracy of predicting the end of the lifespan, a large number of access times are stored as a specific number of times and the average is taken. It is desirable to assume that it has arrived.

However, as shown in FIG. 1, the access pattern of an actual nonvolatile semiconductor memory device includes a certain degree of variation. Therefore, data of n times the fixed specific sector unit is written a small number of times, for example, 10 fixed specific sector units are written once, and the average access time for the plurality of times is calculated. As a result, for example, suppose that the end of life is reached when 7 of the 10 recent average access times of a specific sector written in one go are below the life-span safety threshold. .

If the average access time is equal to or less than the life safety threshold in step S18 (S18: YES), the life judgment unit 15 sets the threshold counter to 1 and counts that the average access time has become less than or equal to the life safety threshold. (S19). FIG. 7 is a schematic diagram showing the internal structure of the threshold counter. The threshold counter is a FIFO (First In First Out) counter. If the average access time is not equal to or less than the lifetime safety threshold in step S18 (S18: NO), the threshold counter is set to 0 (S20), and the process returns to step S15 to repeat the process.

For example, when determining whether the lifespan has been reached based on the average access time of 10 times, the inside of this threshold counter is divided into 10 sections, and if 7 of these are 1, it is determined that the lifespan has been reached. After step S19, the lifespan determining unit 15 determines whether the count value by the threshold counter exceeds a specified value (S21). In step S21, if the count value by the threshold counter does not exceed the specified value (S21: NO), the process returns to step S15 and repeats the process. In step S21, if it is determined that the count value by the threshold counter exceeds the specified value (S21: YES), the lifespan determining unit 15 determines that the lifespan has been reached and outputs a lifespan reaching alarm (S22). The lifespan determining unit 15 notifies that the nonvolatile semiconductor storage device 30 has reached the end of its lifespan by outputting a lifespan reaching alarm. The end-of-life alarm may be a lamp, a buzzer, or the like to notify an abnormality. The host controller 50 instructs replacement of the nonvolatile semiconductor storage device 30, data backup, etc. based on the end-of-life alarm (S23), and ends the process.

Another method for determining this is that even if normal continuous commands are frequently output to the nonvolatile semiconductor memory device 30 and there is no room to write large amounts of data, the access time when writing in one fixed specific sector unit may be used. You can memorize many of them to get the same effect. For example, consider a case where 35 of the 50 access times obtained recently when writing in one fixed specific sector unit become less than the lifetime safety threshold and reach the end of the lifetime. In this case, the inside of the threshold value counter is divided into 50 divisions, and if 35 of these divisions become 1, it is assumed that the life span has been reached and the process proceeds to step S22.

The combination of the amount of one write and the size of the FIFO counter depends on the timing at which the continuous read/write commands of the non-volatile semiconductor memory device are issued, and whether the free space of the non-volatile semiconductor memory device can be written in a large amount. It will be judged by how. This kind of judgment is possible because, as shown in Figure 1, in the case of non-volatile semiconductor storage devices, due to the characteristics of NAND device storage, a specific access time component by an internal control chip is used as an access time component. By having only one.

As described above, according to the embodiment, the life prediction device 10 measures the access time from the outside and makes a determination without installing a special life prediction device or the like inside the nonvolatile semiconductor storage device 30. This allows the lifespan to be predicted. The lifetime prediction device 10 can predict the lifetime of a commercially available nonvolatile semiconductor memory device 30 including a nonvolatile semiconductor memory 31 and a controller 32.

In addition, the life prediction device 10 uses a threshold counter to count when the average access time becomes less than or equal to the life safety threshold and compares it with a specified value. can be predicted. Furthermore, the life prediction device 10 predicts the life of the nonvolatile semiconductor storage device 30 before it reaches the end of its life and breaks down by setting the life reaching safety threshold to a value obtained by multiplying the access time at reaching the life by a safety factor. can do.

Furthermore, the life predicting device 10 can also easily predict the progress of the life, the remaining life, etc. based on the access time at the start of evaluation and the access time at the end of the life with respect to the response time during writing.

Next, the characteristics of the life prediction device 10, the life prediction method, and the life prediction program according to each embodiment will be explained.
The lifespan prediction device 10 includes an access time calculation section 13, a comparison processing section 14, and a lifespan determination section 15, and predicts the lifespan of a nonvolatile semiconductor memory device 30 including a nonvolatile semiconductor memory 31 and a controller 32. The access time calculation unit 13 calculates the average or most frequent access time based on the response time of a write process for writing data into the nonvolatile semiconductor memory device 30 in a predetermined access pattern. The comparison processing unit 14 determines whether the average or most frequent access time calculated by the access time calculation unit 13 is less than or equal to the lifetime safety threshold. The lifespan determination section 15 determines the lifespan of the nonvolatile semiconductor storage device 30 based on the determination result by the comparison processing section 14 . Thereby, the life predicting device 10 can predict the life of the nonvolatile semiconductor storage device from the outside based on the process of comparing the calculated average or most frequent access time with the life reaching safety threshold.

Furthermore, the lifespan determining unit 15 counts the cases in which the average or most frequent access time is less than or equal to the lifespan attainment safety threshold, and determines that the nonvolatile semiconductor memory device 30 has reached the end of its lifespan when the average or most frequent access time is equal to or greater than the specified value. Thereby, the lifetime prediction device 10 can predict the lifetime of the nonvolatile semiconductor memory device 30 by taking into account variations in the average or most frequent access time.

Furthermore, the lifetime safety threshold is set to a value obtained by multiplying the access time when the nonvolatile semiconductor memory device 30 reaches the lifetime by a safety factor. Thereby, the life predicting device 10 can predict the life of the nonvolatile semiconductor memory device 30 before the nonvolatile semiconductor memory device 30 reaches the end of its life and breaks down.

The lifespan prediction method includes an access time calculation step, a comparison processing step, and a lifespan determination step, and predicts the lifespan of the nonvolatile semiconductor memory device 30 including the nonvolatile semiconductor memory 31 and the controller 32. The access time calculation step calculates the average or most frequent access time based on the response time of a write process for writing data into the nonvolatile semiconductor memory device 30 in a predetermined access pattern. The comparison processing step determines whether the average or most frequent access time calculated in the access time calculation step is less than or equal to the life safety threshold. The lifespan determination step determines the lifespan of the nonvolatile semiconductor memory device 30 based on the determination result from the comparison processing step. According to this lifespan prediction method, the lifespan of the nonvolatile semiconductor storage device 30 can be predicted from the outside based on the process of comparing the calculated average or most frequent access time with the safety threshold for reaching the lifespan.

The lifespan prediction program causes a computer to execute an access time calculation step, a comparison processing step, and a lifespan determination step to predict the lifespan of the nonvolatile semiconductor memory device 30 including the nonvolatile semiconductor memory 31 and the controller 32. The access time calculation step calculates the average or most frequent access time based on the response time of a write process for writing data into the nonvolatile semiconductor memory device 30 in a predetermined access pattern. The comparison processing step determines whether the average or most frequent access time calculated in the access time calculation step is less than or equal to the life safety threshold. The lifespan determination step determines the lifespan of the nonvolatile semiconductor memory device 30 based on the determination result from the comparison processing step. According to this lifespan prediction program, the lifespan of the nonvolatile semiconductor storage device 30 can be predicted from the outside based on a comparison process between the calculated average or most frequent access time and the safety threshold for reaching the lifespan.

The above description has been based on the embodiments of the present invention. Those skilled in the art will understand that these embodiments are illustrative and that various modifications and changes are possible and within the scope of the claims of the present invention. It is about to be done. Accordingly, the description and drawings herein are to be regarded in an illustrative rather than a restrictive sense.

10 life prediction device, 13 access time calculation unit, 14 comparison processing unit,
15 lifespan judgment unit, 30 nonvolatile semiconductor memory device,
31 non-volatile semiconductor memory, 32 controller.

Claims (5)

A lifespan prediction device that predicts the lifespan of a nonvolatile semiconductor storage device including a nonvolatile semiconductor memory and a controller,
an access time calculation unit that calculates an average or most frequent access time based on a response time of a write process for writing data in a nonvolatile semiconductor memory device in a predetermined access pattern;
Using the characteristic that the access time at the end of the lifespan , which is the access time for write processing of a fixed specific sector amount after the end of the lifespan, is always a constant time , the safety threshold for reaching the lifespan is determined based on the access time at the end of the lifespan. a comparison processing unit that determines whether the average or most frequent access time calculated by the access time calculation unit is equal to or less than the set life safety threshold;
a lifespan determination unit that determines the lifespan of the nonvolatile semiconductor storage device based on the determination result by the comparison processing unit;
A life prediction device comprising: The lifespan determining unit counts cases in which the average or most frequent access time is less than or equal to the lifespan attainment safety threshold, and determines that the nonvolatile semiconductor storage device has reached its lifespan when the average or most frequent access time is equal to or greater than a specified value. The life prediction device according to claim 1. 3. The lifetime prediction device according to claim 1, wherein the lifetime reaching safety threshold is set to a value obtained by multiplying an access time when the nonvolatile semiconductor memory device reaches its lifetime by a safety factor. . A lifespan prediction method for predicting the lifespan of a nonvolatile semiconductor storage device including a nonvolatile semiconductor memory and a controller, the method comprising:
an access time calculation step of calculating an average or most frequent access time based on a response time of a write process for writing data in a nonvolatile semiconductor memory device in a predetermined access pattern;
Using the characteristic that the access time at the end of the lifespan , which is the access time for write processing of a fixed specific amount of sectors after the end of the lifespan, is always a constant time, the safety threshold for reaching the lifespan is determined based on the access time at the end of the lifespan. a comparison processing step of determining whether the average or most frequent access time calculated by the access time calculation unit is equal to or less than the set life safety threshold;
a lifespan determination step of determining the lifespan of the nonvolatile semiconductor storage device based on the determination result from the comparison processing step;
A lifespan prediction method comprising: A lifespan prediction program that predicts the lifespan of a nonvolatile semiconductor storage device including a nonvolatile semiconductor memory and a controller,
an access time calculation step of calculating an average or most frequent access time based on a response time of a write process for writing data in a nonvolatile semiconductor memory device in a predetermined access pattern;
Using the characteristic that the access time at the end of the lifespan , which is the access time for write processing of a fixed specific sector amount after the end of the lifespan, is always a constant time , the safety threshold for reaching the lifespan is determined based on the access time at the end of the lifespan. and determining whether the average or most frequent access time calculated by the access time calculation unit is equal to or less than the set life safety threshold;
a lifespan determination step of determining the lifespan of the nonvolatile semiconductor storage device based on the determination result from the comparison processing step;
A lifespan prediction program that causes a computer to execute.

Images