A refresh compensation method, device and memory

Abstract

The embodiment of the present disclosure provides a compensation refreshing method and device and a memory, the method comprises the following steps: obtaining a row address of a weak row and a corresponding compensation coefficient; determining a compensation time interval based on the compensation coefficient; matching a self-refresh of the weak row according to the row address of the weak row, and determining a self-refresh moment of the weak row; and performing compensation refreshing on the weak row according to the self-refresh moment of the weak row and the compensation time interval.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of semiconductor technology, and relates to but is not limited to a compensation refresh method and device and a memory. BACKGROUND

[0002] Dynamic random access memory (DRAM) as a volatile storage medium, its data will disappear after power failure, that is, the data needs to be refreshed in time to maintain the stability of the data even when powered on. If not refreshed in time, the charge in the storage cell in the DRAM will slowly drain over time, resulting in data loss.

[0003] In theory, at a double-speed refresh rate, as long as the effective retention time of the charge in the DRAM storage cell exceeds the double-speed refresh interval, the stability of the data can be ensured. However, if the data retention capability of the DRAM internal storage cell is weak and less than the double-speed refresh interval, data loss may occur in actual application. SUMMARY

[0004] Therefore, the embodiments of the present disclosure provide a compensation refresh method, device and memory, and a memory.

[0005] In a first aspect, the embodiments of the present disclosure provide a compensation refresh method, which comprises: obtaining a row address of a weak row and a corresponding compensation coefficient; determining a compensation time interval based on the compensation coefficient; matching a self-refresh of the weak row according to the row address of the weak row, and determining a self-refresh time of the weak row; and performing compensation refresh on the weak row according to the self-refresh time of the weak row and the compensation time interval.

[0006] In some embodiments, the determining of the compensation time interval based on the compensation coefficient comprises: determining a compensation refresh number based on the compensation coefficient; and determining the compensation time interval based on the compensation refresh number and a preset standard time interval.

[0007] In some embodiments, the determining of the compensation refresh number based on the compensation coefficient comprises: multiplying the compensation coefficient by a temperature coefficient to obtain the compensation refresh number; and the temperature coefficient is related to temperature, and the higher the temperature, the higher the value of the temperature coefficient.

[0008] In some embodiments, the obtaining of the row address of the weak row and the corresponding compensation coefficient comprises:

[0009] The preset standard time interval is divided by the data retention time of the weak row, and the obtained quotient is rounded down to obtain the corresponding compensation coefficient; wherein, the preset standard time interval is the minimum data retention time of the normal row which does not need to perform the compensation refresh.

[0010] In some embodiments, the row address of the weak row, the compensation refresh number and the compensation time interval are all stored in a management list; the management list also stores a start flag.

[0011] In some embodiments, the matching of the self-refresh of the weak row according to the row address of the weak row comprises: in the process of performing the self-refresh, sequentially matching the row address of the weak row with a current row address to match the self-refresh of the weak row; the current row address is the row address of the current self-refresh.

[0012] In some embodiments, the sequential matching of the row address of the weak row with the current row address comprises: setting the start flag to a first value in the management list; obtaining the current row address; matching the current row address with the row address of the weak row one by one; if the current row address belongs to the row address of the weak row, resetting the compensation refresh number corresponding to the weak row in the management list and setting the start flag corresponding to the weak row to a second value.

[0013] In some embodiments, the compensation refresh of the weak row according to the self-refresh time of the weak row and the compensation time interval comprises: confirming the start flag corresponding to the weak row in the management list; if the start flag is the second value, starting from the self-refresh time of the weak row, performing the compensation refresh of the weak row once every compensation time interval until the compensation refresh number is reached, and resetting the start flag to the first value.

[0014] In some embodiments, starting from the self-refresh time of the weak row, performing the compensation refresh of the weak row once every compensation time interval until the compensation refresh number is reached comprises: traversing the management list to determine the compensation refresh number and the compensation time interval corresponding to the weak row; starting from the self-refresh time of the weak row, performing the compensation refresh of the weak row once every compensation time interval, and reducing the compensation refresh number corresponding to the weak row in the management list by 1 until the compensation refresh number corresponding to the weak row is reduced to 0.

[0015] In some embodiments, after the row address and the corresponding compensation coefficient of the weak row are acquired, the method further comprises: judging whether the management list is empty; if the management list is empty, the compensation refresh is not performed.

[0016] In the second aspect, the embodiments of the present disclosure provide a compensation refresh device, which comprises: a recording module configured to save the row address and the corresponding compensation coefficient of the weak row; a compensation management module configured to acquire the row address and the corresponding compensation coefficient of the weak row from the recording module; determine the compensation time interval based on the compensation coefficient; match the self-refresh of the weak row according to the row address of the weak row, and determine the self-refresh time of the weak row; and perform compensation refresh on the weak row according to the self-refresh time of the weak row and the compensation time interval.

[0017] In some embodiments, the compensation management module comprises: an initialization module configured to acquire the row address and the corresponding compensation coefficient of the weak row; determine the compensation time interval based on the compensation coefficient; a decision module configured to match the self-refresh of the weak row according to the row address of the weak row, and determine the self-refresh time of the weak row; and an execution module configured to perform compensation refresh on the weak row according to the self-refresh time of the weak row and the compensation time interval.

[0018] In some embodiments, the compensation management module further comprises: a register module configured to register a management list; and the management list comprises a group of item information corresponding to each of the weak rows.

[0019] In some embodiments, the group of item information comprises: the row address of the weak row, the number of compensation refreshes, the compensation time interval, and a preset starting flag.

[0020] In the third aspect, the embodiments of the present disclosure provide a semiconductor memory comprising the compensation refresh device of any one of the above.

[0021] In some embodiments, the memory comprises a dynamic random access memory.

[0022] The compensation refresh method, device and memory provided by the embodiments of the present disclosure, wherein the compensation refresh method comprises determining the compensation time interval of the weak row, and performing compensation refresh on the weak row based on the self-refresh time of the weak row and the compensation time interval. In this way, the original refresh circuit of the DRAM is not changed, and only the weak storage unit is additionally increased with appropriate refresh times, so that the compensation refresh can be completed through simple steps, thereby ensuring the stability of the data of the weak storage unit and avoiding the loss of data. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 Flowchart of the compensation refresh method provided by the embodiment of the present disclosure Figure 1 ;

[0024] Figure 2 Schematic diagram of the compensation refresh device provided by the embodiment of the present disclosure Figure 1 ;

[0025] Figure 3 Flowchart of the compensation refresh method provided by the embodiment of the present disclosure Figure 2 ;

[0026] Figure 4 Schematic diagram of the compensation refresh device provided by the embodiment of the present disclosure Figure 2 ;

[0027] Figure 5 Flowchart of the compensation refresh method provided by the embodiment of the present disclosure Figure 3 ;

[0028] Figure 6 Flowchart of the compensation refresh method provided by the embodiment of the present disclosure Figure 4 ;

[0029] Figure 7 Schematic diagram of the compensation refresh device provided by the embodiment of the present disclosure Figure 3 ;

[0030] Figure 8 Schematic diagram of the compensation refresh device provided by the embodiment of the present disclosure Figure 4 ;

[0031] Figure 9 Schematic diagram of the compensation refresh device provided by the embodiment of the present disclosure Figure 5 ;

[0032] Figure 10 Structure diagram of a memory provided by the embodiment of the present disclosure. DETAILED DESCRIPTION

[0033] Exemplary embodiments of the present disclosure will be described herein below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be implemented in various forms and should not be limited by the specific embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the present disclosure to those skilled in the art.

[0034] In the following description, numerous specific details are set forth to provide a more thorough understanding of the present disclosure. However, it will be apparent to one of skill in the art upon

[0035] In the drawings, the size of layers, regions, elements, and the like can be exaggerated for clarity. Like reference numerals can represent like elements throughout the several figures.

[0036] It will be understood that when an element or layer is referred to as being "on" or "connected to" another element or layer, it can be directly on or connected to the other element or layer or intervening elements or layers can be present. In contrast, when an element is referred to as being "directly on" or "directly connected to" another element or layer, there are no intervening elements or layers present. It will also be understood that, although the terms first, second, third, etc. can be used herein to describe various elements, components, regions, layers and / or sections, these elements, components, regions, layers and / or sections should not be limited by these terms since such terms are only used to distinguish one element, component, region, layer or section from another element, component, region, layer or section.

[0037] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present disclosure. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising", when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0038] In the related art, a DRAM controller (Controller or Control Circuits) at a system on chip (SOC) end sends a refresh instruction to each row (Row) in an array region (Array) in a DRAM according to a refresh rate of the DRAM to maintain stability of data; when the system enters a suspend state, the DRAM controller stops sending any command, and the DRAM internally starts a self-refresh function to refresh each row in the array region in a timing manner.

[0039] Regardless of the refresh interval of the DRAM controller or the internal self-refresh function of the DRAM, the refresh interval depends on the refresh rate of the DRAM itself; the refresh rate is automatically set according to temperature change, and the refresh interval of the refresh rate of one time is usually 32 milliseconds (ms) according to a specification (spec); the higher the temperature, the higher the refresh rate, that is, the smaller the refresh interval. When the refresh rate of one time is at a normal temperature, as long as the charge effective retention time of a memory cell in the DRAM is more than 32 ms, the stability of data can be ensured, but if the data retention capability of the memory cell in the DRAM is weak and less than 32 ms, data loss may occur in actual application. For a die with very poor data retention capability, it is naturally treated as a failed IC, but for a die with a small number of memory cells and a data retention capability close to or slightly less than 32 ms, it is a pity to treat it as a failed IC.

[0040] Based on this, the embodiment of the disclosure provides a compensation refresh method, which is applied to a compensation refresh device. The compensation refresh device includes a recording module and a compensation management module, and the compensation management module includes an initialization module, a judgment module, an execution module, and a register module.

[0041] Figure 1 A flowchart of the compensation refresh method provided by the embodiment of the disclosure is shown in FIG. 1, which includes steps S101 to S104. Figure 1 As shown in FIG. 1, the compensation refresh method includes steps S101 to S104, wherein:

[0042] In step S101, a weak row address (Weak Row Address) and a corresponding compensating factor (Compensating Factor) are obtained.

[0043] In the embodiment of the disclosure, the initialization module can be used to perform step S101.

[0044] In the embodiments of the present disclosure, the weak row can be a row in which the storage cells have a data retention capability that is critical around or slightly less than a preset standard time interval. The preset standard time interval can be the minimum data retention time of a normal row (or normal storage cell) that does not need to be compensated for refresh. The normal row (or normal storage cell) can be a row (or storage cell) having a data retention time greater than or equal to the preset standard time interval. The data retention time can be the retention time of the stored charge in the normal row (or normal storage cell). In addition, the weak storage cell can be a storage cell having a data retention time that is critical around or slightly less than the preset standard time interval. Therefore, the weak row can be a row in which the weak storage cells are located, that is, the weak row includes at least one weak storage cell.

[0045] In the embodiments of the present disclosure, since the DRAM originally has a separate and continuously operating self-refresh circuit, the self-refresh circuit sequentially refreshes each row in the DRAM to keep the data valid. Each row is refreshed once, and a certain time is required after a round of refresh of all rows. The time interval between each round is the minimum data retention time of the normal row that does not need to be compensated for refresh. Therefore, the preset standard time interval is the interval from the self-refresh of the same row to the next self-refresh. The preset standard time interval can change with temperature, but it is a relatively fixed parameter in a certain time period. This is because, as the temperature increases, the preset standard time interval decreases. In a certain time period, the change in temperature can be ignored.

[0046] For example, in a certain time period at room temperature, the preset standard time interval is 32 ms, and the DRAM is refreshed at a double-speed refresh rate (i.e., the refresh interval is 32 ms). In this case, the weak row is a row in which the storage cells have a data retention capability that is critical around or slightly less than 32 ms.

[0047] It should be noted that the data retention capability of the normal storage cell in the DRAM changes with the change in temperature, and the refresh rate also changes accordingly, and the preset standard time interval also changes accordingly. The higher the temperature, the smaller the data retention capability of the normal storage cell in the DRAM, and the higher the refresh rate, that is, the preset standard time interval is smaller.

[0048] In the embodiments of the present disclosure, the compensation coefficient can be determined according to the ratio of the minimum data retention time (i.e., the standard time interval) of the normal storage unit and the data retention time of the weak storage unit, specifically, the compensation coefficient = the minimum data retention time of the normal storage unit / the data retention time of the weak storage unit (the result is rounded down). For example, the minimum data retention time of the weak storage unit in the weak row is 13 ms, and at this time, the compensation coefficient of the weak row = 32 / 13 = 2 (2.46 is rounded down).

[0049] In step S102, a compensation time interval is determined based on the compensation coefficient.

[0050] In the embodiments of the present disclosure, the initialization module can be used to perform step S102.

[0051] In the embodiments of the present disclosure, the compensation time interval can be the time interval of the inserted compensation refresh. For example, the compensation refresh is inserted once in the self-refresh interval of the weak row, and for the weak row, the refresh interval becomes 32 ms / 2 = 16 ms, that is, the time interval of the inserted compensation refresh is 16 ms.

[0052] In the embodiments of the present disclosure, the compensation refresh can be the refresh of the weak row by the compensation circuit between adjacent two self-refreshes. For example, the standard time interval is 32 ms at normal temperature, and several refresh operations are performed in the interval of 32 ms, that is, the compensation refresh is inserted between adjacent two self-refresh intervals of the weak row.

[0053] It should be noted that because the compensation refresh is inserted in the self-refresh process of the DRAM, the two refresh operations are the same, so after the self-refresh circuit sends a refresh command, the compensation refresh is not performed at the same time, but the compensation circuit sends a refresh command after the compensation time interval (for example, after 16 ms); wherein, the self-refresh circuit and the compensation circuit are used to send the refresh command. In addition, there are many rows in the DRAM, and the self-refresh needs to be performed in the preset standard time interval (for example, 32 ms), and for any row, the self-refresh is performed once in the interval of the preset standard time interval (for example, 32 ms).

[0054] In some embodiments, step S102 can include determining a compensation refresh number based on the compensation coefficient; and determining a compensation time interval based on the compensation refresh number and the preset standard time interval.

[0055] In the embodiments of the present disclosure, the compensation time interval = normal storage unit minimum data retention time / (refresh compensation number + 1). The refresh compensation number = n * compensation coefficient, wherein n is a temperature coefficient, and n is set to 1 at normal temperature. For example, in the case of n = 1 (at normal temperature) and compensation coefficient = 1, the refresh compensation number = 1 * 1 = 1 is calculated, and the compensation time interval = 32 / (1 + 1) = 16 ms.

[0056] In step S103, the self-refresh of the weak row is matched according to the row address of the weak row, and the self-refresh time of the weak row is determined.

[0057] In some embodiments, the step S103 can be performed by a decision module.

[0058] In some embodiments, the self-refresh of the weak row is matched according to the row address of the weak row, including: in the process of self-refreshing, the row address of the weak row is matched with the current row address one by one to match the self-refresh of the weak row. Specifically, in the process of self-refreshing, the compensation refresh circuit monitors the row address of the current row refreshed by the self-refresh circuit, and compares it with the obtained row address of the weak row. If the row addresses are consistent, it is considered that the current row is the weak row, and the self-refresh of the current row is the self-refresh of the weak row. Wherein, the current row address is the row address of the current self-refreshing row.

[0059] In the embodiments of the present disclosure, if the row address of the weak row is consistent with the row address of the current row, the time when the self-refresh circuit refreshes to the current row is taken as the self-refresh time of the weak row; if the row address of the weak row is not consistent with the row address of the current row, it is considered that the current row is not the weak row, and the compensation refresh is not performed on the row.

[0060] For example, the self-refresh circuit sequentially refreshes each row in order, the compensation refresh circuit monitors the row address of the current row refreshed by the self-refresh circuit, and when the self-refresh circuit refreshes to row a, the row address of row a is compared with the obtained row address of the weak row. If the row address of row a is consistent with the row address of one of the weak rows (for example, the row address of the weak row 0), it is considered that row a is the weak row, and the time when the self-refresh circuit refreshes to row a is taken as the self-refresh time of the weak row; if the row address of row a is not consistent with any of the obtained row addresses of the weak rows, it is considered that row a is not the weak row, and the compensation refresh is not performed, and thus there is no need to record the time when the self-refresh circuit refreshes to the row.

[0061] In some embodiments, the row address of the weak row is matched with the current row address one by one, which can include: in the management list, setting the start flag to a first value; obtaining the current row address; matching the current row address with the row address of the weak row one by one; in the management list, resetting the compensation refresh number corresponding to the weak row, and setting the start flag corresponding to the weak row to a second value.

[0062] In step S104, the weak row is compensated and refreshed according to the self-refresh time of the weak row and the compensation time interval.

[0063] In the embodiments of the present disclosure, the step S104 can be executed by an execution module.

[0064] In the embodiments of the present disclosure, the compensation and refresh of the weak row means that the compensation circuit sends a refresh command every time the compensation time interval passes after the self-refresh time of the weak row. For example, the compensation time interval of the weak row 0 is 16 ms, and the compensation and refresh of the weak row 0 is inserted every time 16 ms passes after the self-refresh time of the weak row 0. The compensation time interval of the weak row 1 is 11 ms, and the compensation and refresh of the weak row 1 is inserted every time 11 ms passes after the self-refresh time of the weak row 1.

[0065] In some embodiments, the step S104 can include: in the management list, confirming the starting flag corresponding to the weak row; starting from the self-refresh time of the weak row, and every time the compensation time interval passes, the compensation and refresh of the weak row is performed once until the compensation and refresh number is reached, and the starting flag is reset to the first value.

[0066] It should be noted that, please refer to Figure 2 In the process of refreshing the DRAM (i.e., the memory 100), when the DRAM controller 230 at the system chip 260 side or the internal self-refresh circuit 240 of the DRAM refreshes the row in the array region 250, the compensation management module 720 can insert the compensation and refresh, thereby ensuring the charge effectiveness of the storage unit in the DRAM, and further ensuring the stability of the DRAM data.

[0067] In addition, when the DRAM controller 230 at the system chip 260 side controls the refresh operation, the self-refresh circuit is closed, and the compensation and refresh circuit needs to obtain the row address in the external refresh instruction and compare it with the row address of the weak row, so as to synchronize the refresh of the DRAM controller 230 and the compensation and refresh of the weak row, and avoid conflicts.

[0068] In the embodiments of the present disclosure, the compensation and refresh of the weak row is performed based on the self-refresh time of the weak row and the compensation time interval, so that the original refresh circuit in the DRAM does not change the refresh of each row, and only the weak storage unit is additionally increased by a proper number of refreshes. The compensation and refresh can be completed by simple steps, thereby ensuring the stability of the data of the weak storage unit and avoiding the loss of data.

[0069] In the embodiment of the present disclosure, before step S101 is performed, the compensation refresh method further includes step S101a of saving the row address of the weak row and the corresponding compensation coefficient.

[0070] In the embodiment of the present disclosure, please continue to refer to Figure 2 The recording module 710 is added in the one-time programmable memory (Efuse) in the DRAM, and the recording module 710 is used to perform step S101a, that is, the row address of the weak row and the corresponding compensation coefficient are saved by the recording module.

[0071] In the embodiment of the present disclosure, the DRAM is tested first to obtain the row address of the weak row, the minimum data retention time of the normal storage unit and the data retention time of the weak storage unit in the DRAM; during the test, the compensation coefficient corresponding to the weak row address (or the weak row) is obtained according to the minimum data retention time of the normal storage unit and the data retention time of the weak storage unit. Next, the row address of the weak row and the corresponding compensation coefficient obtained by the test are saved to the recording module.

[0072] In the embodiment of the present disclosure, the compensation coefficient = the minimum data retention time of the normal storage unit / the data retention time of the weak storage unit (the result is rounded down), that is, the compensation coefficient is the minimum data retention time of the normal storage unit divided by the data retention time of the weak storage unit, and the quotient is rounded down. Since the compensation refresh needs to keep the data retention data of each weak storage unit in the weak row valid, the inserted compensation refresh needs to be based on the weak storage unit with the shortest data retention time in the weak row.

[0073] For example, at normal temperature, the data retention time of the normal storage unit is greater than or equal to 32 ms, and the minimum data retention time of the normal storage unit is 32 ms; the data retention time of the weak storage unit in the weak row 0 (Weak Row Address 0) is between 16-32 ms (16 ms < data retention time < 32 ms); at this time, the compensation coefficient of the weak row 0 = 1.

[0074] For another example, at normal temperature, the data retention time of the normal storage unit is greater than or equal to 32 ms, and the minimum data retention time of the normal storage unit is 32 ms; the data retention time of the weak storage unit in the weak row 1 (Weak Row Address 1) is between 32 / 3-16 ms (32 / 3 ms < data retention time ≤ 16 ms); at this time, the compensation coefficient of the weak row 1 = 2.

[0075] For example, at normal temperature, the data retention time of a normal storage unit is greater than or equal to 32 ms, and the minimum data retention time of the normal storage unit is 32 ms; the data retention time of a weak storage unit in a weak row 2 is between 8 and 32 / 3 ms (8 ms < data retention time ≤ 32 / 3 ms), and at this time, the compensation coefficient of the weak row 2 is 3. As can be seen, the weak row and the compensation coefficient are one-to-one corresponding, that is, the weak row address and the compensation coefficient are also one-to-one corresponding.

[0076] It should be noted that the present disclosure only takes three weak rows as an example, and in fact, the number of weak rows can be 0, 1 or more.

[0077] In the embodiments of the present disclosure, the DRAM can be tested by a suitable test device and test method such as an automatic test equipment (ATE) and a wafer test. In the embodiments of the present disclosure, the row address of the weak row and the corresponding compensation coefficient of the DRAM obtained by testing can facilitate subsequent accurate compensation and refresh of the weak row without affecting the refresh of the normal row.

[0078] It should be noted that the number of weak rows saved by the recording module should not be too large, and can be several to tens, which should be comprehensively evaluated according to product characteristics, current process level and implementation cost.

[0079] In some embodiments, as shown in FIG. 10, Figure 3 the step S102 in the method 1000 can be implemented by the steps S1021 and S1022 in the method 1001. Figure 1 Figure 2 The step S1021 determines the compensation refresh times based on the compensation coefficient.

[0080] The step S1021 includes: multiplying the compensation coefficient and a temperature coefficient to obtain the compensation refresh times, wherein the temperature coefficient is related to temperature, and the higher the temperature, the higher the value of the temperature coefficient.

[0081] In some embodiments, the step S1021 includes: multiplying the compensation coefficient and a temperature coefficient to obtain the compensation refresh times, wherein the temperature coefficient is related to temperature, and the higher the temperature, the higher the value of the temperature coefficient.

[0082] ​In the embodiments of the present disclosure, the refresh compensation times = n*compensation coefficient, wherein n is the temperature coefficient. For example, when n = 1 is set at normal temperature, the refresh compensation times = compensation coefficient. According to the compensation coefficient corresponding to the weak row (i.e., the compensation coefficient of the weak row 0 = 1, the compensation coefficient of the weak row 1 = 2, and the compensation coefficient of the weak row 2 = 3) obtained in step S101 (i.e., saved in step S101a), it is determined that the refresh compensation times of the weak row 0, the weak row 1 and the weak row 2 are 1, 2 and 3 respectively. That is, the weak row 0 is inserted with 1 compensation refresh, the weak row 1 is inserted with 2 self-refresh compensations, and the weak row 2 is inserted with 3 self-refresh compensations.

[0083] It should be noted that the higher the temperature, the smaller the data retention capability of the weak storage unit in the DRAM, and the larger the value of n, so that the refresh compensation times of the weak row are more.

[0084] In step S1022, the compensation time interval is determined based on the compensation refresh times and the preset standard time interval.

[0085] In some embodiments, in step S1022, the compensation time interval is obtained by dividing the preset standard time interval by the sum of the compensation refresh times and 1. That is, the compensation time interval = minimum data retention time of the normal storage unit / (refresh compensation times + 1).

[0086] For example, the minimum data retention time of the normal storage unit at normal temperature is 32 ms. The refresh compensation times of the weak row 0 = 1, and the compensation time interval = 32 / 2 = 16 ms, that is, 1 compensation refresh is inserted in the self-refresh interval of the weak row 0, and the time interval of the inserted compensation refresh is 16 ms. For another example, the refresh compensation times of the weak row 1 = 2, and the compensation time interval = 32 / 3, that is, 2 compensation refreshes are inserted in the self-refresh interval of the weak row 1, and the time interval of the inserted compensation refresh is 32 / 3 ms. For another example, the refresh compensation times of the weak row 2 = 3, and the compensation time interval = 32 / 4 = 8 ms, that is, 3 compensation refreshes are inserted in the self-refresh interval of the weak row 2, and the time interval of the inserted compensation refresh is 8 ms.

[0087] It can be understood that by determining the refresh compensation times and the compensation time interval, the compensation refresh can be performed after the self-refresh, so as to avoid the conflict between the self-refresh and the compensation refresh.

[0088] In some embodiments, the row address of the weak row, the compensation refresh times and the compensation time interval are all stored in a management list (Manage list); the management list also stores a start flag. The management list is saved to a register module in the compensation management module.

[0089] In the embodiments of the present disclosure, the row address of the weak row, the start flag, the compensation refresh number and the compensation time interval are collectively referred to as running parameters, and the four running parameters are taken as an item information group (Item) to manage the list including the item information group of each weak row.

[0090] For example, as shown in Figure 4 , the row address of the weak row 0, the corresponding start flag, the corresponding compensation refresh number and the corresponding compensation time interval are taken as the item information group 0 (Item 0). The row address of the weak row 1, the corresponding start flag, the corresponding compensation refresh number and the corresponding compensation time interval are taken as the item information group 1 (Item 1). The row address of the weak row 2, the corresponding start flag, the corresponding compensation refresh number and the corresponding compensation time interval are taken as the item information group 2 (Item 2). The management list includes the item information group 0, the item information group 1 and the item information group 2.

[0091] It should be noted that the embodiments of the present disclosure only list three item information groups, and in fact, the number of item information groups is the same as the number of weak rows. When the number of weak rows is 1, 2, 4 or more, the number of item information groups is also 1, 2, 4 or more.

[0092] In the embodiments of the present disclosure, one record table (Registers) in the register module saves one running parameter, and therefore, four record tables (Registers) save four running parameters in one item information group, that is, the number of record tables is four times the maximum number of weak rows that can be recorded.

[0093] For example, please continue to refer to Figure 4 , the record table 0 (reg0) to the record table 3 (reg3) save the four running parameters in the item information group 0. The record table 4 (reg4) to the record table 7 (reg7) save the four running parameters in the item information group 1. The record table 8 (reg8) to the record table 11 (reg11) save the four running parameters in the item information group 2. It can be seen that the number of record tables is 12, the number of weak rows is 3 (one weak row corresponds to four running parameters), and the number of record tables is four times the number of weak rows.

[0094] It should be noted that since the original information of the weak row (the row address of the weak row and the compensation coefficient) is in the recording module, and the recording module is in the one-time programmable memory, the data in the recording module is not convenient to use directly at runtime, and needs to be additionally put into the register module for dynamic management. The register module used for management is for the compensation management module (i.e., the judgment module and the execution module), and its essence is a parameter table. The compensation circuit needs to monitor whether the self-refresh circuit has refreshed the weak row, so as to synchronize. This requires continuously taking the row address in the self-refresh circuit to compare with the weak row recorded in the management list, for example, comparing the row address in the self-refresh circuit with the row address of the weak row recorded in the management list.

[0095] In some embodiments, after step S101, the compensation refresh method further includes step S102a of judging whether the management list is empty; if the management list is empty, the compensation refresh is not performed.

[0096] In the embodiments of the present disclosure, after the compensation management module is initialized, it can be determined whether there is a weak row record in the initialization module. If there is, a signal is sent to the execution module to perform the following compensation refresh. If there is no weak row record, it means that there is no weak row in the DRAM, and the compensation refresh is not needed, and no signal is sent. The compensation management module initialization refers to that after the DRAM is powered up, the initialization module in the compensation management module loads the row address of the weak row in the recording module into the register module, and calculates the corresponding compensation refresh times and compensation time interval according to the compensation coefficient in the recording module.

[0097] In some embodiments, Figure 1 The step S103 in the method includes: in the process of self-refreshing, sequentially matching the row address of the weak row with the current row address to match the self-refresh of the weak row; and the current row address is the row address currently performing self-refresh.

[0098] In some embodiments, as shown in Figure 5 The step of sequentially matching the row address of the weak row with the current row address can be implemented by steps S1031 to S1034.

[0099] In step S1031, the start flag is set to the first value in the management list.

[0100] In the embodiments of the present disclosure, the start flag is set to the first value in the management list, and the first value can be 0.

[0101] In step S1032, the current row address is obtained.

[0102] In the embodiments of the present disclosure, the row address in the external refresh instruction or the row address in the self-refresh circuit can be acquired. For example, when the DRAM controller at the SOC end refreshes each row in the array region in the DRAM to keep the data stable, the row address in the refresh instruction is acquired; when the self-refresh function in the DRAM is started, each row in the array region is refreshed in a timing manner, and the row address in the self-refresh circuit is acquired.

[0103] In step S1033, the current row address is matched with the row addresses of the weak rows one by one.

[0104] In the embodiments of the present disclosure, the compensation refresh circuit monitors the row address refreshed by the self-refresh circuit, and matches and compares the row address with the row addresses of the weak rows in the management list one by one. If the row addresses are consistent, step S1034 is performed; if the row addresses are inconsistent, step S1032 is performed.

[0105] In step S1034, the compensation refresh times corresponding to the weak row in the management list are reset, and the starting flag corresponding to the weak row is set to the second value.

[0106] In the embodiments of the present disclosure, if the current row address belongs to the row addresses of the weak rows, it is determined in the management list whether the compensation refresh times of the weak row are 0. If the compensation refresh times are 0, the compensation refresh times corresponding to the weak row are reset, and the starting flag corresponding to the weak row is set to the second value. The second value can be 1.

[0107] For example, when the self-refresh circuit refreshes to row a, the address of row a is compared with the acquired row addresses of the weak rows (including the row address of the weak row 0, the row address of the weak row 1 and the row address of the weak row 2) one by one. If the address of row a is consistent with the address of the weak row 0, row a is considered as the weak row 0. In the management list, the compensation refresh times corresponding to the weak row 0 are found according to the address of the weak row 0, and it is determined whether the compensation refresh times of the weak row 0 are 0. If the compensation refresh times are 0, the compensation refresh times of the weak row 0 need to be reset, and the starting flag of the weak row 0 is set to 1 to perform the compensation refresh operation. The judgment method of the weak row 1 and the weak row 2 is similar to that of the weak row 0, which is not described herein again.

[0108] It should be noted that please continue to refer to Figure 5 After it is determined that the current behavior is the weak row (i.e., step S1034), the self-refresh circuit continues to refresh to the next row, and the compensation refresh circuit continues to monitor the row address refreshed by the self-refresh circuit to the next row, and compares the row address with the acquired row addresses of the weak rows one by one (i.e., step S1032) to implement the loop.

[0109] If the current row address does not belong to the row addresses of the weak rows, that is, the row addresses are inconsistent, the row address refreshed to the next row by the self-refresh is continuously monitored, and step S1032 is executed. For example, when the self-refresh circuit refreshes to row b, the address of row b is compared with the acquired row addresses of the weak rows (including the row address of the weak row 0, the row address of the weak row 1, and the row address of the weak row 2) one by one. If the address of row b is different from the row addresses of the weak rows, the row address refreshed to the next row by the self-refresh is continuously determined.

[0110] In some embodiments, as shown in FIG. 10, Figure 6 Figure 1 Step S104 in FIG. 10 can be implemented by steps S1041 and S1042 in FIG. 11. Figure 6

[0111] Step S1041, in the management list, the starting flag corresponding to the weak row is confirmed.

[0112] In the embodiments of the present disclosure, for example, when refreshing to the weak row 0, the starting flag corresponding to the weak row 0 is searched according to the address of the weak row 0. It is judged whether the starting flag is the second value (that is, whether the starting flag is 1). If the starting flag is the second value, step S1042 is executed; if the starting flag is not the second value, no compensation refresh is performed.

[0113] Step S1042, starting from the self-refresh time of the weak row, every time the compensation time interval is passed, the weak row is refreshed once, until the compensation refresh number is reached, and the starting flag is reset to the first value.

[0114] In some embodiments, step S1042 includes S1042a and S1042b.

[0115] S1042a, the management list is traversed to determine the compensation refresh number and the compensation time interval corresponding to the weak row.

[0116] In the embodiments of the present disclosure, the management list is traversed to determine the compensation refresh number and the compensation time interval corresponding to the weak row according to the row address of the weak row. For example, the management list is traversed to determine that the compensation refresh number and the compensation time interval corresponding to the weak row 0 are 1 and 16 ms respectively according to the row address of the weak row 0.

[0117] S1042b, starting from the self-refresh time of the weak row, every time the compensation time interval is passed, the weak row is refreshed once, and in the management list, the compensation refresh number corresponding to the weak row is reduced by 1, until the compensation refresh number corresponding to the weak row is reduced to 0.

[0118] ​​In the embodiment of the present disclosure, the self-refresh and the compensation refresh are synchronized starting from the time when the self-refresh reaches the weak row, and the compensation refresh of the weak row is started. At the self-refresh time of the weak row, the compensation refresh of the weak row is performed every compensation time interval, and the compensation refresh number corresponding to the weak row in the management list is reduced by 1 until the compensation refresh number corresponding to the weak row is reduced to 0, and the compensation refresh is stopped.

[0119] For example, the self-refresh and the compensation refresh are synchronized starting from the time when the self-refresh reaches the weak row 0, and the compensation refresh of the weak row 0 is started. At the self-refresh time of the weak row 0, the compensation refresh of the weak row 0 is performed after 16 ms, and the compensation refresh number corresponding to the weak row in the management list is reduced by 1 until the compensation refresh number corresponding to the weak row is reduced to 0, and the compensation refresh is stopped. The compensation refresh method of the weak row 1 and the weak row 2 is similar to the compensation refresh method of the weak row 0, which will not be described here.

[0120] In the embodiment of the present disclosure, after the compensation refresh is stopped, the start flag is cleared, that is, the start flag is set to 0. In this way, by confirming whether the start flag is the second value, the compensation refresh of the weak row can be avoided repeatedly in the preset standard time interval, and after the compensation refresh is stopped, the start flag is cleared to the first value, so that the compensation refresh can be continued in the subsequent refresh process.

[0121] Based on this, for those particles with a small amount of storage units and slightly weak data retention capability, the compensation refresh method provided by the embodiment of the present disclosure can make them be used normally, thereby avoiding waste.

[0122] Figure 7 The schematic diagram of the compensation refresh device provided by the embodiment of the present disclosure is shown in FIG. 7. Figure 7 As shown in FIG. 7, the compensation refresh device 700 comprises:

[0123] The recording module 710 is configured to save the row address of the weak row and the corresponding compensation coefficient;

[0124] The compensation management module 720 is configured to obtain the row address of the weak row and the corresponding compensation coefficient from the recording module; determine the compensation time interval based on the compensation coefficient; match the self-refresh of the weak row according to the row address of the weak row, and determine the self-refresh time of the weak row; and perform the compensation refresh of the weak row according to the self-refresh time of the weak row and the compensation time interval.

[0125] In the embodiments of the present disclosure, the DRAM can be tested first to obtain the row address of the weak row, the minimum data retention time of the normal storage unit and the data retention time of the weak storage unit; during the test, the compensation coefficient corresponding to the weak row address (or the weak row) is obtained according to the minimum data retention time of the normal storage unit and the data retention time of the weak storage unit. Next, the row address of the weak row and the corresponding compensation coefficient obtained by the test are saved in the record module 710 (as shown in Figure 7 and Figure 8 indicated), that is, step S101a in the above-mentioned method embodiment is performed.

[0126] In the embodiments of the present disclosure, the compensation coefficient = minimum data retention time of normal storage unit / data retention time of weak storage unit (the result is rounded down), that is, the compensation coefficient is the minimum data retention time of the normal storage unit divided by the data retention time of the weak storage unit, and the quotient is rounded down.

[0127] In the embodiments of the present disclosure, a weak row record module includes two pieces of information, the row address of the weak row and the compensation coefficient corresponding to the row address of the weak row, as shown in Figure 9 The record module in the one-time programmable memory includes the row address of the weak row 0 (Weak Row Address 0) and the corresponding compensation coefficient 0, the row address of the weak row 1 (Weak Row Address 1) and the corresponding compensation coefficient 1, and the row address of the weak row 2 (Weak Row Address 2) and the corresponding compensation coefficient 2.

[0128] In some embodiments, please continue to refer to Figure 8 The compensation management module 720 includes: an initialization module 721 configured to obtain the row address of the weak row and the corresponding compensation coefficient; determine the compensation time interval based on the compensation coefficient; a judgment module 722 configured to match the self-refresh of the weak row according to the row address of the weak row, and determine the self-refresh time of the weak row; and an execution module 723 configured to compensate and refresh the weak row according to the self-refresh time of the weak row and the compensation time interval.

[0129] In the embodiments of the present disclosure, please continue to refer to Figure 8The initialization module 721 is configured to perform steps S101 and S102 in the method embodiments described above, i.e., determining the compensation refresh number and the compensation time interval according to the row address of the weak row in the weak row record module 710, the compensation coefficient corresponding to the row address of the weak row, and the preset standard time interval. The judgment module 722 is configured to perform step S103 in the method embodiments described above, i.e., determining whether the row address of the current row is a weak row. The execution module 723 is configured to perform step S104 in the method embodiments described above, i.e., based on the weak row being the current row, inserting the compensation refresh number and the compensation time interval according to the refresh time under the last normal refresh of the weak row, wherein the last normal refresh refers to the first of the two self-refreshes in the same row.

[0130] In some embodiments, the initialization module 721 is further configured to determine the compensation refresh number based on the compensation coefficient; and determine the compensation time interval based on the compensation refresh number and the preset standard time interval.

[0131] In some embodiments, the initialization module 721 is further configured to obtain the compensation refresh number by multiplying the compensation coefficient by a temperature coefficient; wherein the temperature coefficient is related to temperature, and the higher the temperature, the higher the value of the temperature coefficient. That is, the refresh compensation number = n * compensation coefficient, wherein n is the temperature coefficient. For example, when n is set to 1 at room temperature, the refresh compensation number = compensation coefficient.

[0132] In some embodiments, the initialization module 721 is further configured to obtain the compensation time interval by dividing the preset standard time interval by the sum of the compensation refresh number and 1. That is, the compensation time interval = minimum data retention time of normal storage unit / (refresh compensation number + 1). Wherein, the preset standard time interval is the minimum data retention time of the normal row that does not need to be compensated for refresh.

[0133] In some embodiments, please continue to refer to Figure 8 The compensation management module 720 further includes a register module 724 configured to register a management list; the management list includes a project information group corresponding to each weak row.

[0134] In some embodiments, the project information group includes the row address of the weak row, the compensation refresh number, the compensation time interval, and a preset start flag.

[0135] In the embodiments of the present disclosure, the register module is used to save the management list, and the management list includes a project information group of each weak row. For example, the management list includes a project information group 0, a project information group 1, and a project information group 2, each of which includes 4 record tables, and one record table is used to save one running parameter; wherein the row address of the weak row, the compensation refresh number, the compensation time interval, and the start flag are all called running parameters.

[0136] In some embodiments, please continue to refer to Figure 8 , the decision module 722 is further configured to match the row addresses of the weak rows with the current row address in turn to match the self-refresh of the weak rows in the process of self-refresh; the current row address is the row address of the current self-refresh.

[0137] In some embodiments, please continue to refer to Figure 8 , the decision module 722 is further configured to set the start flag to the first value in the management list; obtain the current row address; match the current row address with the row addresses of the weak rows one by one; if the current row address belongs to the row addresses of the weak rows, reset the compensation refresh number corresponding to the weak row and set the start flag corresponding to the weak row to the second value in the management list.

[0138] In the embodiments of the present disclosure, the start flag is set to the first value in the management list, wherein the first value can be 0. The row address in the external refresh instruction or the self-refresh circuit is obtained. In the process of self-refresh, the compensation refresh circuit monitors the row address of the row refreshed by the self-refresh circuit and matches and compares the row addresses one by one, if the row addresses are consistent, the compensation refresh number corresponding to the weak row is reset and the start flag corresponding to the weak row is set to the second value. Wherein the second value can be 1. If the row addresses are not consistent, the row address of the next row refreshed by the self-refresh is continuously monitored.

[0139] In some embodiments, please continue to refer to Figure 8 , the execution module 723 is further configured to confirm the start flag corresponding to the weak row in the management list; if the start flag is the second value, the self-refresh time of the weak row is taken as the start, and the weak row is refreshed once every compensation time interval until the compensation refresh number is reached, and the start flag is reset to the first value.

[0140] In the embodiments of the present disclosure, the start flag corresponding to the weak row is confirmed in the management list; it is judged whether the start flag is the second value (i.e. whether the start flag is 1). If the start flag is the second value, the compensation refresh is continuously executed, i.e. the self-refresh time of the weak row is taken as the start, and the weak row is refreshed once every compensation time interval until the compensation refresh number is reached, and the start flag is reset to the first value. If the start flag is not the second value, the compensation refresh is not performed.

[0141] In some embodiments, please continue to refer to Figure 8The execution module 723 is further configured to traverse the management list, determine the compensation refresh times and the compensation time interval corresponding to the weak row, and perform compensation refresh on the weak row every time the compensation time interval elapses since the self-refresh time of the weak row, and decrease the compensation refresh times corresponding to the weak row in the management list by 1 until the compensation refresh times corresponding to the weak row are decreased to 0.

[0142] In the embodiments of the present disclosure, the management list is traversed, the compensation refresh times and the compensation time interval corresponding to the weak row are determined according to the row address of the weak row, the self-refresh and the compensation refresh are synchronized since the time of the self-refresh to the weak row, and the compensation refresh on the weak row is started. At the self-refresh time of the weak row, the compensation refresh is performed on the weak row every time the compensation time interval elapses, and the compensation refresh times corresponding to the weak row in the management list are decreased by 1 until the compensation refresh times corresponding to the weak row are decreased to 0, and the compensation refresh is stopped.

[0143] In some embodiments, the initialization module 721 is further configured to determine whether the management list is empty, and the execution submodule is configured to not perform the compensation refresh if the management list is empty.

[0144] In the embodiments of the present disclosure, after the compensation management module is initialized, it can be determined whether there is a weak row record in the initialization module, and if yes, a signal is sent to the execution module to perform the following compensation refresh. If there is no weak row record, it means that there is no weak row in the DRAM, and the compensation refresh is not needed, and no signal is sent.

[0145] In the embodiments of the present disclosure, since the recording module and the compensation management module are added in the compensation refresh device, the weak row can be compensated and refreshed without changing the original refresh circuit in the DRAM, that is, the weak row is additionally increased with appropriate refresh times after the self-refresh time of the weak row, so as to ensure the stability of the data of the weak row, and further avoid the loss of data.

[0146] The descriptions of the above device embodiments are similar to those of the above method embodiments, and have similar beneficial effects to the method embodiments. In some embodiments, the device provided by the embodiments of the present disclosure has functions or includes modules which can be used to execute the methods described in the above method embodiments. For technical details not disclosed in the device embodiments of the present disclosure, please refer to the descriptions of the method embodiments of the present disclosure.

[0147] In addition, the embodiments of the present disclosure also provide a memory, as shown in Figure 10 The memory 100 includes the compensation refresh device 700.

[0148] In some embodiments, please continue to refer to Figure 10 The memory 100 includes a dynamic random access memory.

[0149] In other embodiments, with reference to Figure 10 , the memory 100 can include a magnetoresistive random access memory (MRAM), a static random access memory (SRAM), or a ferroelectric random access memory (FRAM).

[0150] In several embodiments provided in the present disclosure, it should be understood that the disclosed apparatus and method can be implemented in a non-targeted manner. The apparatus embodiments described above are merely illustrative, for example, the division of units is merely a logical function division, and actual implementation can have another division manner, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling between the various components shown or discussed.

[0151] The disclosed features in several method or apparatus embodiments provided in the present disclosure can be combined arbitrarily without conflict to obtain new method embodiments or apparatus embodiments.

[0152] The above is only some embodiments of the present disclosure, but the protection scope of the present disclosure is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present disclosure, which should be covered within the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be subject to the protection scope of the claims.

Claims

1. A compensation refresh method, characterized in that, the method comprises: obtaining a row address of a weak row and a corresponding compensation coefficient; determining a compensation time interval based on the compensation coefficient; matching self-refresh of the weak row according to the row address of the weak row, and determining a self-refresh time of the weak row; compensating refresh of the weak row according to the self-refresh time of the weak row and the compensation time interval; wherein, the determining of the compensation time interval based on the compensation coefficient comprises: determining a compensation refresh number based on the compensation coefficient; determining the compensation time interval based on the compensation refresh number and a preset standard time interval; the determining of the compensation refresh number based on the compensation coefficient comprises: multiplying the compensation coefficient by a temperature coefficient to obtain the compensation refresh number; wherein, the temperature coefficient is related to temperature, and the higher the temperature, the higher the value of the temperature coefficient. 2.The method of claim 1, characterized in that, the obtaining of the row address of the weak row and the corresponding compensation coefficient comprises: dividing a preset standard time interval by a data retention time of the weak row, and obtaining the corresponding compensation coefficient by rounding down the quotient; wherein, the preset standard time interval is the minimum data retention time of a normal row that does not need to perform the compensation refresh. 3.The method of claim 1, characterized in that, the row address of the weak row, the compensation refresh number and the compensation time interval are all stored in a management list; and the management list also stores a start flag. 4.The method of claim 3, characterized in that, the matching of the self-refresh of the weak row according to the row address of the weak row comprises: in the process of self-refresh, sequentially matching the row address of the weak row with a current row address to match the self-refresh of the weak row; and the current row address is a row address of a row currently performing self-refresh. 5.The method of claim 4, characterized in that, the sequentially matching of the row address of the weak row with the current row address comprises: setting the start flag to a first value in the management list; obtaining the current row address; matching the current row address with the row address of the weak row one by one; if the current row address belongs to the row address of the weak row, resetting the compensation refresh number corresponding to the weak row in the management list, and setting the start flag corresponding to the weak row to a second value. 6.The method of claim 5, characterized in that, the compensating refresh of the weak row according to the self-refresh time of the weak row and the compensation time interval comprises: confirming the start flag corresponding to the weak row in the management list; if the start flag is the second value, starting from the self-refresh time of the weak row, performing the compensation refresh of the weak row every time the compensation time interval elapses, until the compensation refresh number is reached, and resetting the start flag to the first value. 7.The method of claim 6, characterized in that, starting from the self-refresh time of the weak row, every time the compensation time interval is passed, the compensation refresh is performed on the weak row once until the compensation refresh number is reached, including: traversing the management list to determine the compensation refresh number and the compensation time interval corresponding to the weak row; starting from the self-refresh time of the weak row, every time the compensation time interval is passed, the compensation refresh is performed on the weak row once, and in the management list, the compensation refresh number corresponding to the weak row is reduced by 1 until the compensation refresh number corresponding to the weak row is reduced to 0.

8. The method of claim 3, wherein, after the row address and the corresponding compensation coefficient of the weak row are obtained, the method further comprises: determining whether the management list is empty; if the management list is empty, the compensation refresh is not performed.

9. A compensation refresh device, comprising: a recording module configured to save the row address and the corresponding compensation coefficient of the weak row; a compensation management module configured to obtain the row address and the corresponding compensation coefficient of the weak row from the recording module, determine a compensation time interval based on the compensation coefficient, match the self-refresh of the weak row according to the row address of the weak row, and determine the self-refresh time of the weak row, and perform compensation refresh on the weak row according to the self-refresh time of the weak row and the compensation time interval. wherein the compensation time interval is determined based on the compensation coefficient, comprising: determining a compensation refresh number based on the compensation coefficient; determining the compensation time interval based on the compensation refresh number and a preset standard time interval; the compensation refresh number is determined based on the compensation coefficient, comprising: multiplying the compensation coefficient by a temperature coefficient to obtain the compensation refresh number; wherein the temperature coefficient is related to temperature, and the higher the temperature, the higher the value of the temperature coefficient.

10. The device of claim 9, wherein the compensation management module comprises: an initialization module configured to obtain the row address and the corresponding compensation coefficient of the weak row, and determine the compensation time interval based on the compensation coefficient; a decision module configured to match the self-refresh of the weak row according to the row address of the weak row, and determine the self-refresh time of the weak row; an execution module configured to perform compensation refresh on the weak row according to the self-refresh time of the weak row and the compensation time interval.

11. The device of claim 9, wherein the compensation management module further comprises: a register module configured to register a management list; the management list comprises a project information group corresponding to each weak row.

12. The device of claim 11, wherein the project information group comprises: the row address of the weak row, the compensation refresh number, the compensation time interval, and a preset start flag.

13. A semiconductor memory, comprising the compensation refresh device of any one of claims 9 to 12. ​ ​ ​ ​ ​ 14. The memory of claim 13, wherein: the memory comprises a dynamic random access memory.

Images