An adaptive energy consumption optimization storage chip control method, device and medium

Abstract

The application discloses a kind of self-adapting energy consumption optimization storage chip control method, equipment and medium, it is related to computer structure technical field, including, the running parameter in candidate parameter set is carried out feasibility determination, and reliable behavior prediction is carried out through state evolution mapping relationship, and form extended feasible set;Running cost and maintenance cost comprehensive evaluation are carried out to extended feasible set, and multi-objective optimization sorting is executed, and output target running parameter combination;Prediction deviation analysis is carried out to execution result package and updates prediction mapping relationship and historical parameter library, and output adaptive control basis;The application realizes that control model continues self-evolution with chip aging process by collecting the historical behavior data of target storage partition, current junction temperature information and current state distribution, and considers energy consumption optimization and the adaptive storage chip control of life reliability.

Description

Technical Field

[0001] This invention relates to the field of computer architecture technology, and in particular to an adaptive energy-optimized memory chip control method, device, and medium. Background Technology

[0002] With the large-scale application of high-density non-volatile memory chips (such as 3D NAND, PCM, and novel resistive random access memory), memory subsystems have evolved from simple read / write functional units to intelligent components with complex control logic and refined management mechanisms. Existing technologies generally integrate functional modules such as temperature monitoring, voltage regulation, write strategy selection, and wear leveling within the memory chip controller. Operating parameters are dynamically adjusted through preset strategies or empirical thresholds to achieve a trade-off between performance, reliability, and power consumption. For example, voltage compensation mechanisms based on temperature thresholds, lifetime management mechanisms based on erase / write cycles, and power consumption regulation strategies based on load intensity have all become mainstream control methods.

[0003] Existing technologies often rely on decision-making based on single physical quantity thresholds or a limited number of operational indicators when establishing control strategies. This lacks a systematic characterization of the "state distribution" of storage partitions under different historical behavioral contexts, making it difficult to reflect the complex statistical characteristics of storage cell groups under the influence of charge retention, aging drift, and thermal stress. Furthermore, existing solutions often match parameters based on the current instantaneous state or preset rules, lacking joint analysis of historical operating trajectories, junction temperature variations, and state distribution evolution trends. This makes it difficult to achieve refined energy consumption regulation and reliability prediction in scenarios where long-term aging and short-term environmental fluctuations overlap. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an adaptive energy consumption optimization memory chip control method to solve the problem of difficulty in finely adjusting energy consumption and predicting reliability in scenarios where long-term aging and short-term environmental fluctuations overlap.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides an adaptive energy consumption optimization memory chip control method, which includes: collecting historical behavior sets and current junction temperature information, reading the current state distribution, extracting the state distribution feature morphology, and forming a real-time state packet;

[0008] Perform historical parameter retrieval and resource granular matching on real-time status packets, and combine them with environment matching records to output a set of candidate parameters;

[0009] Feasibility is determined for the operating parameters in the candidate parameter set, and reliability behavior is predicted through state evolution mapping relationship to form an extended feasible set;

[0010] The extended feasible set is comprehensively evaluated for operational and maintenance costs, and multi-objective optimization ranking is performed to output the target combination of operational parameters;

[0011] Based on the target combination of operating parameters, the target operation is executed and the actual energy consumption trajectory and post-operation state distribution are collected simultaneously to form an execution result package.

[0012] Perform prediction deviation analysis on the execution result package and update the prediction mapping relationship and historical parameter library to output the basis for adaptive control.

[0013] As a preferred embodiment of the adaptive energy consumption optimization memory chip control method of the present invention, the steps of collecting historical operating behavior sets and current junction temperature information, reading the current state distribution, extracting the state distribution feature morphology, and forming a real-time state packet are as follows.

[0014] After receiving the storage operation trigger instruction of the target storage partition, the system retrieves historical operation count information, historical control parameter records and corresponding historical state distributions and integrates them to form a set of running historical behaviors. At the same time, it reads the current junction temperature information fed back by the internal temperature sensor of the chip.

[0015] Perform a fast scan on the target storage partition to obtain the current state distribution, and perform statistical processing to obtain the state distribution feature shape;

[0016] The historical behavior set, current junction temperature information, current state distribution, and state distribution feature morphology are integrated and encapsulated to generate a real-time state package.

[0017] As a preferred embodiment of the adaptive energy consumption optimization memory chip control method of the present invention, the specific steps of performing historical parameter retrieval and resource granularity matching on the real-time status packet, and combining it with environmental matching records to output a candidate parameter set, are as follows.

[0018] Collect historical control parameter records and perform similarity retrieval based on the state distribution characteristics in the real-time state packet to obtain a set of historical control parameter records;

[0019] Based on the historical operation behavior set, the historical operation count information in the historical control parameter record set is mapped to the count interval and the interval is matched. At the same time, based on the current junction temperature information, the temperature interval markers in the historical control parameter record set are matched, and the environmental matching record is output.

[0020] Perform resource-granular matching on the environment matching records, filter the historical control parameter record set, and extract the corresponding control parameter combinations as candidate parameter sets.

[0021] As a preferred embodiment of the adaptive energy consumption optimization memory chip control method of the present invention, the steps of determining the feasibility of operating parameters in the candidate parameter set and predicting reliability behavior through state evolution mapping to form an extended feasible set are as follows.

[0022] Based on the current state distribution and its characteristic shape, calculate the central tendency and dispersion range of the state distribution after the operation of the running parameters in the candidate parameter set, and compare them with the decision boundary parameters to output a subset of feasible parameters.

[0023] For each parameter combination in the feasible parameter subset, the state evolution mapping relationship is invoked to determine the state evolution trajectory under the conditions of the current junction temperature information and the historical operating state data of the target storage partition, and the state changes over time is deduced.

[0024] Based on the state changes over time and the decision boundary parameters, determine whether the state will cross the decision boundary parameters within the target lifetime, eliminate parameter combinations that cross the decision boundary parameters, and output an extended feasible set.

[0025] As a preferred embodiment of the adaptive energy consumption optimization memory chip control method of the present invention, the state evolution mapping relationship is established by collecting long-term operational data on the state distribution characteristics and continuously tracking the state change trajectory over time and the corresponding decision failure time data under different temperature conditions and historical operating conditions, and then conducting statistical analysis.

[0026] As a preferred embodiment of the adaptive energy consumption optimization memory chip control method of the present invention, the specific steps of comprehensively evaluating the operating cost and maintenance cost of the extended feasible set, performing multi-objective optimization ranking, and outputting the target operating parameter combination are as follows.

[0027] Read the operation process energy consumption record corresponding to each parameter combination in the extended feasible set, combine the operation process energy consumption record with the current junction temperature information and perform conversion to obtain the operation cost assessment value;

[0028] Based on the state changes over time and the decision boundary parameters, the frequency of maintenance operations within the target life cycle is determined, and combined with the single energy consumption record, the maintenance cost assessment value is output.

[0029] The operational cost assessment value and the maintenance cost assessment value are jointly evaluated to form a comprehensive cost value. The parameter combinations in the extended feasible set are sorted according to the comprehensive cost value and multi-objective optimization is performed. The optimal parameter combination is selected as the target operational parameter combination.

[0030] In a preferred embodiment of the adaptive energy consumption optimization memory chip control method of the present invention, the steps for forming the execution result package are as follows:

[0031] The target storage operation process is driven by the combination of target operating parameters, and the actual energy consumption trajectory is obtained by current sampling and time counting.

[0032] Perform a fast scan on the target storage partition to obtain the post-operation state distribution. Integrate and encapsulate the actual energy consumption trajectory, post-operation state distribution, target operating parameter combination, and current junction temperature information to generate an execution result package.

[0033] As a preferred embodiment of the adaptive energy consumption optimization memory chip control method of the present invention, the specific steps of performing prediction deviation analysis on the execution result package and updating the prediction mapping relationship and historical parameter library to output the adaptive control basis are as follows.

[0034] Extract the actual energy consumption trajectory and the post-operation state distribution from the execution result package, and align them with the operation cost assessment value and the state change over time, respectively, to calculate the energy consumption deviation between the actual energy consumption trajectory and the operation cost assessment value.

[0035] The central tendency and discrete measure in the state distribution after the operation are compared with the central tendency and discrete measure in the state change over time, and the state distribution deviation is calculated.

[0036] Based on the energy consumption deviation and the state distribution deviation, the correlation weight of the state distribution characteristic shape in parameter feasibility determination and reliability behavior prediction is adjusted to obtain an adaptive decision basis.

[0037] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the adaptive energy consumption optimization memory chip control method as described in the first aspect of the present invention.

[0038] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the adaptive energy consumption optimization storage chip control method as described in the first aspect of the present invention.

[0039] The beneficial effects of this invention are as follows: by collecting historical behavior data, current junction temperature information and current state distribution of the target storage partition, the state evolution trajectory is further deduced based on the predictive mapping relationship between the state distribution characteristics and aging and retention behavior, and parameter combinations that may cross the decision boundary in the future are eliminated. The reliability risk is transformed from ex-post discovery to ex-ante screening, realizing the continuous self-evolution of the control model with the chip aging process, and adaptive memory chip control that takes into account both energy consumption optimization and life reliability. Attached Figure Description

[0040] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 A flowchart of a method for optimizing memory chip control for adaptive energy consumption.

[0042] Figure 2 A flowchart for generating real-time status packets.

[0043] Figure 3 A flowchart for generating an expanded feasible set.

[0044] Figure 4 A flowchart for outputting the adaptive decision-making basis.

[0045] Figure 5 This is a magnified comparison and discrimination window diagram of the differences in the morphological characteristics of state distribution features.

[0046] Figure 6 A comparison chart of statistical results of state distribution under different state distribution characteristics of the target storage partition. Detailed Implementation

[0047] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0048] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0049] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0050] Reference Figures 1-6 As one embodiment of the present invention, this embodiment provides an adaptive power consumption optimization memory chip control method, including the following steps:

[0051] S1. Collect historical behavior data and current junction temperature information, read the current state distribution, extract the state distribution feature morphology, and form a real-time state packet.

[0052] S1.1. After receiving the storage operation trigger instruction of the target storage partition, retrieve the historical operation count information, historical control parameter records and corresponding historical state distribution, and integrate them to form a set of running historical behaviors. At the same time, read the current junction temperature information fed back by the internal temperature sensor of the chip.

[0053] Specifically, after receiving the storage operation trigger command of the target storage partition, the controller locates the target storage partition identifier and retrieves the historical operation count information (historical operation count information is used to characterize the cumulative electrical stress level of the target storage partition, including the cumulative number of erases, cumulative number of programming, and cumulative number of read operations, etc.) and historical control parameter records (historical control parameter records are used to characterize the control strategy adopted during historical operation, including erase voltage parameters, erase pulse width parameters, erase pulse count parameters, and programming voltage parameters, etc.). It reads the historical state distribution associated with the historical control parameter records and performs consistency verification with the historical operation count information to exclude missing or abnormal records. It integrates the historical operation count information, historical control parameter records, and historical state distribution to form an operating history behavior set. At the same time, it reads the current junction temperature information from the internal temperature sensor of the chip and performs time correspondence with the operating history behavior set to form the operating history behavior set and the current junction temperature information.

[0054] S1.2. Perform a fast scan on the target storage partition to obtain the current state distribution and perform statistical processing to obtain the state distribution feature shape.

[0055] Specifically, after obtaining the historical behavior set and current junction temperature information, the controller performs a fast scan operation on the target storage partition. The fast scan operation reads the storage state of the target storage partition page by page, outputs a read result set, cleans the read result set to remove read anomalies, and then performs statistical processing on the read result set to generate the current state distribution. The current state distribution is used to characterize the storage state distribution of the target storage partition under the current junction temperature information. Based on the current state distribution, the central tendency and discrete measure are calculated, and morphological features reflecting the changes in the distribution shape are extracted. The central tendency, discrete measure, and morphological features together constitute the state distribution feature shape, which is a set of features that can characterize the changes in the distribution shape after statistical processing of the current state distribution, including the central tendency and discrete measure.

[0056] It should be noted that the central tendency is defined as the weighted average of the number of readouts and the voltage value of each readout voltage interval in the current state distribution statistics, and the expression is:

[0057] ;

[0058] in, Indicates the quantity of central tendency. Indicates the voltage range number index. Indicates the total number of voltage ranges. Indicates the first The voltage representative value of each readout voltage range Indicates the first The number of readouts for each readout voltage range.

[0059] Discreteness measure is defined as the degree of dispersion of the current state distribution statistics relative to the central tendency measure, and its expression is:

[0060] ;

[0061] in, It represents a discrete metric.

[0062] S1.3. Integrate and encapsulate the historical behavior set, current junction temperature information, current state distribution, and state distribution feature morphology to generate a real-time state package.

[0063] Specifically, the consistency check is performed on the historical behavior set, current junction temperature information, current state distribution, and state distribution feature to confirm that the target storage partition identifier corresponds to the time. The integrity check is also performed on the historical operation count information and historical control parameter records in the historical behavior set to confirm the continuity of the records. The controller integrates and encapsulates the historical behavior set, current junction temperature information, current state distribution, and state distribution feature. The integration and encapsulation includes attaching the corresponding target storage partition identifier and acquisition time stamp to each item and forming a structured record that can be directly read by historical parameter retrieval and resource granularity matching, generating a real-time status package.

[0064] Furthermore, such as Figure 5 As shown, under the same junction temperature information and the same historical operation count conditions, the target storage partition may still exhibit different state distribution statistical results and state distribution characteristics. By statistically processing the current state distribution and extracting the central tendency and discrete measure, the true charge retention state of the storage cell group can be accurately reflected. Relying solely on temperature and historical stress information cannot accurately characterize the current storage state. However, this invention, by introducing state distribution characteristics, realizes real-time state perception of the target storage partition, providing a reliable basis for historical parameter retrieval and parameter selection.

[0065] S2. Perform historical parameter retrieval and resource granularity matching on the real-time status packet, and combine it with environment matching records to output a set of candidate parameters.

[0066] S2.1. Collect historical control parameter records and perform similarity retrieval based on the state distribution characteristics in the real-time state packet to obtain a set of historical control parameter records.

[0067] Specifically, the system reads historical control parameter records that match the target storage partition identifier, and simultaneously reads the historical state data distribution statistics and historical operation count information associated with the historical control parameter records. This completes the time correspondence and integrity verification between the historical control parameter records and the target storage partition's historical behavior data. It then performs feature extraction processing on the historical state data distribution statistics to obtain central tendency, discrete measure, and morphological features with the same structure as the state distribution feature shape. Finally, it performs similarity measurement calculations on the feature extraction results associated with the historical control parameter records and the state distribution feature shape in the real-time state packet. Specifically, it performs difference calculations on the central tendency, the discrete measure, and the state distribution feature shape, and then performs vector distance calculations. The difference results and vector distance results are then weighted and summed to obtain a comprehensive similarity measurement value. Finally, the system sorts and filters these values ​​from highest to lowest to form a set of historical control parameter records.

[0068] S2.2. Based on the historical operation behavior set, map the historical operation count information in the historical control parameter record set to the count interval and perform interval matching. At the same time, based on the current junction temperature information, match the temperature interval markers in the historical control parameter record set and output the environment matching record.

[0069] Specifically, the system reads the historical operation behavior set and current junction temperature information from the real-time status packet, and maps the current junction temperature information to a temperature range as a temperature matching benchmark. At the same time, it extracts historical operation count information from the historical operation behavior set and maps it to a count range as a count matching benchmark. It reads the temperature range marker and historical operation count information range marker corresponding to each historical control parameter record in the historical control parameter record set one by one, performs range matching between the historical operation count information range marker and the count range to eliminate historical control parameter records with inconsistent count ranges, and then matches the temperature range marker with the temperature range to eliminate historical control parameter records with inconsistent temperature ranges. The historical control parameter records that simultaneously meet the conditions of count range matching and temperature range matching are retained and output as environmental matching records.

[0070] S2.3. Perform resource-granularity matching on the environment matching records, filter the historical control parameter record set, and extract the corresponding control parameter combinations as candidate parameter sets.

[0071] Specifically, the target storage partition identifier is read from the real-time status packet, and the corresponding resource granularity level (including page level, block level, plane level, chip level, or storage array level) is determined. The resource granularity marker of each historical control parameter record in the environment matching record is read one by one and compared with the resource granularity level of the target storage partition. When the target storage partition belongs to a certain resource granularity level, only historical control parameter records whose resource granularity marker is at the same resource granularity level and within the same physical domain as the target storage partition are retained. For example, when the target storage partition is at the block level under Plane0, only block-level historical control parameter records also located under Plane0 are retained. Historical control parameter records with different resource granularity levels or different physical domains are removed. The retained historical control parameter records are summarized to form a set of historical control parameter records with resource granularity matching. Subsequently, the control parameter combinations associated with each historical control parameter record are extracted one by one from the set of historical control parameter records with resource granularity matching. Deduplication processing is performed on the control parameter combinations to eliminate the influence of duplicate control parameter combinations. The deduplicated control parameter combinations are summarized and output as a candidate parameter set.

[0072] S3. Determine the feasibility of the operating parameters in the candidate parameter set, and predict the reliability behavior through the state evolution mapping relationship to form an extended feasible set.

[0073] S3.1. Based on the current state distribution and the characteristic shape of the state distribution, calculate the central tendency and dispersion range of the state distribution after the operation of the running parameters in the candidate parameter set, and compare them with the decision boundary parameters to output a feasible parameter subset.

[0074] Specifically, based on the current state distribution and its characteristic features, the changes in the state distribution after operation are evaluated for each candidate parameter combination. This includes reading the historical state distribution and its characteristic features stored in the historical parameter library corresponding to the candidate parameter combinations, comparing them with the current state distribution, and calculating the range of central tendency and dispersion of the state distribution after the execution of the candidate parameter combinations. Subsequently, the range of central tendency and dispersion are compared with the decision boundary parameters item by item, eliminating candidate parameter combinations that cause the state distribution after operation to exceed the limits of the decision boundary parameters, retaining candidate parameter combinations that meet the constraints of the decision boundary parameters, and outputting a subset of feasible parameters.

[0075] It should be noted that the decision boundary parameters are historical state distributions and decision failure time records that are continuously stored in the historical parameter library under different combinations of operating control parameters during long-term operation. When the central tendency or discrete measure in the state distribution characteristic exceeds a certain range, the decision failure time is more likely to occur earlier. Based on the correspondence between a large number of historical state distributions and decision failure time records, interval statistical analysis is performed on the central tendency and discrete measure in the state distribution characteristic to obtain the maximum allowable range of change without the occurrence of earlier decision failure time. The allowable range is fixed and saved as decision boundary parameters and written into the historical parameter library for direct use in subsequent parameter feasibility determination.

[0076] S3.2. For each parameter combination in the feasible parameter subset, call the state evolution mapping relationship, determine the state evolution trajectory under the conditions of the current junction temperature information and the historical operating state data of the target storage partition, and deduce the state change over time.

[0077] Specifically, a reliability behavior prediction operation is performed on each parameter combination in the feasible parameter subset. This operation reads the state evolution mapping relationship and performs condition matching by combining the current junction temperature information with the historical state distribution in the historical operation behavior set. Based on the state distribution characteristics corresponding to the parameter combination, the temperature range entry corresponding to the current junction temperature information and the historical stress range entry corresponding to the historical operating state data conditions of the target storage partition are located in the state evolution mapping relationship. (Historical operating state data conditions refer to those stored in the historical parameter database at the same time as a certain historical control parameter record, which can characterize the physical stress and environmental stress level of the storage unit at that time.) A set of constraint information is used to filter historical samples with consistent physical stress background during prediction, and to locate the morphological interval entries corresponding to the state distribution characteristics. The intersection of the three determines the state evolution trajectory corresponding to the parameter combination. The state evolution trajectory refers to the trend path of the continuous change of the state distribution characteristics over time under the conditions of the operating parameter combination, current junction temperature information and operating history behavior set. It is used to characterize the state change process of the target storage partition under aging and retention, and serves as the basis for lifetime cycle extrapolation and parameter feasibility determination. Subsequently, the state distribution changes at multiple time points within the target lifetime cycle are extrapolated along the state evolution trajectory, and the state changes over time are output.

[0078] It should be noted that the state evolution mapping relationship is established by long-term operation data collection of state distribution characteristics obtained under different combinations of operating control parameters, and continuous tracking of state changes over time and corresponding decision failure time data under different temperature conditions and historical operating conditions. Through statistical analysis, the correspondence between state distribution characteristics, state drift behavior (state drift behavior is obtained by repeatedly performing fast read scans on the same target storage partition at different time points and extracting state distribution characteristics) and retention failure time is established.

[0079] S3.3. Based on the state changes over time and the decision boundary parameters, determine whether the state will cross the decision boundary parameters within the target lifetime, eliminate parameter combinations that cross the decision boundary parameters, and output the extended feasible set.

[0080] Specifically, the state changes over time are described in chronological order as a distribution of state at multiple time points, and the central tendency boundary and dispersion boundary corresponding to the decision boundary parameters are read one by one. At each time point, the controller compares the central tendency and dispersion of the state changes over time with the decision boundary parameters. If the central tendency or dispersion exceeds the allowable range of the decision boundary parameters at any time point, the corresponding parameter combination is determined to have a risk of crossing the decision boundary parameters within the target life cycle and the parameter combination is removed. Parameter combinations that do not cross the decision boundary parameters at any time point are retained, and the retained parameter combinations are summarized and output as an extended feasible set.

[0081] Furthermore, such as Figure 6 As shown, by locally magnifying and comparing the distribution features of different states, it is verified that within the discrimination window area, different forms show significant differences in this interval. This area corresponds to the sensitive interval compared with the decision boundary parameters during the parameter feasibility determination process. By predicting the state distribution after performing operations on the candidate parameter combinations and comparing it with the decision boundary parameters, parameter combinations that may cause the state to cross the decision boundary can be identified before the parameters are actually executed, thus achieving the pre-screening of reliability risks.

[0082] S4. Perform a comprehensive evaluation of the operational and maintenance costs of the extended feasible set, and perform multi-objective optimization sorting to output the target combination of operational parameters.

[0083] S4.1. Read the operation process energy consumption record corresponding to each parameter combination in the extended feasible set, combine the operation process energy consumption record with the current junction temperature information and perform conversion to obtain the operation cost assessment value.

[0084] Specifically, for each parameter combination in the extended feasible set, an operational cost assessment operation is performed. The operational cost assessment operation uses the parameter combination as the search key to read the corresponding operational process energy consumption record from the historical parameter database. The operational process energy consumption record contains energy consumption information formed by current sampling results and time counting results when executing the target storage operation process. Then, the current junction temperature information is read and mapped to a temperature range. The controller performs temperature conversion on the operational process energy consumption record based on the temperature range to correct the impact of temperature changes on energy consumption, and obtains the energy consumption estimation result corresponding to the current junction temperature information condition. The energy consumption estimation result is associated with the parameter combination and output as the operational cost assessment value.

[0085] S4.2. Based on the state changes over time and the decision boundary parameters, determine the frequency of maintenance operations within the target life cycle, and output the maintenance cost assessment value by combining the single energy consumption record.

[0086] Specifically, the target lifespan is divided into multiple evaluation time nodes. At each evaluation time node, the central tendency and discrete measure corresponding to the change of state over time are read. At each evaluation time node, the central tendency and discrete measure are compared with the decision boundary parameters to determine the degree to which the state approaches the decision boundary parameters. When the central tendency or discrete measure reaches the trigger condition of approaching the decision boundary parameters, the controller includes the evaluation time node in the number of maintenance operations and calculates the frequency of maintenance operations within the target lifespan accordingly. Then, the controller reads the single energy consumption record corresponding to the maintenance operation from the historical parameter database and performs temperature conversion by combining it with the current junction temperature information. The frequency of maintenance operations and the conversion results of the single energy consumption record are combined to calculate and output the maintenance cost assessment value.

[0087] S4.3. Jointly evaluate the operational cost assessment value and the maintenance cost assessment value to form a comprehensive cost value. Perform multi-objective optimization sorting on the parameter combinations in the extended feasible set according to the comprehensive cost value, and select the optimal parameter combination as the target operational parameter combination.

[0088] Specifically, a joint evaluation operation is performed on the operational cost assessment value and the maintenance cost assessment value. The joint evaluation operation normalizes the operational cost assessment value and the maintenance cost assessment value to the same dimension, eliminating the impact of the difference in dimensions on the comparison results. Then, the normalized operational cost assessment value and the maintenance cost assessment value are weighted and summed to form a comprehensive cost value. The weighting ratio is determined based on the statistical analysis of the long-term operational results corresponding to different parameter combinations in the historical parameter database. A correspondence is established between the comprehensive cost value and the parameter combination, and all parameter combinations in the extended feasible set are sorted in ascending order of comprehensive cost value. The parameter combination with the smallest comprehensive cost value is selected as the target operational parameter combination.

[0089] S5. Based on the target combination of operating parameters, execute the target operation and simultaneously collect the actual energy consumption trajectory and the state distribution after the operation to form an execution result package.

[0090] S5.1. Drive the target storage operation process according to the target operating parameter combination, and obtain the actual energy consumption trajectory through current sampling and time counting.

[0091] Specifically, the operation control parameters in the target operating parameter combination are read and written into the control register corresponding to the target storage operation process. Then, the read and write execution sequence of the target storage partition is triggered according to the timing of the target storage operation process, and the operation control parameters are kept unwritten during the execution sequence to ensure operational consistency. During the execution of the target storage operation process, the current sampling signal of the power supply path is continuously collected and the corresponding time count value is recorded for each current sampling. The current sampling signal and the time count value are time-aligned and integrated to obtain a sequence record of energy consumption changing over time, and the actual energy consumption trajectory is output.

[0092] It should be noted that the expression for calculating the actual energy consumption trajectory is:

[0093] ;

[0094] in, This represents the total energy consumption corresponding to the actual energy consumption trajectory. This represents the total number of sampling points in the actual energy consumption trajectory. Indicates the first Secondary current sampling value, Indicates the current sampling sequence number index. This indicates the voltage value of the power supply path. Indicates the first The time interval corresponding to the next current sampling value.

[0095] S5.2. Perform a fast scan on the target storage partition to obtain the post-operation state distribution. Integrate and encapsulate the actual energy consumption trajectory, post-operation state distribution, target operating parameter combination, and current junction temperature information to generate an execution result package.

[0096] Specifically, a fast scan is performed on the target storage partition through the read path to obtain the post-operation status read data. The target storage partition is read and the post-operation status read data set is output. The post-operation status read data set is cleaned to remove read anomalies and then statistically processed. The statistical processing is used to generate the post-operation status distribution and establish a correlation between the post-operation status distribution and the target storage partition identifier and the acquisition time stamp. Subsequently, a consistency check is performed on the actual energy consumption trajectory, post-operation status distribution, target operating parameter combination, and current junction temperature information to confirm that the target storage partition identifier and time correspondence are consistent. The actual energy consumption trajectory, post-operation status distribution, target operating parameter combination, and current junction temperature information are integrated and encapsulated according to the field order to generate an execution result package.

[0097] S6. Perform prediction deviation analysis on the execution result package and update the prediction mapping relationship and historical parameter library, and output the basis for adaptive control.

[0098] S6.1. Extract the actual energy consumption trajectory and the post-operation state distribution from the execution result package, and align them with the operation cost assessment value and the state change over time, respectively, and calculate the energy consumption deviation between the actual energy consumption trajectory and the operation cost assessment value.

[0099] Specifically, the actual energy consumption trajectory and post-operation state distribution are read from the execution result package, and the target storage partition identifier and collection time mark are aligned. Then, the operation cost evaluation value associated with the target operation parameter combination is read, and the actual energy consumption trajectory is converted into an energy consumption representation form consistent with the operation cost evaluation value. The difference between the energy consumption corresponding to the actual energy consumption trajectory and the operation cost evaluation value is calculated to obtain the energy consumption deviation.

[0100] S6.2. Compare the central tendency and discrete measure in the state distribution after the operation with the central tendency and discrete measure in the state change over time, and calculate the state distribution deviation.

[0101] Simultaneously, the state changes over time associated with the target operating parameter combination are read and the time node corresponding to the state distribution acquisition time after the operation is selected. The difference between the central tendency and the discrete measure in the state distribution after the operation and the central tendency and the discrete measure in the state changes over time is calculated to obtain the state distribution deviation.

[0102] S6.3. Based on the energy consumption deviation and the state distribution deviation, adjust the correlation weight of the state distribution characteristic form in parameter feasibility determination and reliability behavior prediction to obtain the adaptive decision basis.

[0103] Specifically, the associated weight records corresponding to the state distribution feature patterns are read, and the associated weights related to operating costs are adjusted according to the energy consumption deviation. The associated weights related to parameter feasibility determination and reliability behavior prediction are also adjusted according to the state distribution deviation, so that the influence of the state distribution feature patterns in subsequent parameter feasibility determination and reliability behavior prediction is corrected as the actual deviation changes. The actual energy consumption trajectory and the state distribution after operation are associated with the target operating parameter combination and the current junction temperature information and written into the historical parameter database to supplement historical operating records. Simultaneously, the parameters in the state evolution mapping relationship are updated according to the state distribution deviation, forming an adaptive decision-making basis for subsequent operating parameter decisions. This adaptive decision-making basis refers to directly calling the updated associated weights and state evolution mapping relationships when performing historical parameter retrieval, parameter feasibility determination, reliability behavior prediction, and multi-objective optimization sorting under similar historical operating behavior sets and current junction temperature information. This allows the operating parameter decision results to be automatically corrected according to the actual operating results and continuously aligned with the chip's current aging stage and thermal environment state.

[0104] This embodiment also provides a computer device applicable to the adaptive energy consumption optimization memory chip control method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the adaptive energy consumption optimization memory chip control method proposed in the above embodiment.

[0105] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0106] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the adaptive power-optimized storage chip control method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0107] In summary, this invention, by collecting historical behavior data, current junction temperature information, and current state distribution of the target storage partition, further deduces the state evolution trajectory based on the predictive mapping relationship between the state distribution characteristics and aging and retention behaviors, and eliminates parameter combinations that may cross the decision boundary in the future, transforms reliability risks from ex-post discovery to ex-ante screening. This enables adaptive storage chip control where the control model continuously evolves with the chip aging process, taking into account both energy consumption optimization and lifetime reliability.

[0108] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An adaptive energy-optimized memory chip control method, characterized in that: include, Collect historical behavior data and current junction temperature information, read the current state distribution, extract the state distribution features, and form a real-time state packet; Perform historical parameter retrieval and resource granular matching on real-time status packets, and combine them with environment matching records to output a set of candidate parameters; Feasibility is determined for the operating parameters in the candidate parameter set, and reliability behavior is predicted through state evolution mapping relationship to form an extended feasible set; The extended feasible set is comprehensively evaluated for operational and maintenance costs, and multi-objective optimization ranking is performed to output the target combination of operational parameters; Based on the target combination of operating parameters, the target operation is executed and the actual energy consumption trajectory and post-operation state distribution are collected simultaneously to form an execution result package. Perform prediction deviation analysis on the execution result package and update the prediction mapping relationship and historical parameter library to output the basis for adaptive control. The specific steps for determining the feasibility of operating parameters in the candidate parameter set and predicting reliability behavior through state evolution mapping to form an extended feasible set are as follows. Based on the current state distribution and its characteristic shape, calculate the central tendency and dispersion range of the state distribution after the operation of the running parameters in the candidate parameter set, and compare them with the decision boundary parameters to output a subset of feasible parameters. For each parameter combination in the feasible parameter subset, the state evolution mapping relationship is invoked to determine the state evolution trajectory under the conditions of the current junction temperature information and the historical operating state data of the target storage partition, and the state changes over time is deduced. Based on the state changes over time and the decision boundary parameters, determine whether the state will cross the decision boundary parameters within the target lifetime, eliminate parameter combinations that cross the decision boundary parameters, and output an extended feasible set. The state evolution mapping relationship is established by collecting long-term operational data on the state distribution characteristics and continuously tracking the state change trajectory over time and the corresponding decision failure time data under different temperature conditions and historical operating conditions, and then conducting statistical analysis.

2. The adaptive energy consumption optimization memory chip control method as described in claim 1, characterized in that: The process involves collecting historical behavior data and current junction temperature information, reading the current state distribution, extracting the state distribution features, and forming a real-time state packet. The specific steps are as follows: After receiving the storage operation trigger instruction of the target storage partition, the system retrieves historical operation count information, historical control parameter records and corresponding historical state distributions and integrates them to form a set of running historical behaviors. At the same time, it reads the current junction temperature information fed back by the internal temperature sensor of the chip. Perform a fast scan on the target storage partition to obtain the current state distribution, and perform statistical processing to obtain the state distribution feature shape; The historical behavior set, current junction temperature information, current state distribution, and state distribution feature morphology are integrated and encapsulated to generate a real-time state package.

3. The adaptive energy consumption optimization memory chip control method as described in claim 2, characterized in that: The specific steps for retrieving historical parameters and matching resources at the resource granularity of real-time status packets, and combining these with environment matching records to output a candidate parameter set, are as follows: Collect historical control parameter records and perform similarity retrieval based on the state distribution characteristics in the real-time state packet to obtain a set of historical control parameter records; Based on the historical operation behavior set, the historical operation count information in the historical control parameter record set is mapped to the count interval and the interval is matched. At the same time, based on the current junction temperature information, the temperature interval markers in the historical control parameter record set are matched, and the environmental matching record is output. Perform resource-granular matching on the environment matching records, filter the historical control parameter record set, and extract the corresponding control parameter combinations as candidate parameter sets.

4. The adaptive energy consumption optimization memory chip control method as described in claim 1, characterized in that: The steps for comprehensively evaluating the operational and maintenance costs of the extended feasible set, performing multi-objective optimization ranking, and outputting the target operational parameter combination are as follows. Read the operation process energy consumption record corresponding to each parameter combination in the extended feasible set, combine the operation process energy consumption record with the current junction temperature information and perform conversion to obtain the operation cost assessment value; Based on the state changes over time and the decision boundary parameters, the frequency of maintenance operations within the target life cycle is determined, and combined with the single energy consumption record, the maintenance cost assessment value is output. The operational cost assessment value and the maintenance cost assessment value are jointly evaluated to form a comprehensive cost value. The parameter combinations in the extended feasible set are sorted according to the comprehensive cost value and multi-objective optimization is performed. The optimal parameter combination is selected as the target operational parameter combination.

5. The adaptive energy consumption optimization memory chip control method as described in claim 1, characterized in that: The specific steps for forming the execution result package are as follows: The target storage operation process is driven by the combination of target operating parameters, and the actual energy consumption trajectory is obtained by current sampling and time counting. Perform a fast scan on the target storage partition to obtain the post-operation state distribution. Integrate and encapsulate the actual energy consumption trajectory, post-operation state distribution, target operating parameter combination, and current junction temperature information to generate an execution result package.

6. The adaptive energy consumption optimization memory chip control method as described in claim 5, characterized in that: The specific steps for performing prediction deviation analysis on the execution result package, updating the prediction mapping relationship and historical parameter library, and outputting adaptive control basis are as follows. Extract the actual energy consumption trajectory and the post-operation state distribution from the execution result package, and align them with the operation cost assessment value and the state change over time, respectively, to calculate the energy consumption deviation between the actual energy consumption trajectory and the operation cost assessment value. The central tendency and discrete measure in the state distribution after the operation are compared with the central tendency and discrete measure in the state change over time, and the state distribution deviation is calculated. Based on the energy consumption deviation and the state distribution deviation, the correlation weight of the state distribution characteristic shape in parameter feasibility determination and reliability behavior prediction is adjusted to obtain an adaptive decision basis.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the adaptive energy consumption optimization memory chip control method according to any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the adaptive energy consumption optimization memory chip control method according to any one of claims 1 to 6.

Images