CN-121744724-B - Control method, equipment and medium for self-adaptive energy consumption optimized memory chip

CN121744724BCN 121744724 BCN121744724 BCN 121744724BCN-121744724-B

Abstract

The invention discloses a control method, equipment and medium of a self-adaptive energy consumption optimizing storage chip, which relate to the technical field of computer structures and comprise the steps of carrying out feasibility judgment on operation parameters in candidate parameter sets, carrying out reliability behavior prediction through a state evolution mapping relation to form an extended feasible set, carrying out comprehensive evaluation on operation cost and maintenance cost of the extended feasible set, executing multi-objective optimizing sequencing, outputting a target operation parameter combination, carrying out prediction deviation analysis on an execution result packet, updating a prediction mapping relation and a historical parameter library, and outputting a self-adaptive control basis.

Inventors

XIA JUNJIE
LI QINGJING

Assignees

深圳科摩思智能科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260228

Claims (8)

1. A self-adaptive energy consumption optimizing memory chip control method is characterized by comprising the following steps, Collecting the operation history behavior set and the current junction temperature information, reading the current state distribution, and extracting the state distribution characteristic form to form a real-time state package; Carrying out historical parameter retrieval and resource granularity matching on the real-time state packet, and outputting a candidate parameter set by combining an environment matching record; Carrying out feasibility judgment on the operation parameters in the candidate parameter set, and carrying out reliability behavior prediction through the state evolution mapping relation to form an extended feasible set; comprehensively evaluating the operation cost and the maintenance cost of the expanded feasible set, executing multi-objective optimization sequencing, and outputting objective operation parameter combinations; Based on the target operation parameter combination, executing target operation and synchronously collecting actual energy consumption track and state distribution after operation to form an execution result packet; carrying out prediction deviation analysis on the execution result packet, updating a prediction mapping relation and a historical parameter library, and outputting a self-adaptive control basis; Wherein, the feasibility judgment is carried out on the operation parameters in the candidate parameter set, the reliability behavior prediction is carried out through the state evolution mapping relation, the extended feasibility set is formed, the specific steps are as follows, Based on the current state distribution and the state distribution characteristic form, calculating the concentrated trend and the dispersion change range of the state distribution after the operation of the operation parameters in the candidate parameter set, comparing the concentrated trend and the dispersion change range with the decision boundary parameters, and outputting a feasible parameter subset; For each group of parameter combinations in the feasible parameter subset, calling a state evolution mapping relation, determining a state evolution track under the condition of the current junction temperature information and the historical running state data of the target storage partition, and deducing the change condition of the state along with time; Judging whether the state crosses the decision boundary parameter in the target life cycle according to the state change condition with time and the decision boundary parameter, eliminating the parameter combination crossing the decision boundary parameter, and outputting an expanded feasible set; The state evolution mapping relation is established by collecting long-term operation data of state distribution characteristic forms, continuously tracking a state change track along with time and corresponding judgment failure time data under different temperature conditions and historical operation state conditions and carrying out statistical analysis.
2. The method for controlling the self-adaptive energy consumption optimized memory chip as claimed in claim 1, wherein the steps of collecting the operation history behavior set and the current junction temperature information, reading the current state distribution, extracting the state distribution characteristic form, forming a real-time state package are as follows, After receiving a storage operation triggering instruction of a target storage partition, retrieving historical operation counting information, a historical control parameter record and corresponding historical state distribution, performing association integration to form an operation historical behavior set, and simultaneously reading current junction temperature information fed back by a temperature sensor in a chip; Performing quick scanning on the target storage partition to acquire current state distribution, and performing statistical processing to acquire state distribution characteristic forms; and integrating and packaging the operation history behavior set, the current junction temperature information, the current state distribution and the state distribution characteristic form to generate a real-time state package.
3. The method for controlling the adaptive energy consumption optimized memory chip as set forth in claim 2, wherein said performing historical parameter retrieval and resource granularity matching on the real-time status packet, and combining the environment matching record, outputting a candidate parameter set, comprises the following steps, Collecting historical control parameter records, and performing similarity retrieval based on state distribution characteristic forms in the real-time state package to obtain a historical control parameter record set; According to the operation history behavior set, mapping the history operation counting information in the history control parameter record set to a counting interval and performing interval matching, and meanwhile, matching the temperature interval mark in the history control parameter record set according to the current junction temperature information, and outputting an environment matching record; And performing resource granularity matching on the environment matching records, screening the history control parameter record set, and extracting corresponding control parameter combinations as candidate parameter sets.
4. The method for controlling the adaptive energy consumption optimized memory chip as set forth in claim 1, wherein the steps of comprehensively evaluating the operation cost and the maintenance cost of the extended feasible set, performing multi-objective optimized sorting, outputting objective operation parameter combinations are as follows, Reading operation process energy consumption records corresponding to each group of parameter combinations in the extended feasible set, combining the operation process energy consumption records with current junction temperature information, and converting to obtain an operation cost evaluation value; Determining the occurrence frequency of maintenance operation in a target life cycle based on the condition of state change along with time and the judgment boundary parameter, and outputting a maintenance cost evaluation value by combining single energy consumption record; and performing joint evaluation on the operation cost evaluation value and the maintenance cost evaluation value to form a comprehensive cost value, performing multi-objective optimization sequencing on the parameter combinations in the expanded feasible set according to the comprehensive cost value, and selecting the optimal parameter combination as a target operation parameter combination.
5. The method for controlling an adaptive energy consumption optimized memory chip as set forth in claim 1, wherein the forming the execution result packet comprises the steps of, Driving a target storage operation flow according to the target operation parameter combination, and acquiring an actual energy consumption track through current sampling and time counting; And performing quick scanning on the target storage partition, acquiring state distribution after operation, integrating and packaging an actual energy consumption track, the state distribution after operation, a target operation parameter combination and current junction temperature information, and generating an execution result packet.
6. The method for controlling an adaptive energy consumption optimized memory chip as set forth in claim 5, wherein the performing of the predictive deviation analysis on the execution result packet and updating the predictive mapping relation and the historical parameter library outputs an adaptive control basis according to the following steps, Extracting actual energy consumption track and state distribution after operation from an execution result packet, respectively aligning with the running cost evaluation value and the state change condition along with time, and calculating the energy consumption deviation amount of the actual energy consumption track and the running cost evaluation value; comparing the concentrated trend quantity and the discrete measurement in the state distribution after operation with the concentrated trend quantity and the discrete measurement in the state change condition along with time, and calculating the state distribution deviation quantity; and adjusting the association weight of the state distribution characteristic form in parameter feasibility judgment and reliability behavior prediction according to the energy consumption deviation amount and the state distribution deviation amount to obtain a self-adaptive decision basis.
7. A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that the processor realizes the steps of the self-adaptive energy consumption optimizing memory chip control method according to any one of claims 1-6 when executing the computer program.
8. A computer readable storage medium having a computer program stored thereon, wherein the computer program when executed by a processor implements the steps of the adaptive energy consumption optimized memory chip control method of any one of claims 1 to 6.

Description

Control method, equipment and medium for self-adaptive energy consumption optimized memory chip Technical Field The invention relates to the technical field of computer structures, in particular to a control method, equipment and medium of a self-adaptive energy consumption optimized memory chip. Background With the large-scale application of high-density nonvolatile memory chips (such as 3DNAND, PCM and novel resistive random access memory), a memory subsystem has evolved from an early pure read-write functional unit to an intelligent component with complex control logic and a fine management mechanism, and functional modules such as temperature monitoring, voltage regulation, write strategy selection and wear balance are generally integrated in a memory chip controller in the prior art, and operation parameters are dynamically adjusted through preset strategies or experience thresholds so as to achieve a trade-off among performance, reliability and power consumption. For example, a voltage compensation mechanism based on a temperature threshold, a lifetime management mechanism based on the number of erasures, and a power consumption adjustment strategy based on the intensity of a load have all become mainstream control methods. In the prior art, in the process of establishing a control strategy, a decision mode based on a single physical quantity threshold or a small quantity of operation indexes is mostly adopted to systematically describe the state distribution form presented by a storage partition under different historical behavior backgrounds, and complex statistical characteristics formed by a storage unit group under the influence of charge retention, aging drift and thermal stress are difficult to reflect. Meanwhile, the existing scheme is often matched according to the current instantaneous state or preset rules during parameter selection, and the joint analysis on the historical running track, junction temperature change environment and state distribution evolution trend is lacked, so that the control decision is difficult to realize fine energy consumption adjustment and reliability prediction linkage when facing long-term aging and short-term environment fluctuation superposition scenes. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the invention provides a self-adaptive energy consumption optimizing memory chip control method which solves the problem that fine energy consumption adjustment and reliability prediction linkage are difficult to realize under the superposition scene of long-term aging and short-term environmental fluctuation. In order to solve the technical problems, the invention provides the following technical scheme: in a first aspect, the present invention provides a control method for a self-adaptive energy consumption optimized memory chip, which includes collecting a running history behavior set and current junction temperature information, reading current state distribution, and extracting state distribution feature morphology to form a real-time state package; Carrying out historical parameter retrieval and resource granularity matching on the real-time state packet, and outputting a candidate parameter set by combining an environment matching record; Carrying out feasibility judgment on the operation parameters in the candidate parameter set, and carrying out reliability behavior prediction through the state evolution mapping relation to form an extended feasible set; comprehensively evaluating the operation cost and the maintenance cost of the expanded feasible set, executing multi-objective optimization sequencing, and outputting objective operation parameter combinations; Based on the target operation parameter combination, executing target operation and synchronously collecting actual energy consumption track and state distribution after operation to form an execution result packet; And carrying out prediction deviation analysis on the execution result packet, updating a prediction mapping relation and a historical parameter library, and outputting an adaptive control basis. The invention is used as a preferable scheme of the self-adaptive energy consumption optimizing memory chip control method, wherein the method comprises the steps of collecting the operation history behavior set and the current junction temperature information, reading the current state distribution, extracting the state distribution characteristic form to form a real-time state package, After receiving a storage operation triggering instruction of a target storage partition, retrieving historical operation counting information, a historical control parameter record and corresponding historical state distribution, performing association integration to form an operation historical behavior set, and simultaneously reading current junction temperature information fed back by a temperature sensor in a chip;