Patents
Literature
Patsnap Eureka AI that helps you search prior art, draft patents, and assess FTO risks, powered by patent and scientific literature data.

106 results about "Hit rate" patented technology

Hit rate is a metric or measure of business performance traditionally associated with sales. It is defined as the number of sales of a product divided by the number of customers who go online, Planned call, or visit a company to find out about the product.

Intelligent customer service session aggregation performance adjusting method, platform and program product

The invention relates to the field of AI platforms, and provides an intelligent customer service session aggregation performance adjustment method, platform and program product, and the method comprises the steps: obtaining a target merchant on an e-commerce session management platform, aggregating the sub-accounts of the target merchant on different e-commerce platforms for the target merchant, and configuring an intelligent customer service matched with the platform features. Through cross-platform aggregation, user inquiry of a target merchant in different e-commerce channels is received, the session efficiency is improved, and cross-platform message management is facilitated; a knowledge base blind area is analyzed based on a knowledge base hit rate thermodynamic diagram, platform differences are identified, a service delay trend of an intelligent customer service of a certain platform is captured in real time based on a cross-platform response delay distribution diagram to dynamically balance cross-platform loads, and a corresponding linkage strategy is executed through session transfer rate trend analysis. And the performance optimization of the intelligent customer service reception service is automatically triggered.
Owner:SHANGHAI AIYONGBAO TECH CO LTD

Display device and advertisement data playing method

The embodiment of the invention provides a display device and an advertisement data playing method. According to the method, storage positions are dynamically allocated according to advertisement data characteristics, memory occupation is reduced, and the access speed of key data is increased; a priority score is generated by calculating the access frequency, the resource type weight and the final access time of the advertisement data, and a low-score data block is preferentially removed when the storage space is insufficient, so that the cache hit rate of high-frequency / hot advertisements is ensured; by downloading recent advertisement data to the local in advance, real-time downloading delay during playing is avoided, large file advertisements are stored in blocks, cached parts are read preferentially during playing, meanwhile, remaining data blocks are downloaded in the background, and rapid playing of the advertisement data is achieved. In addition, cloud advertisement data change is monitored, local cache is dynamically synchronized, data timeliness is ensured, a local index table is quickly queried before the advertisement is played, direct playing or playing while playing is selected according to cache integrity, waiting time is shortened, and the problem that the advertisement playing speed is low is solved.
Owner:VIDAA (NETHERLANDS) INT HLDG LTD

Multi-core scheduling system and method based on interrupt affinity and storage medium

The invention discloses a multi-core scheduling system and method based on interrupt affinity and a storage medium. The technical problem that data locality guarantee and load balancing are difficult to consider at the same time is solved by constructing a self-adaptive optimization closed loop of perception-decision-execution-feedback, and the method comprises the steps that a routing interruption configuration mechanism provides a basis for dynamic adjustment; an interrupt guide task placement mechanism ensures that an interrupt service program and an associated task are executed in the same CPU core, and the cache hit rate is increased; the execution overhead of an interrupt service program is brought into core load statistics through task context switching and a'task + ISR 'two-dimensional precise load evaluation mechanism triggered by ISR inlet / outlet double nodes, and real load awareness is achieved; a load-driven dynamic interrupt routing mechanism is combined with dual control of a hysteresis threshold and cooling time, so that a ping-pong effect is avoided. According to the method, dynamic load balancing is realized on the premise of keeping data locality, and the real-time response performance and throughput of the multi-core system are remarkably improved.
Owner:北京星云越动科技有限公司

Down-sized cluster performance modeling for a tiered data processing service

Methods for modeling performance of tiered storage of a data processing service given a decrease in the storage capacity of a warm storage tier of the tiered storage are disclosed. Metadata of the warm storage tier is used to track hits due to incoming queries on data blocks that are stored in the warm storage tier. The metadata prioritizes data block identifiers that correspond to the data blocks stored in the warm storage tier by frequency of hits due to the incoming queries, or various other prioritization schemes. One or more partitions of the metadata may be set that correspond to respective downsized storage capacity scenarios of the warm storage tier. When an incoming query targets a data block within a given partition of the metadata, a hit counter is incremented to track the hit rate that would be made on the downsized warm storage tier corresponding to that partition.
Owner:AMAZON TECH INC

Operating method of storage controller, storage device, and operating method of storage device

An operating method of a storage controller includes, storing data and a source type of the data in a cache memory, when a cache hit occurs, determining a source type of data corresponding to the cache hit and updating a hit count corresponding to the determined source type among a plurality of hit counts, determining a source type with a dominant hit rate among a plurality of source types based on the plurality of hit counts, operating in a first mode when a result of the determining indicates that the source type with the dominant hit rate is a first type, and operating in a second mode when the result of the determining indicates that the source type with the dominant hit rate is a second type.
Owner:SAMSUNG ELECTRONICS CO LTD

Data caching method and device based on data access request, terminal equipment and storage medium

The invention discloses a data caching method and device based on a data access request, terminal equipment and a storage medium, and relates to the technical field of data caching, and the method comprises the steps: determining the service priority of each piece of storage data in a service period through the service period, and according to a service scene, determining the service priority of each piece of storage data; determining a first weight of the access record, a second weight of the importance level and a third weight of the service priority, and performing weighted summation according to the access record, the first weight, the importance level, the second weight, the service priority and the third weight to calculate the data heat of each piece of storage data; and storing the storage data of which the data popularity is greater than a preset popularity threshold into a preset data cache region. Therefore, the data popularity of the storage data in each business period and business scene is evaluated from multiple dimensions, and then the storage data with the data popularity greater than the preset popularity threshold value is stored in the data cache region, so that the problem of low hit rate of cache data in the prior art is effectively solved.
Owner:ELECTRIC POWER RES INST OF GUANGDONG POWER GRID CO LTD

Server multi-core computing processor load optimization test method, device and equipment and medium

The invention relates to the technical field of server testing, in particular to a server multi-core computing processor load optimization testing method, device and equipment and a medium, and the method comprises the following steps: automatically identifying hardware configuration information of a server; generating a multi-dimensional pressure test plan based on the hardware configuration information; executing the test plan, and monitoring the load rate, the memory bandwidth, the cache hit rate and the I / O performance index of each CPU core in real time; dynamically adjusting a task allocation strategy based on a set load difference threshold to realize task dynamic optimization, and if the difference between the maximum core load rate and the minimum core load rate is monitored to exceed the threshold, triggering a task migration operation; system performance data are collected again, performance indexes before and after optimization are compared and analyzed, and system performance bottlenecks are recognized; and automatically generating a performance test report. The utilization rate and the overall task throughput of the multi-core processor are improved, and the problem that a static task allocation strategy is difficult to adapt to dynamic load changes is solved.
Owner:SHANDONG CHAOYUE DATA CONTROL ELECTRONICS CO LTD

Computing system and semiconductor integrated circuit module

Provided is a semiconductor integrated circuit module including DRAM caches in a stacked structure, capable of preventing an increase in tag memory capacity on a processor side, reducing hit latency, and improving hit rate. A computing system (1000) is provided with a CPU (100), a main memory (300), and a cache DRAM circuit (400) stacked on the CPU (100). The cache DRAM circuit (400) operates as a group-connected cache memory. The address space of the main memory (300) is divided into a plurality of first sections, the address space of the cache DRAM circuit (400) is divided into a plurality of second sections in one-to-one correspondence with the first sections, and each second section comprises a plurality of paths. The cache controller (111) uniformly accesses rows in a plurality of paths corresponding to the address of the accessed main memory (300).
Owner:ULSTREETCAREMORY INC

A token aggregation and distribution method and system based on multi-level cache and semantic preheating

The present application relates to the technical field of semantic processing, and more particularly to a word element aggregation and distribution method and system based on multi-level cache and semantic preheating, which obtains historical word element request interaction logs, analyzes the access characteristics of different business scenarios and context fragments; in combination with semantic dependency relationships, identifies regular repetition and sudden hotspot characteristics of word element access, determines a preloading target set; real-time acquisition of context semantic data of the current inference request, analysis of word element prediction demand parameters to be generated for output; fusion of word element prediction demand parameters and preloading target set, dynamic generation of word element aggregation parameters and distribution scheduling strategy. The present application realizes forward-looking preheating and accurate scheduling of word element resources by deeply mining historical rules and real-time semantic intent, effectively solves the problems of weak semantic perception, low hit rate and response lag of traditional cache strategies, significantly reduces the inference delay of large models, and improves the throughput and stability of the system in high-concurrency scenarios.
Owner:SHANGHAI CHENGJI ELECTRONIC TECHNOLOGY CO LTD

360-degree video edge cache joint bit rate allocation method based on double time scales

The invention discloses a 360-degree video edge caching joint bit rate allocation method based on double time scales, and belongs to the technical field of 360-degree video streaming and edge caching. The method comprises the following steps that: in a long-time scale period, an edge server collects and analyzes historical request data of a user, calculates a comprehensive value priority for candidate video slices according to long, medium and short-term popularity of the slices, and further generates an optimal caching strategy which takes effect in a next period. In a short time scale period, according to the caching strategy, the edge server processes the real-time requests of the user and distributes bit rates to all concurrent requests fairly and efficiently. Through a self-adaptive optimization mechanism combining long-term planning and real-time response, the high cache hit rate of high-value slices can be ensured in different storage spaces, the data request quantity to the cloud is reduced, the pressure of a return link is relieved, and the comprehensive experience quality of a user is remarkably improved.
Owner:GUILIN UNIV OF ELECTRONIC TECH

Memory management method, electronic equipment and related device

The invention provides a memory management method, electronic equipment and a related device. Each file in the one or more files in the terminal equipment is divided according to the size of the memory page, and the hot page set is a part of pages in a plurality of pages of the one or more files. When the terminal equipment performs memory management according to the hot page set, in a plurality of pages of the same file, the swap-in priority of pages belonging to the hot page set is higher than that of pages not belonging to the hot page set, and / or the swap-in priority of the pages belonging to the hot page set is higher than that of the pages belonging to the hot page set. The swap-out priorities of the pages not belonging to the hot page set are higher than the swap-out priorities of the pages belonging to the hot page set. According to the method, the hot page and the non-hot page in the file are distinguished by taking the page managed by the memory as the granularity instead of taking the file as the granularity, the hot page in the file can be reserved in the memory for a long time, the hit rate is increased, the non-hot page in the file can be prevented from occupying the memory for a long time, and the memory management efficiency is improved.
Owner:HUAWEI TECH CO LTD

A method and corresponding device for managing dynamic library

This application discloses a method for managing dynamic libraries, applicable to a computer system. The method includes: requesting at least one huge page memory for an application program marked as in huge page mode; mapping the dynamic libraries that the application program relies on to the virtual address space corresponding to the at least one huge page memory; and loading the dynamic libraries into the at least one huge page memory when the application program is running. The solution provided in an embodiment of the application, using huge page memory to load dynamic libraries, can improve the hit rate of TLB queries.
Owner:HUAWEI TECH CO LTD

A segmented caching method that combines user preferences and objective popularity

This invention relates to the field of mobile communication technology, specifically to a segmented caching method combining user preferences and objective popularity. The method includes: obtaining user ratings and topic preference ratings for each file in a base station file library; determining the request probability of each file in the base station file library from the ratings and topic preference ratings, and forming a first file library from the top m1 files with the highest request probabilities; determining the popularity of each file based on the request probabilities of all users in the base station file library, ranking all files by popularity, and forming a second file library from the top n0 files, removing files that are duplicates of those in the first file library; and dividing the user's cache space into a first space that caches only files from the first file library and a second space that caches only files from the second file library. Compared with other popular caching strategies and optimization algorithms, this invention has a certain advantage in system average cache hit rate.
Owner:SOUTH CHINA NORMAL UNIV

General verification method and system for multiple lookup table entry hitting configuration table entry

The application discloses a general verification method and system for multiple lookup table item hit configuration table items, wherein the method comprises the following steps: generating a random character string as a configuration table item template, and obtaining multiple sets of offset collections; obtaining multiple configuration table items corresponding to the multiple sets of offset collections based on the configuration table item template and the multiple sets of offset collections; specifying multiple configuration table items associated with an original message from the obtained configuration table items; and configuring the original message based on the associated configuration table items to obtain an adapted message, which is used for subsequent verification. The application can obtain multiple configuration table items by using the same template, associate the configuration table items with the message, support multiple different lookup table full keys generated from the same message, simultaneously match multiple flow tables, and accurately constrain the hit rate.
Owner:WUXI STARS MICRO SYSTEM TECHNOLOGIES CO LTD

Precise pushing method and system based on e-commerce platform

The invention is suitable for the technical field of internet data analysis, and provides an accurate pushing method and system based on an e-commerce platform, and the method comprises the steps: firstly, based on the e-commerce platform, according to a sampling time period, rapidly obtaining the latest browsed product of a target user and the information of a plurality of historical browsed products, and then according to the latest browsed product, rapidly pushing a plurality of historical browsed products to the target user. According to the method, the multiple pieces of associated product information corresponding to the latest browsed product are accurately generated, then the target product information is effectively generated based on the multiple pieces of associated product information and the multiple pieces of historical browsed product information, and finally the product pushing instruction is efficiently generated based on the target product information. According to the method, the interest preference and behavior characteristics of the target user can be accurately captured, personalized recommendation of new products is realized, the recommendation strategy is dynamically adjusted in real time, the relevance and hit rate of recommendation are improved, the user experience and satisfaction degree are enhanced, and finally the user conversion rate and sales increase are driven. And accurate and efficient new product pushing is realized.
Owner:FOSHAN POLYTECHNIC

Method for electronic mall server to search for commodity information, storage medium and server

The application discloses a kind of electronic mall server search commodity information method, storage medium and server.The method is: P. the URL of each commodity information stored in cache queue is split into multiple fields according to preset rule, to each field and its field position as a field coordinate is stored in URL database;A. if receiving commodity information acquisition request, then determine the field coordinate of the URL of target commodity information requested to be acquired;B. to each field coordinate of the URL, obtain the number X of the same field coordinate of the field coordinate from URL database and the field mean C corresponding to the field position of its field position;C. if the number X of the field coordinate of the URL is less than the mean, then do not read cache queue, but directly read server disk to find the target commodity information.The method can improve the hit rate of electronic mall server reading cache queue, reduce its read operation to find commodity information, avoid its running stability due to a large number of read operations in short time.
Owner:CHINA SOUTHERN POWER GRID INTERNET SERVICE CO LTD

System and method for generating unbiased evaluation of a portfolio manager

A method for generating an unbiased evaluation of a portfolio manager. The method includes receiving holdings data of the portfolio manager, detecting one or more decisions from the holdings data corresponding to changes in quantity of an instrument, and generating episode data from the holdings data based on the decisions and a machine learning algorithm. A decision type is determined for the decisions based on a phase identifier and a time window. Value added metrics for the portfolio manager are calculated based on the episode data and the decision type using a machine learning algorithm. The method further includes calculating decision metrics for the decisions, including at least one of hit-rate, payoff, and BA score. The disclosed method enables objective evaluation of portfolio manager performance based on decision-making skills rather than solely on investment outcomes.
Owner:ESSENTIA ANALYTICS LTD

A method, system, medium and product for processing a translation lookaside buffer entry

The application discloses a kind of translation lookaside buffer table entry processing method, system, medium and product, applied to processor technical field, comprising: when process switching, determine the non-critical table entry corresponding to the process before switching in translation lookaside buffer, obtain target non-critical table entry;The target non-critical table entry is stored in the first cache corresponding to processor core, and the target non-critical table entry is deleted from the translation lookaside buffer;The recovery deadline corresponding to the process before switching issued by operating system is acquired;Based on the recovery deadline, before switching back to the process before switching, the target non-critical table entry is written back to the translation lookaside buffer from the first cache.This way, in the case of guaranteeing the process before switching fast recovery, the hit rate of translation lookaside buffer table entry can be improved, and then the processor performance is improved.
Owner:SHANDONG YUNHAI GUOCHUANG CLOUD COMPUTING EQUIP IND INNOVATION CENT CO LTD

Down-sized cluster performance modeling for a tiered data processing service

Methods for modeling performance of tiered storage of a data processing service given a decrease in the storage capacity of a warm storage tier of the tiered storage are disclosed. Metadata of the warm storage tier is used to track hits due to incoming queries on data blocks that are stored in the warm storage tier. The metadata prioritizes data block identifiers that correspond to the data blocks stored in the warm storage tier by frequency of hits due to the incoming queries, or various other prioritization schemes. One or more partitions of the metadata may be set that correspond to respective downsized storage capacity scenarios of the warm storage tier. When an incoming query targets a data block within a given partition of the metadata, a hit counter is incremented to track the hit rate that would be made on the downsized warm storage tier corresponding to that partition.
Owner:AMAZON TECH INC

Method and system for improving cache hit rate of edge CDN sink node

PCT designated stageWO2025218188A9TransmissionCache serverEngineering
The present application relates to the technical field of content delivery networks (CDNs), and provides a method and system for improving a cache hit rate of an edge CDN sink node. The method comprises: sink node cache servers each recording each URL request on the basis of a request sequence and saving same in a file; deploying an agent on each cache server to perform collection on the file, and collecting a compressed file every set time and uploading the compressed file to a central cache analysis service cluster; saving the number of access times of all URLs in a region, updating the access frequencies of the URLs in real time and sorting same, and screening for a most popular URL list; acquiring, from a database every set time, top N data records among data records that are sorted in a descending sequence on the basis of historical access times; using a CF to convert the top N data records into resource hotspot bitmap information, and issuing the resource hotspot bitmap information to each sink node cache server; and saving the resource hotspot bitmap information in a memory, and persisting the resource hotspot bitmap information in a magnetic disk.
Owner:CHINA TELECOM CLOUD TECH CO LTD

College purchase review expert recommendation method and device, computer equipment and storage medium

The invention discloses a college purchase review expert recommendation method and device, computer equipment and a storage medium. Obtaining feature vectors, review features and recommendation suggestions of the purchase review experts through the text information and the historical review items of each purchase review expert; according to the purchase project name and the to-be-purchased project, obtaining to-be-purchased project characteristics, recommended expert types and to-be-purchased project feature vectors; and according to the similarity between the purchase review expert feature vector and the to-be-purchased project feature vector, recalling a first number of purchase review experts matched with the to-be-purchased project and corresponding review features and recommendation suggestions from a purchase review expert database, and obtaining a target recommendation list. According to the invention, the recall and sorting strategies are adopted, and the real-time recommendation of the review experts of the to-be-purchased project of the college is realized through the cooperation of the project analyzer, the expert recall device, the expert analyzer and the expert sorter. According to the method, the hit rate and the average precision index are remarkably improved.
Owner:DALIAN UNIV OF TECH

Blockchain contract processing method and apparatus, electronic device, and storage medium

The present disclosure provides a blockchain contract processing method and device, electronic equipment and storage medium, comprising: loading N candidate smart contracts in batches from the to-be-processed smart contracts; in response to identifying that the candidate smart contract needs to perform a target external data read-write operation, judging whether the target external data exists in the local cache; in response to the target external data not existing in the local cache, performing saving processing on the first execution context data of the candidate smart contract, and determining a target contract from the to-be-processed smart contracts except the N candidate smart contracts; replacing the target contract with the candidate smart contract, and loading the target contract to the execution slot of the execution memory corresponding to the candidate smart contract, and executing the target contract. When it is found that the candidate smart contract needs to wait for the database to return, the context is saved and the execution is switched, avoiding the idling of the execution unit, improving the hardware utilization rate, and at the same time, by supporting batch preloading, the cache hit rate and execution continuity are improved.
Owner:BEIJING MICROCHIP SENSING TECH CO LTD

Recommendation system intervention method and system based on injection user automatic generation

The invention provides a recommendation system intervention method based on automatic generation of injection users, which comprises the following steps of: after determining hyper-parameters, circularly generating a total budget of the injection users: creating a batch of new users of a clear information set of interaction targets, and initializing recommendation model parameters; model parameters are updated according to the recommendation loss function, a confrontation function and a KL divergence loss function are calculated by using a single-step look-ahead strategy, and user embedding is updated and injected; calculating preference scores, shielding information interacting with the number of times of super-maximum interaction, selecting high-score information interacting with the number of previous maximum interaction budget times, and updating counting; and inputting intervention data to train a target recommendation system, and evaluating a target information hit rate. The invention further provides a recommendation system intervention system based on automatic generation of the injection user, a storage medium and computer equipment. Therefore, high efficiency and effectiveness of injection user generation can be realized, and the promotion intervention effect of clarification information is maximized.
Owner:INST OF COMPUTING TECH CHINESE ACAD OF SCI

Inference acceleration method for sentiment classification task, storage medium and device

The invention relates to a reasoning acceleration method for an emotion classification task, a storage medium and a device. The method mainly comprises a BERT classification model and further comprises a cache bypass, a Token filter is arranged in the cache bypass, and the method is based on approximate cache, that is, the closer the two inputs are embedded in a high-dimensional space, the higher the possibility that the two inputs belong to the same category is. According to the method, a trained Token filter (Token Filter) is adopted to predict the significance score of each Token in the input, and the low-score Token is filtered out according to the significance score of each Token. And a dot product method is adopted, so that the cache search speed in the GPU memory is improved, the hit rate and the accuracy are improved, and the interpretability of the similarity in an application scene is enhanced. By introducing the novel similarity caching mechanism, the problem that the reasoning time is too long in the sentiment classification task is solved, the reasoning speed can be effectively increased, and the reasoning time is shortened.
Owner:SUZHOU NEW HOPE INFORMATION TECH CO LTD

A database cache query method, device, equipment and storage medium

The application discloses a database cache query method and device, equipment and storage medium, belongs to computer technical field, the technical scheme is: the received query request is parsed, according to the analysis result, the ID identification of each data set to be queried is determined, and the corresponding query condition of each data set is determined according to the query request;The ID identification of data set is matched with the ID identification of data set saved in the cache queue one by one;If the matching degree of the ID identification of data set meets the preset condition, the record mapping information corresponding to the data set ID identification saved in the cache queue is read from the database, the data record mapped by the record mapping information is read, and the data record meeting the corresponding query condition of the current data set is matched from the data record, as the query result and return;Solve the technical problem of improving the hit rate of cache and reducing the query pressure of underlying database under the condition of limited cache space.
Owner:ZHEJIANG DAHUA TECH CO LTD

A method, system and device for efficient storage of pallets in a logistics store

This invention relates to the field of intelligent logistics warehousing and automated storage technology, specifically a method, system, and apparatus for efficient pallet storage in logistics storage. The method includes: collecting actual outbound instructions and pallet flow records as initial data packets; performing time-series correlation analysis to obtain the spatiotemporal coupling characteristics of orders; extracting pallet flow probabilities and calculating predictive cache hit rates; classifying heat levels based on historical outbound frequencies and assigning migration priorities; calculating the physical topology entropy reduction index of the storage area in conjunction with pallet flow probabilities; extracting idle time slices of handling equipment and calculating fragmented time rebalancing conversion rates; constructing a multi-objective optimization function and solving it to obtain a dynamic topology rearrangement strategy; generating migration trigger conditions and cache execution instructions, and executing them when the conditions are met; and calculating scheduling preemption rollback delays and formulating rearrangement interruption recovery strategies when sudden tasks have higher priority. This invention thereby improves the accessibility and scheduling stability of high-frequency pallets.
Owner:SHENZHEN ZHIJIANENG AUTOMATION CO LTD

A deep learning-based software crowdsourcing task recommendation method

The application discloses a software crowdsourcing task recommendation method based on deep learning. At present, many related researches propose to use the deep learning method to recommend the crowdsourcing task text information, but in the existing method, the extraction method of the crowdsourcing task text information lacks universality, and due to the unbalanced distribution characteristics of the crowdsourcing data, the hit rate and diversity cannot be considered in the index of the recommendation result. The method contains three parts: extracting crowdsourcing text features based on a pre-training model Bert, further learning the crowdsourcing text features based on CNN+LSTM, and the output based on the above two models, which can adaptively overcome the loss function of the unbalanced distribution of the crowdsourcing data. Through the application, the developer recommendation of a specific software crowdsourcing platform can be realized simply and efficiently, and the hit rate and diversity of the recommendation result are improved.
Owner:HANGZHOU DIANZI UNIV

Model reasoning request scheduling method and device, equipment and medium

The invention provides a model reasoning request scheduling method and device, equipment and a medium, and relates to the technical field of artificial intelligence, in particular to the technical field of large model and distributed model services. According to the implementation scheme, at least one to-be-coded first data item is determined based on a to-be-scheduled model reasoning request; determining a cache hit rate corresponding to each model instance based on a first data index and the at least one first data item, wherein the first data index comprises a plurality of second data items which are determined by performing deduplication processing on a plurality of historical data items cached by the plurality of model instances; the first data item and a data identifier corresponding to each second data item, the data identifier corresponding to each second data item comprising a plurality of sub-identifiers, and each sub-identifier indicating whether the model instance corresponding to the sub-identifier caches the second data item; determining a target model instance based on the cache hit rate corresponding to each model instance; and scheduling the model reasoning request to the target model instance to execute reasoning.
Owner:BEIJING BAIDU NETCOM SCI & TECH CO LTD

Data management method and device, medium and product

The invention discloses a data management method and device, a medium and a product, and relates to the technical field of computers, when data of a program operated on a certain processor core needs to be inserted into a shared cache, data belonging to different application programs are placed at positions with different priorities according to the types of the application programs. It is ensured that the shortest time for the data to stay in the private cache is that the data travels from the highest priority position matched with the processor core to which the data belongs to to the lowest priority position, and the intermediate process is not accessed again. Furthermore, the storage time of the data blocks with higher hit rate in the cache system is longer, so that the overall hit rate of the cache system is improved.
Owner:INSPUR SUZHOU INTELLIGENT TECH CO LTD