643 results about "Memory pool" patented technology

The invention provides a data management method, device and equipment and a readable storage medium, and the method comprises the steps: obtaining key value cache data generated by a first type of computing power unit in a pre-filling stage, transmitting the key value cache data to a memory pool, and caching the key value cache data in the memory pool; and in response to a data acquisition demand of the second type of computing power unit in a decoding stage, transmitting the specified key value cache data cached in the memory pool to the second type of computing power unit according to the data acquisition demand. According to the technical scheme of the invention, the memory pool is arranged and the aggregated transmission channel is utilized, so that the transmission efficiency of the key value cache data between the computing power devices is effectively improved, the data transmission delay is reduced, and the network transmission protocol overhead is reduced, thereby improving the data interaction efficiency of the computing power devices in a pre-filling stage and a decoding stage, and improving the user experience. The utilization of computing power resources is optimized, and the throughput and performance of large model reasoning are improved.

The invention discloses a memory fault repairing method and device, equipment, a medium and a computer program product, and relates to the technical field of computers.The memory fault repairing method includes the steps that memory error information sent by a memory controller is obtained, a fault target memory page can be positioned, and a standby memory page is obtained from a standby memory pool; the memory address mapping table is updated, the physical address of the target memory page with the fault is mapped to the physical address of the standby memory page, the repairing process does not depend on triggering of a system management interrupt mechanism, memory repairing in the system running stage is achieved, the system downtime caused by memory fault repairing is shortened, and therefore the memory repairing efficiency is improved. The problem of high delay in memory fault processing can be solved, and the technical effect of improving the memory fault repairing efficiency is achieved.

The invention provides a long document semantic coherence processing method and device, a storage medium and computer device.The method comprises the steps that block coding is conducted on a long document through a sliding window converter, and the sliding window converter comprises a window dividing unit, a position coding unit and a converter coding unit; the window division unit adjusts block positions according to sentences and paragraph boundaries and sets overlapping areas between adjacent blocks; the encoded document content is stored and managed through a memory management module, the memory management module comprises a chapter-level memory pool and a gating memory unit, the chapter-level memory pool stores the document encoding content, and the gating memory unit maintains entities and event timelines; performing deep semantic understanding on the document content through a semantic processing module, and calculating a multi-dimensional attention weight capture dependency relationship between paragraphs; providing a retrieval result through a retrieval module; and generating semantically coherent output content through a content generation module. According to the scheme, the semantic coherence of long document processing is improved.

The invention discloses a long video multi-modal understanding and question-answering method and system based on large model and retrieval enhancement generation. The method comprises the following steps: 1) a multi-modal feature extraction module; 2) a multi-modal synchronization and alignment mechanism; 3) constructing a structured memory pool; 4) querying a drive generation mechanism; 5) incremental updating and memory compression strategy; and 6) unifying the multi-modal representation space. The invention provides a long video multi-mode understanding method fusing a large language model and retrieval enhancement generation, and aims to break through the limitation of a traditional method in the aspects of single-mode processing and semantic fragmentation. According to the method, video image features are extracted through a visual model (such as YOLO and ViT), voice transcription and environment voice description are obtained in combination with an audio model (such as Whisper and Qwen-Audio), and unified coding of vision, voice and audio in a long video is achieved. Then, a structured memory pool is constructed through semantic consistency segmentation and timestamp alignment technologies to store time slice data of different modalities.

The invention discloses a compatible operation method for an Android camera HAL in a container based on memory access virtualization, and the method comprises the steps: taking a DMA-Buf memory heap of a Linux kernel host system as a target memory heap, taking an ION memory heap of an Android container system as a source memory heap, building a process exclusive memory management context, an FD cache table and a synchronous fence pool, and carrying out the execution of the process exclusive memory management context, the FD cache table and the synchronous fence pool; kernel registration and node binding of virtual ION equipment, pre-allocation of a first memory pool and access hook registration of kernel layer equipment are completed, and when an ION file descriptor is obtained in an HAL process, legality of a container process is verified, and context binding is initialized to the file descriptor; intercepting a memory allocation request of the HAL process, analyzing and adapting parameters, and preferentially multiplexing first memory pool resources to obtain an ION handle; when a data sharing request is processed, matching an FD cache or generating a new FD through an ION handle; when the HAL process releases the memory, resources are recycled according to a memory source, and when the HAL process exits, the context is cached or destroyed, so that cross-system memory operation compatibility is realized.

The invention discloses a multi-source data fusion type intelligent data management system, which comprises a data preprocessing module used for accessing a heterogeneous data source and generating a standardized data frame; the cognitive sub-graph construction module is used for generating cognitive sub-graphs and forming a cross-source evolutionary multi-cognitive hypergraph; the event aggregation module is used for aggregating entities and relationships according to event fingerprints and outputting a multi-source observation chain; the causal checking module is used for performing causal comparison and conflict detection and outputting a checking result, a positioning report and a treatment suggestion; the fusion updating module is used for distributing weights and performing weighted fusion, outputting fusion representation and confidence score, recording conflicts and updating the hypergraph; and the memory pool module is used for storing the hypergraph, the check result and the fusion representation, executing weight adjustment, forgetting and elimination, and outputting a service interface. According to the method, intelligent fusion and dynamic management of multi-source heterogeneous data are realized by constructing a cross-source self-evolution multi-cognitive hypergraph and an active evolution type fusion memory pool.

The invention provides a memory pooling method and system for model reasoning acceleration and a computer program product, and relates to the technical field of computers. The model is a separation pooling architecture, the separation pooling architecture comprises a pre-filling node pool, a decoding node pool and a CXL memory pool which are separated, and the memory pooling method for model reasoning acceleration comprises the steps that a reasoning request is allocated to a first pre-filling node in the pre-filling node pool based on a first scheduling strategy; the first pre-filling node processes the reasoning request to obtain a key value cache; storing the key value cache in a CXL memory pool; selecting a first decoding node in the decoding node pool based on a second scheduling strategy; the first decoding node obtains a key value cache from a CXL memory pool based on the reasoning request; and the first decoding node generates a reasoning result according to the obtained key value cache.

This application generally relates to memory provisioning and particularly describes systems and methods for distributed memory pools. For example, a computing / memory device in a distributed system may be configured to execute application logics using local cache for data maintained in non-local memory in another computing / memory device of the distributed system. The computing / memory device may further be configured to provide memory for other applications executed on other computing / memory devices of the distributed system. The local and non-local memory allocation for the applications in the computing / memory devices of may be dynamically and adaptively adjusted according to various system parameters and indicators.

The invention provides a city calculation basic model and equipment based on a Mama time sequence, and relates to the technical field of data processing. In the model provided by the embodiment of the invention, a Mama framework is combined with multi-task time sequence analysis. Aiming at the non-stationarity of urban time series data, a frequency prompt network is designed for decoupling in combination with a Koopman operator theory and dynamic mode decomposition, a steady mode and non-stationary fluctuation are analyzed from the angles of time and frequency, and a frequency mode memory pool (FPMP) is introduced for realizing cross-scene feature reuse. The advantage of a Mama model in capturing long-term time dependence is utilized, and an attention mechanism similar to Mama is designed to simulate channel dependence in a multivariable time sequence. A learnable supplementary sequence and a multi-task prompt mark are introduced to support dynamic multi-task adaptation, and the contradiction between urban data isomerism and task diversity is relieved. Shared knowledge and specific task knowledge are balanced through shared experts and specific task experts of the multi-task attention module, the negative migration problem in multi-task learning is relieved, and the generalization ability of the model is enhanced.

The invention discloses an impurity sorting and removing system based on machine vision, and the system comprises a data collection and preprocessing layer which is responsible for obtaining multi-dimensional physical characteristics of tea leaves and impurities, and constructing three-dimensional scene representation; the feature analysis and decision-making layer is used for extracting light field fusion features by utilizing a ResNeSt-50 backbone network, directly outputting a preliminary classification confidence coefficient through a full connection layer, forming a first layer of decision-making candidates, simulating branches by combining rigid body dynamics to predict a blade movement track, performing Bayesian correction on the preliminary classification confidence coefficient after a shielding probability graph is sampled and calculated through Monte Carlo Dropout, and forming a second layer of decision-making candidates; meanwhile, a Cook-Torrent BRDF model is integrated to carry out material reflection correction, a feature map after material correction is output, and a correction value is converted into decision threshold offset; and the knowledge storage and update layer is used for constructing a bimodal memory pool composed of a gradient direction matrix and an LRU elastic cache, and generating a pseudo sample supplement long-tail category in combination with the four-layer full-connection pulse neural network.

Memory has been playing a major role in the performance, scalability and applicability of General Compute systems, and more recently, in realizing Generative Artificial Intelligence (GenAI) and High-Performance Computing (HPC) systems that scale to thousands of GPUs, CPUs and special-purpose Accelerators. Embodiments herein disclose efficient software-defined protocol terminations and protocol translations utilizing Compute Express Link (CXL), including translations between CXL.mem and CXL.cache protocols. Also disclosed are CXL-based systems, Resource Provisioning Units (RPUs), and Memory Fabric Switches enabling dynamic memory pooling and sharing, host-to-host communication utilizing CXL.mem, CXL.cache and CXL.io, intent-based protocol translations, and optionally seamless interactions between CXL, UALink, NVLink, and / or Ethernet protocols, utilizing a broad range of semantics including IO, Cache, and Memory, optimizing memory access and reducing latency. Some embodiments also enable scalability, flexibility and security in high-performance architectures suited for data centers and next-generation computing environments.

Memory has been playing a major role in the performance, scalability and applicability of General Compute systems, and more recently, in realizing Generative Artificial Intelligence (GenAI) and High-Performance Computing (HPC) systems that scale to thousands of GPUs, CPUs and special-purpose Accelerators. Embodiments herein disclose efficient software-defined protocol terminations and protocol translations utilizing Compute Express Link (CXL), including translations between CXL.mem and CXL.cache protocols. Also disclosed are CXL-based systems, Resource Provisioning Units (RPUs), and Memory Fabric Switches enabling dynamic memory pooling and sharing, host-to-host communication utilizing CXL.mem, CXL.cache and CXL.io, intent-based protocol translations, and optionally seamless interactions between CXL, UALink, NVLink, and / or Ethernet protocols, utilizing a broad range of semantics including IO, Cache, and Memory, optimizing memory access and reducing latency. Some embodiments also enable scalability, flexibility and security in high-performance architectures suited for data centers and next-generation computing environments.

The invention discloses an intelligent parking dynamic scheduling optimization method based on multi-source data fusion, and the method comprises the following steps: S1, collecting multi-source data, and carrying out the preprocessing of the multi-source data; s2, constructing a heterogeneous graph structure, and generating a node feature matrix and a heterogeneous adjacency tensor; s3, extracting a state fusion vector, constructing a joint scoring matrix, and generating a candidate path set; s4, setting target parameters, executing path search and pheromone iteration, and outputting a regional path set; s5, adjusting a node preference weight, executing regional path reordering, and outputting an optimal vehicle parking path; and S6, recording the whole scheduling process, updating the network parameter weight through the sliding window memory pool, and generating an incremental learning sample set. According to the invention, high-precision, low-delay and personalized intelligent parking dynamic scheduling under the driving of multi-source data can be realized, and the parking efficiency and the user experience are effectively improved.

The invention discloses an incomplete multi-mode crisis event detection method based on a memory pool and a mode perception expert system, and relates to the field of machine learning, and the method specifically comprises the steps: building a prompt strongly related to an event instance based on the memory pool, and complementing a missing mode through a large model; a CNN encoder, a ResNet encoder and a CLIP encoder are utilized to construct a double-flow encoder, and text and visual feature extraction is achieved; capturing fine-grained event information and clues; the distribution difference between a complemented sample and a complete sample is identified based on a modal perception expert system, and the quality of fusion representation is improved; and inputting the mixed features into the classification head to complete a crisis event detection task. According to the invention, through the prompt construction module based on the memory pool, a large model is introduced to realize careful completion of a missing mode, and the distribution difference between a completion sample and a complete sample is accurately identified based on a mode perception expert system, so that a multi-mode data missing scene can be effectively coped with, and the accuracy and timeliness of crisis event detection are improved.

Examples include techniques to utilize near memory compute circuitry for memory-bound workloads. Examples include the near memory compute circuitry being resident on an input / output (I / O) arranged to couple with a plurality of memory devices configured as a memory pool that is accessible to a host central processing unit (CPU) through the I / O switch. The near memory compute circuitry may receive a request to obtain data from the memory pool and generate a result that is made available to the host CPU to facilitate acceleration of a memory-bound workload.

The invention relates to the technical field of printing task scheduling, in particular to a printing task scheduling method and system based on a main control chip and a terminal, and the method comprises the steps of initial task classification, dynamic priority adjustment, task queue recombination, resource allocation execution, termination mechanism triggering and the like. The high-priority task response speed is optimized through system load monitoring and a waiting time coefficient, and a real-time task pool and an elastic buffer pool are divided to improve the scheduling efficiency. According to the method, the real-time task pool and the elastic buffer pool are divided through an algorithm of'dynamic memory pool partitioning + task life cycle binding ', and the memory blocks are pre-allocated according to the task types, so that the real-time performance and the scheduling efficiency of the high-concurrency printing tasks are improved, rapid response of high-priority tasks can be realized, delay of key tasks is reduced, and the efficiency of the high-concurrency printing tasks is improved. The resource utilization rate and the scheduling fairness are improved, and the method is suitable for high-concurrency printing scenes.

The invention provides a self-adaptive information retrieval and optimization system based on retrieval enhancement generation, which comprises a prospective answer module used for receiving a query statement and context information and inputting the query statement and the context information into a large language model so as to generate a prospective preliminary answer; the confidence coefficient detection module is used for setting a confidence coefficient score threshold value to carry out keyword-level confidence coefficient evaluation on the content in the prospective preliminary answer, correcting a low-confidence-coefficient part in the answer according to an evaluation result, and outputting a complete answer content as a confidence coefficient answer; the knowledge graph extension query module is used for executing named entity recognition operation on the confidence answer and obtaining an extension query answer in combination with a pre-constructed small knowledge graph; and the context memory pool module is used for storing the context information input for the first time and related information fragments obtained in each round of retrieval process, preferentially performing information retrieval from the memory pool in the subsequent retrieval process, and accessing the external index database only when no matched content exists in the memory pool.

The invention belongs to the technical field of computers, and particularly discloses a data migration method based on a heterogeneous network memory pool and a data center, the method is applied to the data center, the data center comprises a management node and a plurality of cabinets, each cabinet comprises a plurality of nodes, CXL is adopted between the nodes in the same cabinet, RDMA is adopted between the nodes of different cabinets, and the management node is connected with the management node. The method comprises the following steps: monitoring access popularity of a node to a memory page; under the condition that the access popularity of the first node to the target memory page is greater than a popularity threshold value, migrating the target memory page from a second node to which the target memory page belongs to the first node through the management node, and initializing the access popularity of the first node to the target memory page; according to the method and the device, the access times and the access frequency of each cross-cabinet page are tracked, so that the high-heat target memory page with relatively high access times and access frequency can be migrated from the node to which the target memory page belongs to the CXL memory of the node of the cabinet to accelerate access.

The invention discloses a neural radiation field rendering method based on dynamic hash coding, and the method comprises the steps: employing the feature sequence data as the input, calculating the density value and color value of each sampling point through the forward propagation of a neural network, carrying out the volume rendering integral operation according to the ray tracing principle in the direction of a ray, and obtaining the feature sequence data; judging a final color output result of the current pixel point; according to an error value between the color output result and a real image, updating a network parameter weight through a back propagation algorithm, and if the error value is greater than a convergence threshold, continuing to iterate the training process to adjust a feature coding strategy to obtain an optimized neural radiation field model parameter; and after the rendering performance configuration parameters are obtained, optimizing a storage allocation strategy of feature data through a memory pool management mechanism, and if the current memory occupancy rate exceeds a safety threshold, starting a data compression algorithm to reduce the storage space requirement, and obtaining a real-time rendering output result. According to the invention, high-quality real-time rendering of the dynamic scene is realized.

The invention discloses a context-aware-based gap containerization TLS compatible method, which comprises the following steps of: distinguishing a simple TLS variable and a complex TLS variable in a source gap system, and establishing a TLS variable source feature; the method comprises the following steps: constructing a dynamic telescopic TLS memory pool matched with a variable access permission in a target system, expanding a kernel state TLS segment descriptor, constructing a bidirectional mapping table of musl TLS offset and a memory pool address, and optimizing and constructing a clustering structure aligned with a CPU cache line for a hotspot variable to improve the access efficiency; tLS variable access tracks are recorded through a dynamic binary translation engine, TLS segment descriptors are updated, the descriptors are read during thread switching to recover TLS context and refresh cache to guarantee consistency, address and offset conversion is completed according to target system alignment requirements by combining source features and a bidirectional mapping table for TLS variables of nested structures, pointers and array types, and the TLS variables of the nesting structures, the pointers and the array types are aligned. And completing dynamic conversion execution of the source system application program in the target system.

The invention provides a 3DGS model real-time processing method and device based on an unreal engine, and relates to the technical field of three-dimensional modeling, and the method comprises the steps: writing 3DGS data into a partitioned memory pool structure; writing data in the partitioned memory pool structure into a memory buffer area, performing coordinate system alignment processing according to a world coordinate system of an unreal engine, and performing spatial partitioning on the aligned processing data according to a point cloud density and a region density difference coefficient; converting the data in each spatial block into a point cloud data structure compatible with the Niagara particle system, and completing data format conversion in a graphics processor by using a calculation shader to obtain second processing data; and based on the constructed multi-level resource pool management strategy and a runtime memory monitoring mechanism, performing rendering processing on the second processing data according to a track prediction result of the camera, and generating a rendering result under an unreal engine. The processing efficiency of three-dimensional modeling in the unreal engine can be improved.

The embodiment of the invention provides a data migration method, an access processing method, a data migration system and electronic equipment. The data migration method is applied to a switch, the switch communicates with a plurality of memory devices in a memory pool based on an interconnection protocol, and each memory device is provided with a plurality of memory areas. When data migration needs to be carried out on a first memory area in a memory pool, a switch sequentially migrates a plurality of data pages obtained by dividing data in the first memory area into available second memory areas according to pages, and in the migration process, a page bitmap determined for the first memory region is utilized to record migration states of the plurality of data pages. By adopting the scheme, the problem of large-area fault of the server caused by the memory problem can be effectively solved, the page-level granularity is adopted to migrate the data in the first memory region page by page in the migration process, the access of the server to the first memory region is continuously processed in the migration process, and the influence on the performance of the server is reduced.

The invention discloses an auditing data security management and control system based on big data, and relates to the field of data management and control, and the system comprises a distributed crawler which collects auditing logs, operation records and network traffic in real time. And the stream batch integrated engine performs metadata alignment and timestamp synchronization on heterogeneous data to generate a unified audit event stream. The intelligent desensitization module carries out gradient desensitization on the sensitive fields and constructs reversible blood relationship mapping. According to the hierarchical storage, data are stored in an encrypted memory pool, an SSD and a cold standby cluster according to data sensitivity and access frequency, and consistent hash load balancing is achieved. Multi-modal safety monitoring analyzes data in real time through a three-level assembly line. And the hierarchical response module writes a permission chain according to the risk trigger action and the key audit event hash value, and constructs a tamper-proof evidence chain. The method has the advantages that the data security, privacy protection and efficient emergency response are ensured through intelligent desensitization, hierarchical storage and security analysis of the audit data, and the non-tampering property of the data is ensured by utilizing the block chain technology.

The embodiment of the invention provides a metadata asynchronous disk refreshing method based on a data stock calculation separation architecture, which comprises the following steps: receiving a modification instruction of target metadata in a shared storage node, locking the target metadata according to the modification instruction, and modifying the target metadata in a shared memory of the shared storage node; constructing a data log of the target metadata operation, and writing the data log into a disk; copying the modified target metadata into a memory pool, mounting the modified target metadata into a task queue of a background metadata disk brushing thread, and unlocking the modified target metadata; performing disk refreshing operation on the modified target metadata in the memory pool through a background metadata disk refreshing thread according to a task sequence mounted in the task queue; and after the modified target metadata is brushed, data logs in the disk are emptied. According to the embodiment of the invention, the holding time of the lock in the metadata modification process is shortened, and the lock conflict time is shortened.

The invention discloses a memory pooling system and a server system, and relates to the technical field of servers, the memory pooling system comprises a computing node, a memory exchange node and a memory pool node; the computing node is interconnected with the memory exchange node through a first bus, and is interconnected with the memory exchange node through a network; the memory pool node is interconnected with the memory exchange node through a first bus, and is interconnected with the memory exchange node through a second bus; the memory exchange node comprises a data center security control module; and the memory exchange node manages the memory pooling system through the data center security control module. According to the memory pooling system, the hardware cost can be reduced, and the maintainability of the whole server system is improved.

The invention discloses a process communication method, device and equipment and a computer readable storage medium, which are applied to the technical field of communication, and comprise the following steps: determining a target shared memory pool based on the size of a basic block and the number of basic blocks; determining a target process corresponding to the target shared memory pool, and setting a data callback function corresponding to the target process; wherein the data callback function comprises a function for processing a data change event in the target shared memory pool; and starting a data processing thread, waiting for a data change event triggered based on other processes, and triggering a data callback function based on the data change event to realize inter-process data communication. Compared with the prior art that the synchronization of the inter-process data cannot be realized in time, the data callback function corresponding to the target process is set, so that the change of the data can be quickly responded based on the data callback function, the memory state does not need to be continuously polled, and the inter-process communication efficiency is improved.

The invention discloses an operating system containerization kernel compatibility method based on shadow structure mapping, which comprises the following steps of: constructing a multi-dimensional mapping rule base by analyzing the difference of data structures between a container system and a standard system, and creating a shadow structure comprising a data area, a metadata area and an expansion area when the container system runs on a host system; the host system allocates a shadow memory space and a memory pool, establishes mapping between a shadow structure and a physical memory page and bidirectional mapping between the shadow structure and a host structure, intercepts system call of a container application, obtains a target structure and allocates a shadow structure and a host structure memory, and sends the shadow structure and the host structure memory to the host system; according to the method, conversion from the target structure to the host structure is completed according to the multi-dimensional mapping rule base, the host machine completes calling through the host structure, then the calling result is reversely converted into the target structure and returned to the container application, and seamless compatibility of the container system and the host system is achieved under the condition that a user mode program of the container system is not modified.

The invention discloses a resource management method and device, equipment and a medium, and relates to the technical field of storage, the resource management method comprises the following steps: carrying out performance test and data analysis on different storage architectures according to an I / O path of a distributed storage system and an asynchronous programming model, and determining processor static parameters under the different storage architectures according to the results of the performance test and the data analysis; the method comprises the following steps: collecting performance indexes of a distributed storage system through an asynchronous programming model, predicting load change conditions of the system according to the performance indexes of the distributed storage system, and dynamically adjusting processor resources according to the load change conditions of the system; managing a large-page memory pool and a cache according to the large-page memory management model and the business model type of the distributed storage system; and optimizing a memory application and release mechanism in the storage service of the distributed storage system. Through dynamic scheduling of processor resources and a large-page memory pool data sharing mechanism, balanced allocation of storage and calculation resources is realized, and the system performance and the resource utilization rate are improved.

The invention relates to the technical field of memory control, and particularly provides a memory allocation method and system, a terminal and a storage medium, and the method comprises the steps: configuring a memory pool based on a data processing flow and memory demand characteristics; obtaining a basic unit of each calculation unit for processing data each time, and configuring an address range and a current state for each memory block based on the determined size of the memory block in the memory pool; constructing a balanced binary search tree, and inserting all the memory blocks into nodes of the balanced binary search tree based on the sizes of the memory blocks; acquiring a memory allocation request, searching an optimal memory block in the balanced binary search tree based on the optimal adaptation strategy for memory allocation, and updating the current state of the optimal memory block to be occupied; and obtaining a memory release request, releasing the corresponding memory block back to the memory pool, and updating the current state of the released memory block to be idle. Accurate configuration, efficient utilization and flexible management of the memory are achieved, and scientificity and effectiveness of memory management are effectively improved.

The present disclosure discloses a single-target tracking method based on a credit allocation network. The credit allocation network is at first provided, which can be updated online using a guiding focus loss function. The credit allocation network generates credit scores for the tracking results by learning features of the target object, which ensures that the reliable samples are updated for storage in the memory pool. In order to better adapt to the changes in the appearance of the target during the tracking process, a new memory selection strategy is provided to collect high-quality tracking results as the memory samples during the tracking process, which further improves the reliability and adaptability of the memory pool. The present disclosure uses the guiding focus loss function to update the credit allocation network online, which can select more reliable memory samples for the memory pool and improve the robustness of the tracking results.