402 results about "Hyperscale" patented technology

The invention discloses an expert network automatic hybrid distributed algorithm based on cloud side end cross-domain data, and belongs to the field of large model distributed training. According to the method, an innovative method is introduced in the field of large model distributed training, an OSDP is used for training an MOE model in a distributed mode, distribution of different experts in a GPU cluster is optimized, and higher parallel computing efficiency is ensured. The combined motivation is used for improving the training efficiency under the condition that the model performance is not sacrificed. Through hybrid parallel accelerated training and hardware resource saving, the training cost of a large-scale model is reduced, and training of a super-large-scale deep learning model becomes more economical and feasible. And an automatic hybrid distributed algorithm based on an expert network is introduced, so that cloud, edge and end collaborative cross-domain heterogeneous data fusion is realized, and the complex requirements of the industrial intelligent field in a multi-base and multi-data-source environment can be met.

Provided in the present application is a data routing system. The system comprises a first topological network, a second topological network and a third topological network, wherein the network structure of the first topological network is a toroidal network topology structure consisting of a first node group, the first node group comprises a first number of network nodes, and the network nodes in the toroidal network topology structure have connection relationships with upstream nodes and downstream nodes in each dimension; the second topological network is a fat-tree network topology structure consisting of the network nodes in the first topological network and a switch; there is a cascade relationship between the network nodes in the first topological network and the switch; the first topological network is nested in the third topological network, and the network structure of the third topological network is a fully connected network topology structure consisting of a second node group; and the second node group comprises a second number of first node groups, and there is a connection relationship between the first node groups. Thus, large-scale or ultra-large-scale networking can be realized at a relatively low cost.

The invention discloses a VLSI circuit layout optimization method and system based on deep learning in the technical field of integrated circuits, and the method comprises the steps: abstracting a large-scale integrated circuit into a hypergraph structure, receiving the hypergraph structure, generating a node embedding matrix through a hypergraph neural network, and optimizing a segmentation loss function through unsupervised training, obtaining a probability distribution matrix, determining segmented subsets based on the probability distribution matrix, mapping each subset to a physical layout unit, generating a physical region layout of the chip according to an area equilibrium constraint and a time sequence constraint, performing constraint feedback on a physical layout result, and dynamically adjusting parameters of the segmentation loss function. The method can be used for circuit layout of a super-large-scale integrated circuit, and an optimization method and means are provided for providing a high-quality VLSI circuit layout scheme.

The invention relates to the technical field of computer software, and discloses a large-scale social network simulation method and system based on a dynamic topological structure. The method comprises the following steps: distributing a user portrait and a memory to a corresponding agent, and constructing a social network for the agent; distinguishing a core agent from a conventional agent according to the influence measurement; the core agent is driven by a large language model and interacts with other core agents through a natural language, so that the attitude of a conventional agent to an event is influenced; the conventional agent updates the attitude through the trust weight, and adjusts the social network by using a dynamic link prediction engine; updating the memory of the intelligent agent, pushing a library and a social network; and continuously performing a dynamic evolution process at each time step until the deduction of the event is completed. According to the method, detailed description of individual behaviors is ensured, the expandability of the whole system is also considered, and the decision-making efficiency and the simulation precision of the super-large-scale social network simulator are remarkably improved.

The invention relates to the technical field of data storage and management, in particular to a distributed storage and management method, system and equipment for super-large-scale medical and invasive data. The method comprises the following steps: acquiring super-large-scale medical wound data based on a distributed storage architecture, segmenting the data into a plurality of small blocks, and distributing the small blocks to a plurality of storage nodes; performing compression and duplicate removal processing on the data in the storage nodes to generate an efficiently stored data sequence; based on the data sequence, carrying out automatic hierarchical storage on the data by adopting an intelligent algorithm; based on a hierarchical storage result, a node comprehensive load value is calculated by using a multi-factor collaborative optimization load balancing algorithm, and data blocks in high-load nodes are screened for preferential migration; a distributed storage architecture and an intelligent data management system are combined, and data consistency management is achieved through a distributed consistency algorithm. The reliability and expansibility of the system are enhanced, and the system is suitable for scenes such as science and technology research and development and big data analysis.

The invention provides a giant satellite base measurement and control data transmission task planning method and device, electronic equipment and a medium, and relates to the technical field of multi-satellite task planning. A measurement and control data transmission integrated task scheduling strategy for a super-large-scale constellation aims at fully combining complex constraints of measurement and control and data transmission tasks in a giant satellite base; ground station resources are efficiently and reasonably allocated for the giant satellite base and a task execution scheme is optimized through the powerful learning and decision-making capability of a deep reinforcement learning algorithm, so that the execution effects of two task types are considered, the task satisfaction rate of the system is remarkably improved, and the utilization rate of time resources is improved; in conclusion, through multi-dimensional resource collaborative optimization and dynamic task planning, the bottlenecks of a traditional scheduling strategy in the aspects of task capacity, response timeliness and the like are broken through, and therefore the high-reliability in-orbit operation and mass data transmission capacity of a constellation system are ensured.

An end-side-cloud three-in-one active chat robot system for high-emotional merchants comprises an end side, a local edge server and a cloud end, and the end side is used for collecting multi-modal data of a user and performing local lightweight real-time processing and response execution; the local edge server is used for receiving and fusing the multi-modal features and the context information from the end side, and carrying out sentiment calculation, dialogue management and active trigger decision making with medium complexity; the cloud end is used for operating a super-large-scale model and providing global knowledge management, long-term user portrait storage and model training optimization; and the end side, the local edge server and the cloud end carry out cooperative communication through an encrypted channel to form a distributed intelligent processing architecture. According to the method, global optimization is realized by integrating end-side lightweight sensing, edge multi-modal fusion and cloud long-term memory. A composite finite state machine (FSM) active questioning mechanism is combined with sentiment calculation, and is different from traditional rule type triggering. Off-line and on-line fusion scheduling and multi-agent role playing are combined to be applied to a high-emotional-quotient interaction scene, and the simulation is improved. Multi-modal emotion perception circulation is introduced, and the problem of single text emotion misjudgment is solved.

Technologies for providing energy efficiency technology in extreme mMIMO systems in a cellular network cellular network (e.g., 5G wireless network, 6G wireless network) are described. The method collects data representing conditions for potential energy saving (ES) modes, comprising morphology data and traffic pattern data. The method determines, using the collected data, one or more energy saving (ES) modes for one or more components of a cellular network in at least one of a time (T) domain, a frequency (F) domain, or a space (S) domain, wherein the one or more ES modes cause at least one adjustment to the one or more components in the at least one of the time (T) domain, the frequency (F) domain, or the space (S) domain.

The invention provides a super-large-scale high-expandability grid generation method and device and a storage medium, and the method comprises the steps: obtaining a geometric model of a numerical simulation task, carrying out the surface grid division, obtaining a corresponding overall surface grid connected domain, carrying out the segmentation of the overall surface grid connected domain, obtaining a plurality of sub connected domains corresponding to the geometric model, and carrying out the segmentation of the sub connected domains; the method comprises the following steps: acquiring a plurality of sub-connected domains, distributing each sub-connected domain to each node, enabling each node to carry out volume mesh division on the sub-connected domains to obtain a mesh data file, searching other connecting surfaces corresponding to the connecting surfaces in the sub-connected domains, and finally updating mesh information in the mesh data file of each node according to a corresponding relationship among the connecting surfaces. And each node writes the updated grid information into the grid model file of the geometric model in parallel, so that the generation of the grid model for numerical simulation is realized, the memory occupation of each node during grid generation is reduced, the grid generation stability is ensured, and the file writing efficiency is greatly improved.

The invention discloses a robot hierarchical path planning method and device in a complex environment, a medium and a product, and the method comprises the steps: responding to a path planning request of a robot in an airport terminal, dividing the airport terminal into a plurality of dynamic Voronoi regions according to an underlying environment grid map and people flow density information, a minimum path cost algorithm with people flow density weighting is adopted to construct a global path anchor point sequence, the global path anchor point sequence is further split into a plurality of local path planning requests, a double-heuristic cost function is constructed, an improved algorithm is adopted to generate local sub-paths, and according to surrounding environment information collected by a robot in real time, the local sub-paths are calculated. And optimizing each local sub-path by adopting a reinforcement learning algorithm, generating motion control instructions for the robot at different control time points, and realizing dynamic obstacle avoidance of the robot. The path planning efficiency and the motion safety of the mobile robot in a super-large-scale dynamic environment are remarkably improved, and the technical problems that a traditional method is high in calculation complexity and poor in real-time performance are effectively solved.

A simulation method, apparatus and terminal for long-time-scale hierarchical operation of a power grid, and a medium. In the technical solution provided in the present application, on the basis of a hierarchical subordination relationship, simulation of a large-scale power grid is divided into three levels: a national level, a provincial level, and a prefectural level; by means of a hierarchical decoupling method, operation simulation of an ultra-large-scale and complex power grid is simplified into three stages: national-level power grid operation simulation, provincial-level power grid operation simulation, and prefectural-level power grid operation simulation; a lower-level simulation model only needs to perform operation simulation on the basis of key parameters output by an upper-level simulation model, and other than a limited number of key parameter interactions, the operation processes of simulation models at all levels are independent of each other; thus, the computational complexity of operation simulation of the power grid is simplified, and the association between upper-level and lower-level power grids is also taken into consideration, so that when facing massive scenario computation requirements, the operation time can be effectively reduced, and the operation efficiency can be improved.

The invention discloses a beam focusing optimization method based on dense connection expansion self-coding element learning, and relates to the technical field of super-large-scale IRS assistance, and the method comprises the following steps: S1, analyzing a super-large-scale intelligent reflector-assisted multi-user communication system, and obtaining a problem optimization formula; s2, constructing a beam focusing optimization algorithm network based on dense connection expansion self-coding element learning according to a problem optimization formula, and taking a matrix form of a phase shift matrix and a precoding matrix as the input of the beam focusing optimization algorithm network based on dense connection expansion self-coding element learning, and obtaining an optimized pre-coding matrix and an optimized phase shift matrix, and obtaining the maximum spectrum efficiency value of the super-large scale intelligent reflector assisted near field communication system according to the optimized pre-coding matrix and the optimized phase shift matrix. According to the invention, the spectrum efficiency of the system can be improved and the processing overhead of high-dimensional signals can be reduced.

The invention relates to a super-large scale array simulation method and device based on a wireless channel simulation instrument. The method comprises the following steps: quickly transmitting a phase sequence output by a phase shifter to the simulation instrument; the method comprises the following steps: discretely generating a track of a terminal, storing the track in a memory of a channel simulator, multiplying an additional phase of a phase shifter and a steering vector according to strict time synchronization, and outputting amplitude and phase gains brought by spatial correlation in an array unit; according to the single-array channel impact response at the moment t, generating a channel impact response of a single array in an array unit by utilizing a specified channel modeling method according to a terminal track; calculating a steering vector of an array in the array unit to each position of the UE; the data processing module performs dot product summation according to the phase shifter data transmitted by the high-speed interface and the array steering vector read from the memory; and the channel impact response and the input signal are multiplied to complete a fading channel simulation process, so that two-stage weight dynamic adjustment of the Massive MIMO system and a super-large-scale antenna array simulation test are realized.

Disclosed in the present application is a data routing system. The system is used for realizing data routing between computing nodes. The system comprises a first topological network and a second topological network, wherein the first topological network is nested within the second topological network. The network structure of the first topological network is an enhanced torus network topology structure formed by a first node group, wherein each network node in the enhanced torus network topology structure has a connection relationship separately with an upstream node, a downstream node and a non-adjacent node in each dimension. The network structure of the second topological network is a full mesh network topology structure formed by a second node group, wherein the second node group comprises a second number of first node groups, and there is a connection relationship between the second number of first node groups. In the system, by means of a hierarchical network structure in which an enhanced torus network topology is nested within a full mesh network topology, a large-scale or ultra-large-scale network is established, thereby effectively reducing the network cost and power consumption, and improving the network scalability.

The disclosure presents techniques for optimizing the allocation of virtual machines (VMs) in a hyperscale cloud computing environment to enhance overall efficiency. The system comprises a VM agent operating within each VM environment, and a node agent running on each computing node. Both agents report utilization metrics to a central VM balancing service. This service processes the received metrics through either predefined rules or a pretrained machine learning model to evaluate VM utilization. Based on the analysis, the service identifies underutilized VMs that are candidates for migration. When a VM qualifies for migration and a suitable receiving node is available, the system generates and outputs a migration plan, aiming to improve resource utilization across the network.

An efficient open set encryption traffic identification method for super-large scale traffic relates to the technical field of data security, solves the technical problems of insufficient resource efficiency, feature expression and open set identification in the prior art, and comprises the following steps: S1, inputting a network traffic data set, and generating dimension reduction data of an optimal dimension through a PCA dimension reduction algorithm; s2, inputting the dimension reduction data of the optimal dimension into an improved MobileNetV3 model, and in combination with an improved SE attention mechanism and a composite structure of Dense connection and residual connection, extracting features and outputting a classification result; s3, judging whether a classification result belongs to a known category, a drift sample or an unknown category by using an open set recognition algorithm based on confidence; s4, if the classification result is a drift sample or an unknown category, incremental learning based on knowledge distillation is started, parameters of the improved MobileNetV3 model are updated, and old knowledge is reserved. According to the invention, the network security protection capability can be effectively improved.

The invention discloses an automatic generation method and device for a large-scale unstructured grid of a commercial aircraft. The method comprises the following steps: carrying out geometric segmentation on a computational domain enclosed by a near-field bounding box and a far-field bounding box through a constructed inscribed geometric surface to obtain far-field regions; generating a first surface grid on the surfaces of the aircraft body and the aircraft parts in the near-field region, generating a second surface grid on the inscribed geometric surface of the near-field region and the far-field region, and generating a third surface grid on the surface of the far-field bounding box; generating a non-structural body grid based on the second surface grid and the third surface grid; automatically generating a basic unstructured mixture grid in the near-field region based on the first surface grid; according to the commercial aircraft large-scale non-structural hybrid grid generation method and device, the calculation domains are divided into the calculation sub-domains, automatic generation of the grids of the calculation sub-domains is achieved, then automatic combination of the grids is achieved, the technical bottlenecks that super-large-scale non-structural grids are difficult to generate, the automation degree is low and the like are effectively solved, and the commercial aircraft large-scale non-structural hybrid grid generation method and device are suitable for large-scale aircraft large-scale non-structural hybrid grids. And the simulation efficiency is improved.

The invention relates to the technical field of large language models, in particular to a high-concurrency large language model high-speed reasoning deployment method, which comprises the following steps of: extracting a weight coefficient corresponding to a user identity identifier, an emergency degree numerical value mapped by a request type and a resource occupation value converted by an estimated calculation amount according to the user identity, the request type and the estimated calculation amount; according to the method, the dynamic priority is generated by performing comprehensive operation on the user identity, the request type and the estimated calculation amount, so that differentiated services for reasoning tasks are realized, interactive requests with high urgency degree can bypass batch processing tasks with long time consumption, the response delay and delay jitter of key services are reduced, and the user experience is improved. Meanwhile, according to the complexity of the request text and the geographic coordinates of the user, the task is intelligently routed to the model node with the most suitable scale in the distributed network, so that the wide area network transmission delay is greatly reduced through edge processing, and the computing power waste caused by using a super-large-scale model to process the simple task is also avoided.

The invention relates to a power network structure weakness detection method based on multi-diagonal-block matrix decomposition, and belongs to the technical field of super-large-scale integrated circuits. According to the method, an original large-scale sparse matrix is converted into a band edge diagonal block structure and divided into a plurality of sub-matrixes capable of being solved independently by constructing a layering and blocking strategy for eliminating tree drive, and redundancy of full-matrix calculation is avoided. Meanwhile, a local approximate inverse algorithm of column norm truncation is designed, and target elements are calculated on the premise that a preset error threshold value is met. According to the method, the parallel computing architecture of the multi-core processor is fully utilized, the computing complexity is reduced by 2-3 orders of magnitude, the computing efficiency is effectively improved, the memory occupation is remarkably reduced, and a high-precision and high-efficiency solution is provided for detecting the weakness of the power network structure of the super-large-scale integrated circuit.

The invention provides an MCM-GPU-oriented resource storage optimization method, device and equipment, and the method comprises the steps: carrying out the static analysis of an operator-level data flow diagram of a calculation task in a compiling stage, and extracting the data dependence relation and life cycle characteristics of each operator; generating a storage optimization scheme based on the data dependency relationship and life cycle characteristics of each operator; in the operation stage, task and data collaborative allocation is carried out according to a data prefetching strategy and a data replacement strategy, task and data collaborative allocation and the calculation task are executed in parallel, and a data flow diagram and an operator dependency relationship of the task are analyzed by utilizing a compiling period; in combination with the data locality characteristics of the MCM-GPU multi-level storage architecture and the concurrency requirements of different modules for data access, an appropriate GPU module is selected to carry out data optimization management, so that frequent data exchange between a GPU and a CPU is reduced, and the performance potential of the MCM-GPU architecture in a super-large-scale task is brought into full play.

The invention discloses a universal 3D codebook design method and device for super-large-scale MIMO (Multiple Input Multiple Output). The 3D codebook design method comprises angle design and distance design. According to the design method, a design scheme of an angle and a distance is obtained based on a target of minimizing a correlation function. Although the angle and distance design scheme is initially designed for the near-field area, the design principle and the implementation effect of the angle and distance design scheme are also suitable for the far-field area, and the angle and distance design scheme has good universality. Compared with the existing work, the 3D codebook design theory and method can realize higher discrimination of the wave beam in each sampling angle and sampling distance, and reveals the performance gain brought by the excavation of the degree of freedom of the vertical dimension under the ideal condition.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for processing an input through each of a plurality of layers of a neural network using a plurality of hardware accelerators to generate an output. The plurality of layers includes a fully connected layer having a plurality of parameters arranged in a row dimension and a column dimension. One of the methods includes generating a plurality of parameter blocks by partitioning a plurality of parameters along a row dimension and a column dimension; determining a ratio of the number of parameters along the row dimension to the number of parameters along the column dimension; and determining whether to use row fragmentation or column fragmentation in conjunction with the plurality of hardware accelerators to calculate an output for the fully connected layer, and then using the row fragmentation or the column fragmentation to calculate an output for the fully connected layer.

The invention relates to an AI processing method for NAS equipment and an NAS system, the method is used for an NAS control unit in the NAS equipment, the NAS control unit realizes butt joint with an AI model in a mode of communication connection with an AI processing module, and the method comprises the following steps: the NAS control unit calls a user operation layer in a first form, generates a first AI demand, and sends the first AI demand to the AI model; the AI model receives a first AI demand, converts the first AI demand into an AI instruction, and feeds back the AI instruction to the user operation layer in the first form; the user operation layer in the first form is converted into a user operation layer in a second form based on an AI instruction; and the NAS control unit calls the user operation layer in the second form as an AI processing result. According to the scheme, a super-large-scale model or data set can be processed without being limited by local hardware configuration, and the functions of data security and privacy protection can be achieved in a multi-user access environment at low cost.

Systems and methods for making use of non-persistent storage as the journaling storage media for a virtual storage system are provided. According to one embodiment, in order to meet the needs of Extreme Low Latency Workloads while also seeking to provide predictable performance and the lowest possible latency, ephemeral storage of the virtual storage system is used to preserve state information (e.g., in the form of boot arguments and an operation log journal) across a host failure recovery scenario in which the virtual storage system is expected to be redeployed within a compute instance brought up by a cloud environment of a hyperscaler on the same host, thereby providing improved data durability (fewer host failure scenarios that result in lost data) as compared to the use of ephemeral memory of the compute instance and lower write latency than the use of persistent storage provided by the cloud environment.

The invention provides a data routing system. The system comprises a first topology network, a second topology network and a third topology network. The network structure of the first topology network is a surrounding network topology structure formed by a first node group, the first node group comprises a first number of network nodes, and the network nodes in the surrounding network topology structure are respectively connected with upstream nodes and downstream nodes in each dimension. And the second topology network is a fat tree network topology structure formed by network nodes and switches in the first topology network. And each network node and the switch in the first topology network have a cascade relationship. The first topology network is nested in the third topology network, and the network structure of the third topology network is a full-interconnection network topology structure formed by the second node groups. The second node groups comprise a second number of first node groups, and the first node groups are connected. Therefore, large-scale or super-large-scale networking can be realized with relatively low cost.

The invention discloses a high-order multi-scale finite element method and system for chip multi-physics field simulation. According to the method, a designed chip model is embedded in a regular coarse background grid, and the problem of numerical instability caused by unit cutting can be caused by embedded simulation. Therefore, ill-conditioned coarse nodes of a coarse unit are judged based on a boundary interpolation function; furthermore, a reduced coarse unit is constructed by removing an ill-conditioned boundary interpolation function, so that a good condition number is ensured. Therefore, a numerical value shape function capable of reflecting internal material distribution is constructed on the coarse background grid for subsequent simulation. On the premise of keeping the solving precision, the calculation cost is reduced, the numerical stability of embedded simulation is improved, and the method has huge potential in super-large-scale boundary complex structure simulation. The method is suitable for multiple simulation fields of statics, dynamics, thermal fields and the like, and the numerical stability, simulation precision and calculation efficiency advantages of the method are verified in statics and thermal field simulation.

The invention discloses a near-field channel state information feedback method under a super-large scale array based on Transform, which comprises the following steps: in a downlink XL-MIMO system in a near-field area, a user side performs two-dimensional matrix processing on acquired CSI data and transmits the processed CSI data to a coding end; two-channel coding compression is carried out on the CSI matrix, a first channel uses an encoder to extract channel related characteristics, and a second channel enables an original CSI matrix to pass through a linear network so as to reserve original information; compressing the output of the two channels, and splicing to generate a low-dimensional code word to be fed back; the base station side decoding end splits the low-dimensional code word and decompresses the low-dimensional code word of the corresponding channel; decompressed dual-channel code words are respectively input to two ends of a decoder; and carrying out size change on the output of the decoder to obtain the estimated original CSI. According to the method, angle and distance information of a near-field channel is fully utilized, and efficient CSI compression and high-precision reconstruction are realized.

The invention relates to the technical field of information analysis and processing, in particular to a universal quality analysis coding system based on LLM. The efficiency bottleneck of traditional quality analysis is broken through, automation, intellectualization and high efficiency of quality data analysis are achieved, the quality analysis efficiency, the semantic understanding depth and the consistency of analysis results can be remarkably improved through the system, particularly, outstanding advantages are shown when super-large-scale data sets are processed, and the method is suitable for popularization and application. Researchers in more disciplinary fields can quickly master and apply the quality analysis method, and the method has higher adaptability and flexibility in cross-field application. Compared with a traditional quality analysis method, the method has the advantages that analysis and processing of super-large-scale research samples can be achieved, an efficient, scientific and innovative research method and technical support are provided for quality research, high academic value is achieved, and the automatic and intelligent progress of quality research is promoted.

The invention relates to an information-missing-oriented super-large-scale integrated circuit security design detection method, and belongs to the field of hardware Trojan horse detection. The method comprises the following steps: analyzing a gate-level netlist design file of the super-large-scale integrated circuit with missing information, creating directed graph representation, performing feature representation coding on logic gate nodes, and constructing circuit directed graph data; performing structural reasoning on the directed graph data by using a graph variation auto-encoder GraphVAE, complementing a circuit structure with information loss by using a decoder, and constructing complemented circuit directed graph data; performing logic gate type, fan-in and fan-out quantity, betweenness centrality and neighborhood type distribution calculation on the complemented data, and combining and splicing the complemented data into circuit gate characteristics; and constructing a graph neural network model GNN, further extracting deep-level features through graph convolution operation, and performing reasoning training by adopting a multi-layer perceptron to realize hardware Trojan classification. According to the method, the hardware Trojan of the gate-level netlist under the condition of information loss can be effectively detected.

The invention relates to a multi-mode multi-agent RAG-based power dispatching method and system, and the method comprises the steps: generating a dispatching plan through the cooperative work of a plurality of heterogeneous agents with different division of labor, and receiving query input containing a power dispatching task; the planner agent decomposes the power dispatching task into a plurality of sub-tasks which are mutually dependent and generates a corresponding execution plan; according to the execution plan, the searcher agent sequentially executes two stages of fine-grained multi-modal retrieval on each sub-task from the multi-modal power knowledge base, coarse-grained whole-page retrieval is carried out in the first stage, fine-grained area fine reading and modal expansion are carried out in the second stage, and evidence related to the current sub-task is obtained and added into an evidence pool; and according to all evidences in the evidence pool, the responder agent answers the initially input query to generate a scheduling plan. According to the method, the problems of timeliness and multi-scene adaptability in super-large-scale power dispatching are effectively solved, and an innovative path is provided for guaranteeing dynamic safety and stability of a power grid.