1095results about "Program synchronisation" patented technology

A parallel processing unit comprises a plurality of processors each being coupled to a memory access hardware circuitry. Each memory access hardware circuitry is configured to receive, from the coupled processor, a memory access request specifying a coordinate of a multidimensional data structure, wherein the memory access hardware circuit is one of a plurality of memory access circuitry each coupled to a respective one of the processors; and, in response to the memory access request, translate the coordinate of the multidimensional data structure into plural memory addresses for the multidimensional data structure and using the plural memory addresses, asynchronously transfer at least a portion of the multidimensional data structure for processing by at least the coupled processor. The memory locations may be in the shared memory of the coupled processor and / or an external memory.

The invention discloses an android container rendering optimization method based on cross-domain hard real-time Fen synchronization, which comprises the following steps: by taking a swan-mong system as a host and an android system as a container, creating a hash table, a synchronous thread and a shared memory corresponding to a GPU core when the host is started, acquiring VSync cycle registration callback, transmitting shared memory FD to the container, and finishing shared memory mapping and alignment by the container. Registering a GPU queue to complete callback; when the Android application is started, a container obtains a queue and a physical address of a rendering buffer area, creates a Fen and binds the Fen to the queue, after the queue is submitted, metadata is written into a shared memory to inform a host, after the host receives the metadata, nodes are created and stored in a hash table, and an overtime timer is registered; after the GPU completes the command queue, the container calls back an update state and a verification value to notify the host, and after the host is verified to be valid, the corresponding hash table is updated, the timer is reset, and asynchronous screen loading is triggered; and the host executes buffer area synthesis and submission of the display equipment to finish on-screen, so that the stability of the rendering frame rate is improved, and the reliability of cross-domain synchronization is ensured.

A new transaction barrier synchronization primitive enables executing threads and asynchronous transactions to synchronize across parallel processors. The asynchronous transactions may include transactions resulting from, for example, hardware data movement units such as direct memory units, etc. A hardware synchronization circuit may provide for the synchronization primitive to be stored in a cache memory so that barrier operations may be accelerated by the circuit. A new wait mechanism reduces software overhead associated with waiting on a barrier.

The invention provides a distributed event processing method and electronic equipment, which can be applied to the field of distributed technologies. The distributed event processing method comprises the following steps: receiving an event processing request at a request receiving entrance of a distributed system, and searching a priority and a mutual exclusion execution type corresponding to an event type from a preset event registration library according to an identifier representing the event type in the event processing request; based on the priority and the timestamp of the event processing request, inserting an event object obtained after packaging the event processing request into a priority scheduling queue; after an event object used for execution is extracted from the priority scheduling queue, conflict detection is carried out on a mutual exclusion execution type corresponding to the event type based on a global mapping table maintained in the distributed system; and under the condition that the conflict is not detected, calling a service processor corresponding to the event type to execute the event object, and updating the current execution count of the event type in the global mapping table based on the execution state.

The invention provides a dynamic tool integration method and system for an intention-driven AI dialogue system, and the method comprises the steps: S5, extracting resource occupation information from a transaction log according to a state persistence record, carrying out the resource release and temporary context cleaning of each heterogeneous component, and obtaining a released resource list; s6, searching a breakpoint state snapshot and a transaction log from the distributed database according to the released resource list, and reconstructing a tool chain execution flow historical state to obtain a fault context analysis result; and S8, according to the context recovery state and the semantic feature vector, updating a tool chain execution stream, and reallocating computing resources through a heterogeneous component coordination mechanism to obtain a task continuity confirmation result. According to the method, the recovery efficiency and the resource utilization rate of the tool chain in an abnormal scene are remarkably improved, and the task continuity and the system robustness in a high-concurrency environment are guaranteed.

The invention relates to the technical field of multi-agent collaboration, in particular to a multi-agent collaboration method and system based on shared memory data exchange, and the method comprises the steps: creating and initializing a shared data area which is mapped and accessed by a plurality of agent processes in a physical memory; constructing an annular buffer structure on the shared data area, and realizing a lock-free read-write pointer propulsion mechanism based on atomic operation and a memory barrier so as to support a plurality of intelligent agents to perform data read-write concurrently; asynchronous notification of data updating is carried out between agent processes through a lightweight event notification mechanism; and each agent performs autonomous decision making and task scheduling based on the shared data so as to realize distributed collaboration. A data block version control and verification mechanism is introduced into the annular buffer structure, and when it is detected that the agents are abnormal, resources of the abnormal agents are automatically recycled, and a data consistency recovery process is triggered. Multiple times of data copying between a kernel mode and a user mode are eliminated, and communication delay is reduced.

The invention relates to a memory shader. The programmable atomic memory shader execution circuitry is a seamless component of the hierarchical memory system, and can receive and execute programmable atomic operation calls from any number of processors. Programmable atomic memory shader execution circuitry close to memory allows execution circuitry to access shader programs stored in memory, thereby eliminating latency that may occur when an upstream processor exchanges shader instructions, data, and memory lock / unlock commands with the execution circuitry. The locked / unlocked state of the programmable atomic memory shader execution circuitry (e.g., in the L2 or L3 cache memory) allows the system to quickly lock the memory resources, perform one or more operations in an atomic manner over one or more cycles, and then quickly unlock the memory resources.

Apparatuses, systems, and techniques to facilitate graph code synchronization between application programming interfaces. In at least one embodiment, one or more circuits are to perform an application programming interface (API) to cause graph code to wait on a semaphore used by another API.

The invention provides a visual form construction method, which comprises the following steps: receiving user form operation data, carrying out business logic calculation, state change control and inter-service coordination operation on the user form operation data based on domain-driven design, and generating domain events, state data and service governance data. Domain events and state data are recorded to an event storage system, a query optimization view is constructed, event logs and query view data are generated, and dual optimization of data persistence and query efficiency is ensured. According to a preset security policy rule, real-time security monitoring is performed on the service governance data, a system health state index and security audit data are generated, and a system-level observability and security protection mechanism is established. And performing interface rendering and synthesis based on the event log, the query view data, the system health state index and the security audit data to generate a user interactive form interface. Through fusion processing of multi-source data, deep integration of interface rendering and security policies is realized.

The invention discloses a distributed task scheduling system and method for multi-robot collaborative drawing, and the system employs a distributed architecture of a master control server and a plurality of robot clients, and achieves the decoupling of drawing task generation and execution through a task queue. The master control server receives the task request, asynchronously generates a robot control instruction from the image data through a built-in task calculation module, and maintains a robot state management module; and the robot client registers the unique identifier and keeps communication with the server. During scheduling, the server actively queries the state management module, selects an available robot to issue an instruction and atomically updates the state of the available robot to be unavailable; and after the task is completed, the client feeds back a receipt, and the server updates the state to be available. According to the method, the problems of high concurrency, repetition prevention and loss prevention of task scheduling in a multi-robot system are solved, and efficient and reliable parallel execution of non-standardized artistic creation tasks and elastic expansion of the system are realized.

The invention relates to the technical field of new energy power generation, and discloses a generating capacity calculation method and system based on wind and light resources and equipment data. The method comprises the following steps: acquiring and fusing wind and light resource data and equipment performance parameters to construct a standardized data set; based on the data set, a resource-equipment dynamic coupling potential field is established, and the potential field accurately represents the comprehensive power generation potential in the space-time dimension by fusing the resource space gradient and the equipment response characteristics; carrying out triple integral calculation on the potential field, and controlling an error by adopting a self-adaptive algorithm; and finally, realizing self-calibration of the system through model verification and parameter optimization. According to the method, the defects of insufficient data fusion, model staticization, contradiction between calculation precision and efficiency and the like of a traditional method are overcome, and the accuracy, the adaptive capacity and the engineering practical value of generating capacity calculation are remarkably improved.

The invention relates to the technical field of task scheduling, and discloses a scheduling control method and device for a multi-core real-time processing system and a computer storage medium, and the method comprises the steps: when a target process has a high-bandwidth operation request, executing a first bandwidth preemption operation based on a bandwidth lock mechanism, scheduling a target processor core corresponding to the target process to preempt a first shared resource required by the high-bandwidth operation request; the first shared resource comprises a CPU shared resource; based on a dynamic bandwidth preemption priority mechanism, executing a second bandwidth preemption operation according to the process priority and the task information of the target process so as to schedule a task execution unit of the target process to preempt a second shared resource; the second shared resources at least comprise global shared resources. Visibly, the scheduling reliability and accuracy between the processor cores and the main devices of the multi-core CPU in the multi-core real-time processing system can be improved, the problems of inner competition and outer competition are solved at the same time, and therefore the real-time performance of the system is improved.

The invention discloses a failover method and device, electronic equipment and a storage medium, and relates to the technical field of distributed storage, and the failover method comprises the following steps: determining a first disk resource group corresponding to a first storage node and a second disk resource group corresponding to a second storage node; and establishing a double-control relationship between the first disk resource group and the second disk resource group. And after the establishment of the double-control relationship is completed, carrying out fault switching on different disk resource groups under the double-control relationship based on the first mechanism and the second mechanism. The technical problem that in a distributed storage system, efficient and safe disk resource management and data access control of a fault controller cannot be effectively achieved under a double-controller architecture is solved, and the technical effects of improving rapidness, smoothness and automatic back-switching of fault switching and remarkably improving the high availability of the distributed storage system are achieved.

The invention relates to the technical field of intelligent driving control, in particular to a multi-camera acquisition system software synchronization method, which comprises the following steps of: distributing an independent acquisition thread and a global annular cache for each camera; each acquisition thread is blocked to wait for a new image to arrive, the image is stored in a corresponding annular cache, and a timestamp is updated; traversing the annular caches of the other cameras by taking the timestamp of one camera as a reference, and searching the image frame with the closest timestamp; whether the time deviation is smaller than a preset threshold value or not is judged, and if all the cameras meet the synchronization condition, a semaphore awakening algorithm processing thread is sent to read the synchronized image data from all the annular caches for processing. According to the method, the synchronization of the multi-camera data is realized on the basis of multiple threads and a semaphore mechanism which are originally provided by the system, the dependence on a complex middleware software framework is avoided, additional hardware synchronization equipment and protocols do not need to be configured, and the scheme has very outstanding advantages on an embedded platform with limited hardware performance.

The invention belongs to the technical field of nuclear power station maintenance management, and particularly relates to a nuclear power maintenance decision-making system and method based on multi-Agent cooperation. The system comprises an input layer, a multi-Agent cooperation layer, a knowledge support layer, a decision processing layer, an output and interaction layer and a feedback learning layer. The input layer receives initial work order information including equipment basic information and fault description; the multi-Agent collaboration layer allocates sub-tasks to predefined six types of professional Agents according to work order information and completes information interaction; the knowledge support layer comprises a nuclear power professional knowledge base, a historical maintenance database and a rule and regulation library; the decision processing layer performs conflict detection, negotiation and decision fusion on Agent output; the output and interaction layer provides a decision result display and man-machine interaction interface; and the feedback learning layer realizes comprehensive improvement of nuclear power maintenance decision-making efficiency and safety. The method has the beneficial effects that a multi-professional Agent collaborative network technical means is adopted, and the technical effects of maintenance decision cross-professional information instant sharing and efficient collaboration are realized.

An agricultural product carbon footprint acquisition and calculation method and system belong to the technical field of carbon emission calculation and agricultural informatization crossing, adopt an ecological pyramid intelligent architecture, and are composed of a data acquisition module, a data preprocessing and storage module, a carbon footprint calculation engine module, a traceability and safety module and an application service interface module. A'carbon footprint collection-accounting-evidence storage-identification-query 'full-chain quantitative management system is constructed, accurate collection of multi-source heterogeneous carbon source data is realized through cloud, edge and end cooperation, a modular carbon footprint calculation engine is combined with a localized factor database, CPU-GPU cooperative calculation and a digital twinning technology are adopted to realize accurate accounting and uncertainty quantification, and the carbon footprint calculation efficiency is improved. A CogVLM multi-mode model is innovated and improved to be embedded into agricultural carbon footprint field knowledge and a lightweight calculation engine, and deep fusion of visual perception and carbon footprint dynamic evaluation is achieved.

Dynamic allocation of work and data on a spatial compute architecture in which a lead dispatch node dynamically assigns a task of an application to a selected compute tile, and routes a corresponding active message to a local data memory pool of the selected compute tile. The active message may include instructions, data, and / or configuration parameters to enable the selected compute tile to execute the task. If there is insufficient space in the local data memory pool, the lead dispatch node routes the active message to shared memory or to a local data memory pool of another compute tile, and subsequently routes the active message to the local data memory pool of the selected compute tile when space is available. Alternatively, the lead dispatch node may forward the task and the active message to another lead dispatch node.

The invention relates to the technical field of data processing systems, and discloses a conference information intelligent data processing system based on multi-modal data fusion, and the system comprises a transaction template generation unit which calls a model from a preset rule base to generate a structured data container with an attribute field; the dual-channel feature extraction unit is used for distributing the voice to service data and a control instruction channel, extracting the service data to an independent cache region and monitoring a control keyword; the state control unit is used for unlocking the container only when the confirmation instruction is received, verifying data and writing a field to generate a transaction submission instruction; according to the system and the method, physical isolation of data flow and control flow is realized through a dual-channel architecture, a control instruction is used as a unique logic gating, non-decisive negotiation noise is filtered from the source, and atomicity and compliance of conversion from unstructured conference data to structured management transactions are ensured.

The invention discloses a shared memory lock implementation method and related device.The method comprises the steps that a synchronization request of a current task is obtained, the synchronization request is a locking request for a target resource in a shared memory or an unlocking request for the target resource, and the target resource in the shared memory is unlocked by the locking request or the unlocking request; the target resources are to-be-locked or to-be-unlocked resources of the current task; if the synchronization request is the locking request, under the condition that the value of a first atomic variable is a first preset value representing that the target resource is not occupied, modifying the value of the first atomic variable from the first preset value to a second preset value representing that the target resource is occupied; and if the synchronization request is an unlocking request, setting the value of the first atomic variable as the first preset value. According to the method, the target resource is locked and unlocked based on the value of the first atomic variable in the shared memory, and system calling is not executed in the locking and unlocking process, so that the overhead of the locking and unlocking process is reduced.

Apparatuses, systems, and techniques to facilitate graph code synchronization between application programming interfaces. In at least one embodiment, one or more circuits are to perform an application programming interface (API) to cause graph code to wait on a semaphore used by another API.

The invention provides a multi-channel optimization method of an underlying user mode file system adaptive to KV storage, and relates to the technical field of file system and storage system optimizing.The method comprises the steps that a local channel is distributed for each user thread, and an I / O request is forwarded to a service thread; i / O requests are divided into metadata requests and non-metadata requests according to types, and the metadata requests and the non-metadata requests are processed by a main thread and a service thread respectively; and the service thread performs task scheduling through a lock-free queue and a Work-Stealing algorithm, and dynamically expands a thread pool and binds computing resources according to a channel load. By means of the mode, efficient scheduling and execution of the high-concurrency I / O request in the user mode file system are achieved, thread contention and resource bottleneck are avoided, and the overall performance and response efficiency of a KV storage system in a high-concurrency scene are improved.

To eliminate the need for complex exclusive control processing in a task side executed by a processor.SOLUTION: A semiconductor device 10 comprises: a memory 13 which stores task management information 15 and in-execution group management information 16; a first PE 11a and a second PE 11b; and a first shared resource 12a and a second shared resource 12b. The first PE 11a or the second PE 11b are configured to: refer to the in-execution group management information 16; identify a group of tasks executable in the first PE 11a or the second PE 11b as an executable group; refer to the task management information 15; and determine the task associated with group identification information of the identified executable group, as the task to be executed next in the first PE 11a or the second PE 11b.SELECTED DRAWING: Figure 1

The embodiment of the invention provides an inference acceleration method and system based on a large language model. The method comprises the following steps of: firstly, reading a query vector, a key vector and a value vector which are calculated by a matrix processing kernel and are subjected to offset addition and quantization operation from a shared cache by virtue of a multi-head attention kernel in a system, respectively calculating an attention score and target key value data, and outputting data to be decoded; the single-node calculation efficiency is guaranteed through customized hardware pipeline design; then, the query vector is recalculated in the decoding stage, the target key value data stored in the high-bandwidth storage key value cache region in the pre-filling stage are directly reused, and resource waste caused by repeated weight loading is avoided; finally, a model parallel strategy is adopted, calculation tasks of the large language model are distributed to all accelerator nodes, distributed execution of large-scale matrix operation among multiple nodes is achieved, the node number can be flexibly expanded according to the model scale in the scheme, and the expansibility of the reasoning acceleration scheme is remarkably improved while reasoning efficiency is guaranteed.

The invention discloses a service request processing method and device, a medium, equipment and a program product, and relates to the technical field of smart home / smart home. According to the method, a first service request is received, the first service request comprises at least one target resource, a lock identifier is generated for the first service request, the lock identifier comprises a serial number used for identifying the receiving sequence of the first service request, and the lock identifier of the first service request is added into distributed queues corresponding to the at least one target resource. Under the condition that the lock identifier of the first service request is arranged at the head in the distributed queue corresponding to the at least one target resource, requesting to allocate a respective resource lock of the at least one target resource for the first service request; according to the method, the uniform lock identifier containing the serial number is generated for the service request, and resource lock coordination is performed based on the distributed queue, so that the consistency of a multi-resource locking sequence is ensured, deadlock is effectively avoided, and the stability and reliability of service processing in a high-concurrency scene are improved.

The invention discloses an urban fine pipe network drainage effect simulation method based on a nonlinear reservoir method, and the method comprises the steps: obtaining the rainfall, terrain, underlying surface and trunk pipe network data of a research region, building a two-dimensional hydrodynamic model and a one-dimensional pipe network model based on the obtained data, dividing a control region for each rainwater node, and carrying out the calculation of the control region. Establishing a coupling relationship between a ground surface grid in the control area and a corresponding rainwater node; calculating the rainfall and infiltration process of the research area to judge whether the rainwater nodes overflow or not, if overflow occurs, performing overflow calculation, if overflow does not occur, deducting a certain amount of water from each earth surface grid, counting the water amount, adding the water amount into the corresponding rainwater nodes respectively, and correcting the water amount by adopting a nonlinear reservoir method; and performing data updating according to the calculation result until the set operation time is reached, and outputting the final calculation result of the research area. The problem that the drainage capacity of the pipe network is difficult to generalize in the absence of fine pipe network data in the prior art is solved.

Apparatuses, systems, and techniques to perform an application programming interface (API) to add one or more graph nodes to a software graph, wherein the API is to cause a null-operation graph node to be added to a software graph, based, at least in part, on a dependency type indicated by the API. In at least one embodiment, one or more nodes are added to a graph in accordance to one or more dependency types.

The embodiment of the invention discloses a thread synchronization method, device, processing unit and system. Wherein the processing unit comprises a shared memory, a thread beam scheduler and an asynchronous acceleration unit. The shared memory is configured to update an internally stored target synchronization barrier based on the barrier update request, and in response to detecting that the target synchronization barrier satisfies a release condition, send thread wake-up information to the thread beam scheduler. The thread beam scheduler is configured to execute a thread sleep operation based on the arrival state of each target thread and trigger a first update request for the target synchronization barrier, and execute a thread wake-up operation based on the thread wake-up information in response to receiving the thread wake-up information. The asynchronous acceleration unit is configured to trigger a second update request for the target synchronization barrier based on a completion state of each target asynchronous transaction operation. According to the embodiment of the invention, the synchronization between the threads and between the threads and the asynchronous transaction operation can be realized on the premise of not introducing too many hardware structures.

The invention discloses a multi-physics field coupling numerical method for simulating wave-ocean current-earthquake-seabed interaction, and relates to the technical field of ocean engineering numerical simulation, and the method comprises the following steps: model construction, load application, coupling solution, and result analysis and application. According to the method, revolutionary simulation of collaborative response of a seabed and a structure in a complex marine environment and multi-physics field high-precision bidirectional coupling of random waves, steady-state ocean currents and seismic excitation are realized by constructing a total-factor coupled high-precision numerical model, the problems of incomplete coupling dimensions and excessive model simplification of a traditional method are solved, and the method has the advantages of being high in practicability and high in reliability. According to the wave-induced seabed liquefaction simulation method, the simulation result and the actual working condition coincidence degree are remarkably improved, the frequency domain transfer function method is introduced to analyze the seabed pore pressure response, the cognitive depth and the prediction precision of the wave-induced seabed liquefaction mechanism are remarkably improved, and in addition, through efficient algorithms such as GPU parallel calculation, self-adaptive grids and frequency domain-time domain mixed solution, the wave-induced seabed liquefaction simulation method has the advantages that the wave-induced seabed liquefaction simulation precision is improved. And the calculation efficiency is greatly improved while the precision is ensured.

The invention provides a process information monitoring method and device, equipment, a medium and a product, and relates to the technical field of information security. The method comprises the steps that historical static process information of a plurality of processes is taken, and a static process fingerprint database is generated according to the historical static process information; taking historical dynamic process information of the plurality of processes, and generating a dynamic process fingerprint database according to the historical dynamic process information; taking real-time static process information and real-time dynamic process information of the target process, judging whether the real-time static process information is in the static process fingerprint database or not, and if the real-time static process information is in the static process fingerprint database, conditionally releasing the target process; and judging whether the real-time dynamic process information exists in the dynamic process fingerprint database or not, and if yes, completely releasing the target process. A double-fingerprint database verification mechanism is adopted, after an abnormal behavior is injected into a target process, the target process is released through matching of a static process fingerprint database, but a dynamic process fingerprint database detects that the target process deviates from a normal behavior during operation, so that data is intercepted, and the security of the data is remarkably improved.