494 results about "Concurrency" patented technology

The invention provides an automated task parallel method suitable for distributed machine learning and a system thereof. The method and the system solve defects of a programming interface in existing distributed machine learning, and tight coupling of system data access behaviors and application logic caused by just providing a reading-writing interface of key value pairs. The defect intensifies network bandwidth resource competition in distributed cluster, and causes that programming personnel is difficult to perform parallelization on a task. The system comprises a work node module, a service node module, a host node module, a tensor module, a scheduling module, a message tracking module, a stage module, a stage group module, and an executing engine module. Through providing higher-level programming abstraction, the system decouples logic of reading-writing access behaviors and an application program. In operation, the system firstly dynamically partitions tasks according to the load condition of a service node, and then machine learning tasks are automatically executed in a parallel manner, so as to greatly reduce burden of programming personnel to compile high-concurrency machine learning applications.

The invention relates to a method for test, simulation and concurrence of software performance. The method for the test, simulation and concurrence of the software performance specifically comprises the steps that (1) user configuration information which is input by a user is read, (2) a user requirement structural body is stored in a shared memory module and mapping is established, (3) service requests of concurrent users are received and at least one test progress is established according to the number of the concurrent users and the user requirement structural body, (4) test threads are established, (5) each test thread is used for processing the service request of a corresponding user and is stopped when a stopping condition is met, (6) operation is ended and the test threads in each test process are stopped after operation of the test threads in each process is ended in sequence, (7) relevant data of each service are stored, analyzed and counted, and then all the processes are finished. According to the method and system for the test, simulation and concurrence of the software performance, the fact that how to simulate user concurrence is explained, bottlenecks are prevented from occurring, and the purpose of a high-concurrency scene by means of a small number of hardware sources is achieved; concurrency stability is guaranteed; support to different user services is achieved; help is provided for positioning and development cycle shorting.

Automated refactorings as implemented in modern IDEs for Java usually make no special provisions for concurrent code. Thus, refactored programs may exhibit unexpected new concurrent behaviors. We analyze the types of such behavioral changes caused by current refactoring engines and develop techniques to make them behavior-preserving, ranging from simple techniques to deal with concurrency-related language constructs to a framework that computes and tracks synchronization dependencies. By basing our development directly on the Java Memory Model we can state and prove precise correctness results about refactoring concurrent programs. We show that a broad range of refactorings are not influenced by concurrency at all, whereas other important refactorings can be made behavior-preserving for correctly synchronized programs by using our framework. Experience with a prototype implementation shows that our techniques are easy to implement and require only minimal changes to existing refactoring engines.

The invention provides a Hadoop cluster-based task scheduling method. The method comprises the following steps of 10, setting that a business has N priorities, enabling each priority to correspond toone scheduling queue, enabling each task to specify the corresponding priority according to business importance, entering the scheduling queue of the corresponding priority, and performing queuing according to a time sequence, wherein N is a positive integer; 20, setting a maximum task concurrency number of a system and a maximum task concurrency number of the queue corresponding to each priority,scheduling the tasks in the scheduling queues according to a priority sequence from high to low at a time interval; 30, automatically distributing the queued tasks to the task idle priority queue forscheduling in a balanced manner according to the priority sequence from high to low; and 40, enabling the tasks of running failure and the tasks exceeding a predetermined maximum running time to re-enter the original priority queue for queuing and scheduling. The invention furthermore provides a computer device. Task priority grouping, queuing scheduling, route strategies and fault transfer are realized, so that the scheduling efficiency of the cluster tasks is greatly improved.

The invention belongs to the technical field of electronic commerce, and in particular relates to a system and a method for processing a high-concurrency data request. A main body of the system provided by the invention is divided into three parts, namely, a load balancing layer, an arbitration application unit and a goods deduction processing unit, wherein the load balancing layer is used for equally distributing a great amount of client requests to arbitration applications at a rear end; the arbitration application unit is used for receiving user requests which are distributed by the load balancing layer, carrying out qualification judgment, merging and recombining the user requests, and then transmitting to the goods deduction processing unit to carry out deduction processing of goods; the goods deduction processing unit is used for receiving goods deduction requests from the arbitration applications, and carrying out inventory quantity deduction on second-kill goods. By virtue of a single-thread batch processing manner, the bottleneck of a database system caused by multi-thread high concurrency is avoided; meanwhile, high business handling capability is reserved; the cost of system resources is also reduced as low as possible.

Various methods and apparatus are described for communicating transactions between one or more initiator IP cores and one or more target IP cores coupled to an interconnect. Tag logic may be located within the interconnect, such as located in an agent, and configured to assign different interconnect tag identification numbers to two or more transactions from a same thread. The tag logic assigns different interconnect tag identification numbers to allow the two or more transactions from the same thread to be outstanding over the interconnect to two or more different target IP cores at the same time, allow the two or more transactions from the same thread to be processed in parallel over the interconnect, and potentially serviced out of issue order while being returned back to the multiple threaded initiator IP core realigned in expected execution order.

The invention provides a petri-network-based control method for an automatic manufacture system. Through operation of a deadlock avoidance algorithm, a robustness enhancing algorithm and a concurrency improving algorithm, a group of transition set can be produced. Any transmission of transmission belonging to the set can meet requirements for zero deadlock, robustness and system concurrency improvement. After each transition is transmitted, the three algorithms need to be calculated again successively and a new transition set is produced. Through said circulation, a group of event sequence is produced dynamically in real time. According to the invention, when resource faults of a system occur, programs that do not need the fault resources are not blocked by programs that need the fault resources, so that smooth processing can continue and system concurrency can be improved.

The invention discloses a concurrency OLAP (On-Line Analytical Processing)-oriented test data hierarchy cluster query processing system and a concurrency OLAP (On-Line Analytical Processing)-oriented test data hierarchy cluster query processing method. The I/O price of a test data access in the concurrency OLAP is reduced by a series of DBMS (Data Base Management System) technologies of hierarchy test count, cluster, an index organization chart, large object-based storage and the like, so the processing capacity of concurrency query is imporved. According to the system and the method, the concurrency query processing and optimization techniques can be realized in a database management system by aiming at the I/O performance and concurrency OLAP processing performance, and the oriented I/O performance is supported to optimize and set a concurrency OLAP so as to process a load, thereby improving the predictable performance displayed by test data waveforms, and realizing the accelerated processing of the test data query of the large-scale concurrency OLAP.

The invention provides a system and method for predicting the non-steady application user concurrency in a cloud environment, and relates to the technical field of service performance optimization in the cloud environment. The system comprises a historical data obtaining module, a non-steady service judgment module and an application user concurrency prediction module; historical data is analyzed; identification of service properties is carried out; therefore, the fact that a concurrency sequence has the tendency or the periodicity is judged; and concurrency prediction is carried out respectively for the two properties. According to the system and the method provided by the invention, for the characteristics of the tendency and periodic application user concurrency of non-steady services in the cloud environment, a prediction model is constructed by utilizing different prediction methods; the concurrency containing two properties is predicted; characteristics of a tendency and periodic concurrency sequence can be automatically identified; furthermore, the periodicity of the sequence can be automatically calculated; manual intervention is unnecessary; the prediction process can be automatically completed; and simultaneously, the non-steady service concurrency prediction accuracy can be effectively increased.

The invention relates to an industrial 5G network technology, in particular to a dynamic resource allocation method for a high-concurrency multi-service industrial 5G network. The method comprises thefollowing steps: establishing a network system model, and constructing a machine learning model for high-concurrency multi-service-oriented industrial 5G network dynamic resource allocation; collecting the state, action and reward information of all industrial equipment in an industrial 5G network at different time slots, and training the machine learning model; circularly evaluating network performance indexes of different services, and training the machine learning model until performance requirements are met; and taking the state information of all the industrial equipment in the current time slot industrial 5G network as the input of the machine learning model, and carrying out resource allocation on a plurality of different types of concurrent services of all the industrial equipment. According to the invention, the problem of resource conflicts caused by different requirements of various types of services such as control commands, industrial audios and videos, perception data and the like on real-time performance, reliability and throughput in the concurrent communication process of large-scale heterogeneous industrial equipment in an industrial 5G network is solved.

The invention discloses a vulnerability mining model construction method based on swarm intelligence. The vulnerability mining model construction method comprises the following steps: step 1, modelinga user individual; step 2, performing intelligent decomposition on a vulnerability mining task based on a vulnerability mining scene; 3, solving an optimal vulnerability mining path; and step 4, gathering and fusing crowd sourcing vulnerability mining results and establishing a feedback learning model. Compared with the prior art, the invention has the advantages that: the invention has positiveeffects; the model reflects crowd-sourcing concepts such as crowd collaboration, machine collaboration, computing resource collaboration and tool sharing; the method has the advantages of natural high-concurrency capability, rapid and efficient search and solution capability, extremely strong environmental adaptability, high robustness and self-recovery capability, strong expandability and high flexibility, the vulnerability mining efficiency is effectively improved, and the vulnerability mining cost is reduced.

The invention discloses a transaction processing method and device, computer equipment and a storage medium, and belongs to the technical field of databases. The method comprises the steps of obtaining a read-write set of a target transaction in an execution stage of the target transaction, verifying whether the target transaction meets a self concurrency consistency level or not according to a logic starting moment and a logic submitting moment of the target transaction in a verification stage, and dynamically updating the logic starting moment based on a predetermined strategy. Thus, the logic starting moment and the logic submitting moment which are finally updated are enabled not to generate data exception which is not allowed by the own concurrency consistency level under the legal condition, and then the target transaction is submitted, so that different transactions in the whole system can correspond to different concurrency consistency levels, and the user experience is improved. The data exception detection accuracy is greatly improved, and the transaction processing efficiency and the performance of a database system are improved.

The invention relates to a cross-platform station automatic monitoring system testing method based on concurrency simulating tools. Two-stage concurrency simulating tools are adopted to finish high-efficiency test, i.e. a 103 scale protocol simulating tool and an LR load pressure testing tool. The testing method realizes the purpose of constructing a large-scale station automatic monitoring system testing environment by using a minimized device emulation computer simulating tool and ensures that a station spacer layer is virtualized into a data transmission request processing center of a single similar clouding computing by adopting several computers, thereby the systematic difference characteristic of the cross-platform monitoring system of a ground shielding station control layer is maximized. The invention can solve many problems caused by the traditional simple mechanical laboratory test mode and realize high-efficiency, reliable and long-term stability simulation test of the cross-platform station automatic monitoring system.

The invention discloses a business service test method and device, a storage medium and an electronic device. The method comprises the steps that a target test result obtained by testing a business service is acquired, wherein the target test result is obtained after a test request is sent to a server of the business service according to a target concurrency number, and the target concurrency number is used for indicating the concurrency number of the test request sent in a target time period; a first concurrency number is determined according to the target test result and the target concurrency number, wherein the first concurrency number is the concurrency number of the test request sent in the first time period, the time length interval of the first time period is the same as the time length interval of the target time period, and the starting time of the first time period is not earlier than the ending time of the target time period; and a test request is sent to the server according to the first concurrent number in the first time period to obtain a first test result. According to the invention, the technical problem of low efficiency of service testing in related technologiesis solved.

The invention relates to the technical field of server high-concurrency processing, and discloses a multi-thread concurrency monitoring method, device and apparatus and a storage medium, which are used for preventing the conditions of memory overflow and too slow response of a server and improving the running stability of the server. The method comprises the steps of creating tasks in batches andsubmitting the tasks to a task queue, and an original value of a task timer of a locking tool is initialized to be the total number of the tasks of sub-threads; traversing the sub-thread tasks and recording the number of to-be-executed tasks; judging whether the task number exceeds a preset threshold value or not; if the preset threshold value is not exceeded, judging whether a queue is full or not; if the queue is not full, submitting the sub-thread task to the sub-thread and blocking the main thread; recording running result information of each task; after the task operation of each sub-thread is finished, subtracting one from the value of a task timer of the locking tool and returning task operation result information to a main thread; and when the value of the task timer is 0, recovering the main thread to execute the corresponding main thread task.

The invention discloses a special circuit for a neural network and a related computing platform and implementation method thereof. The special circuit comprises: a data reading module which comprisesa feature map reading sub-module and a weight reading sub-module which are respectively used for reading feature map data and weight data from an on-chip cache to a data calculation module when a depthwise convolution operation is executed, wherein the feature map reading sub-module is also used for reading the feature map data from the on-chip cache to the data calculation module when executing pooling operation; a data calculation module which comprises a dwconv module used for executing depthwise convolution calculation and a pooling module used for executing pooling calculation; and a datawrite-back module which is used for writing a calculation result of the data calculation module back to the on-chip cache. The use of hardware resources is reduced by multiplexing the read logic andthe write-back logic of the two types of operations. The special circuit provided by the invention adopts a high-concurrency pipeline design, so that the computing performance can be further improved.

The invention discloses a method and a device for adjusting the garbage collection concurrency number in a storage array and a readable medium. The method comprises the steps of detemining the maximumvalue of IO writing requests according to the data processing capacity of a current storage array; dividing the number of the IO writing requests into a plurality of adjacent intervals from zero to the maximum value, and setting a corresponding relationship between each interval and the interval of the garbage collection concurrent number; querying the number of the currently processed IO writingrequests every other preset time, and judging whether the current garbage collection concurrent number conforms to the corresponding interval or not based on the number of the currently processed IOwriting requests and the corresponding relationship; in response to the fact that the garbage collection concurrent number is larger than the maximum value of the corresponding interval, issuing a newgarbage collection request in a delayed mode; and in response to the fact that the garbage collection concurrent number is smaller than the minimum value of the corresponding interval, issuing a newgarbage collection request. According to the scheme provided by the invention, the garbage collection concurrency is automatically adjusted according to the corresponding relationship between the garbage collection concurrency and the read-write IO concurrency determined by the system processing capacity.

The invention provides a task scheduling execution method and device, and the method comprises the steps: obtaining a to-be-executed first task list every first preset duration; determining a second task list preferentially executed in the first task list according to a depth-first search algorithm; and scheduling execution resources to execute the tasks in the second task list. According to the embodiment of the invention, the method comprises the steps: obtaining a to-be-executed first task list every first preset duration; according to the method, the second task list which is executed preferentially is determined according to the depth-first search algorithm, and the execution resources are scheduled preferentially to execute the determined second task list, so that core emergency tasks with high execution priorities of the execution resources can be scheduled preferentially in the high-concurrency time period of the tasks, and the timeliness of task scheduling execution is improved.

The invention discloses a multi-core interaction method and system suitable for a nonvolatile memory and a storage medium. The system comprises an interaction request submitting module, an interactionrequest scheduling module, a kernel thread management module, a CPU load detection module and a system crash recovery module. According to the method, an interaction request environment is initialized through an interaction request submitting module, then batch interaction requests are optimized and distributed through an interaction request scheduling module, and then kernel threads are distributed to each CPU through a kernel thread management module. According to the method, batch interaction requests are processed, the advantages of multi-core asynchronous concurrent processing can be brought into full play, the efficiency is high, and the concurrency is high; besides, the load condition of each CPU can be acquired in real time through the CPU load detection module, and the system data can be recovered through the system crash recovery module, so that the stability of the system is improved, and the method and system can be widely applied to the technical field of computers.

The invention provides a flow limiting method and device for a high-concurrency system, a storage medium and equipment. The method comprises the following steps: judging whether the resource access amount of the high-concurrency system meets a preset flow limiting condition or not; if the resource access amount of the high-concurrency system meets a preset flow limiting condition, carrying out priority division on the received resource access requests, and caching the resource access requests to cache queues with corresponding priorities according to the priorities of the resource access requests; and sequentially processing the resource access requests in each cache queue according to the priority sequence of the cache queues. Flow limiting can be performed in system high-concurrency access, normal operation of a core system is guaranteed, and the user experience is improved.

The invention provides a merging request processing method and device, equipment and a storage medium, and the method comprises the steps: sequentially writing to-be-processed user requests into a preset cache pool for caching, and enabling the user requests to comprise request parameters; when the execution environment reaches a preset execution condition, extracting at least two request parameters from the cache pool, and creating a batch processing request based on the at least two request parameters; obtaining batch feedback data generated by executing the batch processing request; and distributing the batch feedback data to a user terminal corresponding to the user request. By accumulating the user requests, when the execution environment reaches the preset execution condition, the user requests are processed in batches by combining the batch processing operation capability of a database, so that the burden of the database in a high-concurrency scene is reduced, and the high-concurrency processing capability of the system is improved.

The embodiment of the invention relates to the field of communication, and discloses a performance detection method, a performance detection device, a server and a storage medium. The method comprisesthe following steps of obtaining the L performance index values influencing the concurrency of a system to be tested are obtained, wherein L is an integer larger than 1; performing quantitative calculation on the influence of the L performance index values on the concurrency, and obtaining an adjustment value of the concurrency according to the quantitative influence of each performance index value on the concurrency; and if the adjustment value meets the preset condition, taking the concurrency after adjustment of the adjustment value as a performance detection result. Therefore, under the condition that multiple performance indexes influence the concurrency of the system, automatic detection can also be realized, the performance detection efficiency is improved, and the work of manuallyparticipating in detection parameter adjustment and data analysis in the performance detection process is avoided.