36 results about "Run queue" patented technology

The invention relates to a dynamic CPU scheduling method, system and device based on a CPU busy state and a medium. The method comprises the steps that busy degree indexes of all CPU cores are dynamically generated through monitoring; service logic to be executed is analyzed into a task set, and task attributes which cannot be changed are distributed to tasks; determining respective target CPU cores based on busy degree indexes of all current tasks, generating binding parameters, and writing the binding parameters into task attributes of the corresponding tasks; executing the allocation logic according to the task attribute of each task, and submitting the task to a running queue of the target CPU core to wait for execution; and when the single execution duration of the task in the target CPU core reaches a preset threshold value or is terminated by external interruption, resetting the corresponding task into the operation queue. According to the invention, breakthrough improvement is realized in three dimensions of hardware resource utilization rate, system real-time performance and scheduling security, and a solution with response speed and stability is provided for task scheduling in a computing environment.

The invention relates to the technical field of artificial intelligence, and discloses a large language model high concurrency reasoning method and system, and the method comprises the steps: calculating the size of a video memory block through an actuator, and distributing a video memory space; the scheduler is used for converting the request sequence and putting the request sequence into a waiting queue of the scheduler; the scheduler allocates a corresponding video memory block to each request sequence until each request sequence can perform next reasoning; the scheduler calculates the video memory requirement of the request sequence in the waiting queue according to the priority sequence, and transfers the request sequence in the waiting queue to the running queue; according to the number of pre-filling types and the number of decoding types of the request sequence, distributing the number of video memory blocks used for executing pre-filling reasoning or decoding reasoning; therefore, according to the method, continuous batch processing, a dynamic space allocation mechanism and a task scheduling framework are adopted, the parallel reasoning capability of continuous batch processing is fully utilized, the concurrency and throughput of large model reasoning are improved, and the limitation that traditional continuous batch processing needs space pre-allocation is solved.

The invention discloses a value maximization calculation unloading method and system based on dynamic priority scheduling and reinforcement learning. According to the method, an optimization framework with maximization of the'long-term average task value 'of the system as a core target is constructed, and the task value is defined by introducing an exponential value function, so that the problem of'value neglect' existing in a traditional'average delay or energy consumption optimization 'strategy is solved. According to the value-oriented optimization mechanism, the multi-agent system can autonomously learn a strategy which preferentially ensures high-value task execution when resources are limited in the training process. In addition, the invention further provides a dynamic priority function which can accurately measure the actual emergency level of the task at different moments, so that the defect that a static priority model is difficult to deal with the real-time change of the emergency state is overcome. Particularly, the dynamic priority method is not only used for task selection in a waiting queue, but also innovatively introduced into a resource dynamic allocation mechanism of a running queue.

The present application provides a task migration method and a related apparatus, which are applied to an electronic device, the electronic device comprising one or more processors. When the electronic device detects that a highest-priority task in an operating queue on a second processor has been executed, a highest-priority task in a waiting queue on a first processor is migrated to the second processor to be executed. A highest-priority task exists on each of the first processor and the second processor, the highest-priority task on the first processor waiting in a queue, and the highest-priority task on the second processor being currently executed. When the highest-priority task in the operating queue on the second processor has been executed, this can indicate that the highest-priority task does not exist on the second processor at the present moment, and balancing is performed at this moment, which can ensure that the highest-priority task migrated to the second processor is executed in a timely manner. Compared with existing load balancing, migration timeliness and success rate can be improved.

This invention relates to the field of computer task management technology, and in particular to a task management method for Electron desktop applications based on IndexedDB storage. The method includes a task management module that listens for user-issued tasks; a task management trigger judgment method that modifies the queue type of a preset number of tasks to a running queue; a trigger method when the waiting task queue changes, wherein the waiting task queue enters the running task queue in a first-in-first-out manner; and a trigger method when the running task queue changes, executing the background interface or Electron's Node.js service corresponding to the task type of the unexecuted or paused tasks. This invention provides a complete desktop application task management solution that, compared to other storage methods, handles large-scale task processing with ease. IndexedDB's shared-source strategy ensures data security.

The quality of service guaranteed function-free computing system comprises an HTTP gateway, a scheduler and a work pool; the HTTP gateway serializes the incoming DAG request through a serializer, and sequentially adds the serialized function-free function in the DAG request to a waiting queue; the scheduler retrieves the function-free function from the waiting queue and the chain running queue to the work pool for processing according to a scheduling strategy. The quality of service guaranteed function-free system and method of the application process the incoming request through the producer-consumer mode by reforming the scheduler in the platform. In addition, a strategy framework containing rich strategies is added. Different kinds of strategies are set for different kinds of application requests, so that the queuing time is reduced to a certain extent, and the running performance is improved.

The invention discloses a task scheduling method, electronic equipment, a storage medium and a computer program product. The method comprises the following steps: detecting an operation mode of a physical processor to obtain a detection result; in response to the detection result indicating that the physical processor is switched from the physical mode to the virtualization mode, a target scheduling priority to be scheduled is determined based on an operation bitmap in an operation queue structure corresponding to the physical processor, and different bits in the operation bitmap correspond to different scheduling priorities respectively; and scheduling the to-be-executed scheduling task in the target scheduling queue corresponding to the target scheduling priority. The technical problems of poor task scheduling timeliness, low accuracy and low system resource utilization rate are solved.

A method and apparatus for managing threads are provided. The method includes creating a first thread, the first thread being a kernel mode thread, the first thread having a first thread context; creating a second thread using the first thread, wherein the second thread is a user mode thread, and the second thread inherits the first thread context; controlling the first thread to enter an idle loop state after the second thread is enqueued to a run queue; and selecting and executing the second thread from the run queue using a scheduling thread.

The application discloses a frequency control method and device of a CPU core and electronic equipment, and relates to the technical field of terminals. The specific technical scheme is as follows: the electronic equipment can acquire an interactive thread list; the interactive thread list comprises thread information, which is used for indicating an interactive thread for processing a user-perceptible event; and a frequency upper limit value of a first CPU core and at least one second CPU core is adjusted according to a first load parameter and a first power consumption threshold value; wherein a running queue of the first CPU core comprises the interactive thread; the first load parameter is a load parameter of the first CPU core running the interactive thread; the at least one second CPU core is a CPU core other than the first CPU core; and the first power consumption threshold value is a total power consumption upper limit value of the first CPU core and the at least one second CPU core.

The embodiment of the invention provides a thread processing method and device, relates to the technical field of computers, and can solve the problem of priority inversion in a conditional variable scene and improve the execution efficiency of threads. The method is applied to the electronic equipment, the electronic equipment comprises a target processor, the target processor comprises a target running queue, the target running queue comprises a first thread, a second thread and a third thread, and the method comprises the steps that under the condition that the condition of a target condition variable is not met, the first thread enters a sleep state, and the third thread enters a sleep state; the data is not cleared from the target operation queue; the condition of setting the target condition variable by the second thread is met, and the first thread is awakened; under the condition that the condition of the target condition variable is met, the first thread continues to run; after the first thread runs, the third thread runs, and the priority of the third thread is higher than that of the second thread and lower than that of the first thread.

This invention relates to a user-mode thread scheduling method, apparatus, and storage medium based on behavior prediction. A short-term behavior predictor calculates prediction results using a hybrid model. A scheduling decision engine receives these prediction results, calculates a comprehensive benefit score based on the prediction results, and generates scheduling and migration decisions. A core affinity and migration controller executes migration decisions, controls the computation core to run CPU-intensive tasks, controls the I / O core to bind I / O interrupts and run I / O-intensive tasks, and controls the scheduling core to select user-mode thread tasks from a multi-level run queue and allocate them to the computation core or I / O core for execution. The scheduling and migration decisions involve resource allocation and priority adjustment. A lightweight behavior collector can collect the operational characteristics of user-mode threads with low overhead. While maintaining low overhead, a hybrid model is used to dynamically model thread behavior characteristics, predict short-term thread operational trends, and perform resource allocation and priority adjustment accordingly, thereby improving the overall system throughput.

The application provides an artificial intelligence-based task flow accurate scheduling method and device, electronic equipment and storage medium. The artificial intelligence-based task flow accurate scheduling method comprises the following steps: constructing a task link graph of each preset task flow, the preset task flow comprising a plurality of task nodes, the task link graph comprising a preset ID, a cycle type of the preset task flow and an upstream and downstream relationship of the task nodes; judging a dependency relationship between any task nodes in different task link graphs to draw a dependency graph, the dependency relationship comprising a dependent task node and a dependent cycle number; obtaining state information of each task node, the state information comprising a running queue and a running cycle number, the running queue comprising output results of the task nodes in each running cycle; calculating a scheduling priority of each task link graph based on the state information, the dependency graph, the cycle type and a preset priority; and scheduling the preset task flow corresponding to the task link graph according to the scheduling priority to obtain a scheduling result. The application can realize accurate scheduling of the task flow.

A system, computer-implemented method, and computer program product for facilitating the scheduling of quantum computing jobs are provided. According to an embodiment, the system may include a memory storing computer-executable components and a processor executing the computer-executable components stored in the memory. The computer-executable components may include a scheduler component that may determine an execution order for quantum computing jobs based on one or more quantum-based execution constraints. The computer-executable components may also include a run queue component that may store quantum computing jobs based on the execution order. In an embodiment, the scheduler component may determine the execution order based on the availability of one or more qubits having a defined fidelity level.

The invention discloses a batch task scheduling method and device, equipment, a medium and a product, and relates to the technical field of information processing.The method comprises the steps that task information of a to-be-executed task is obtained, the to-be-executed task comprises at least three batch tasks, the task information comprises at least three task data, and the task data and the batch tasks are in one-to-one correspondence; based on the task data of each batch task, determining at least two task groups of the to-be-executed task and reference weight values of the at least two task groups; and based on the reference weight values of the at least two task groups, adding the at least two task groups to a waiting queue, and when a resource release instruction of the running queue is received, processing the at least two task groups in sequence according to the resource release instruction. The batch tasks are grouped according to the task data of the batch tasks, and at least one task group is started according to the resource release condition of the running queue, so that the utilization rate of software and hardware resources is improved while the time limit requirement of the batch tasks is ensured.

Provided in the present disclosure are task scheduling methods, an electronic device, a storage medium and a computer program product. A task scheduling method comprises: detecting the running mode of a physical processor to obtain a detection result; in response to the detection result indicating that the physical processor has switched from a physical mode to a virtualized mode, determining, on the basis of a run bitmap in a run queue structure corresponding to the physical processor, a target scheduling priority for pending scheduling, wherein different bits in the run bitmap respectively correspond to different scheduling priorities; and performing scheduling processing for scheduling tasks to be executed in a target scheduling queue corresponding to the target scheduling priority. The present disclosure solves the technical problems of poor task scheduling timeliness, low accuracy and low system resource utilization rate.

The invention provides a lift pump automatic operation method and system based on queue management, a medium and equipment. The lift pump automatic operation method comprises the following steps that state information of all lift pumps is obtained in real time; establishing a ready queue, a running queue and a start-stop queue of the lift pump based on the lift pump state information; the current operation number of lifting pumps and current water level information are obtained; based on the current water level information, the start-stop demand number of the lifting pumps is obtained; and adjusting the operation state of the lift pump based on a comparison result of the required start and stop number of the lift pump and the current operation number of the lift pump and queue information. According to the lift pump automatic operation method and system based on queue management, the medium and the equipment, a reasonable pump operation strategy is provided, and the problems that in the current production process, equipment use is not uniform, and water level fluctuation is large are solved.

A central processing unit, CPU (10, 20), comprising: a processing circuit arrangement; and a control unit extracting instructions from a memory unit (18), the instructions causing the control unit to: to execute respective threads (210, 212, 214) by means of the processing circuitry, which threads are held in a local run queue (205) associated with the processing circuitry and with a poller, wherein the control unit interleaves the execution of the respective threads (210, 212, 214) and the poller, randomly selects the poller to be executed after the execution of a thread in the local run queue (205) from a shared poller array (240), and wherein the execution of the poller by the processing circuitry causes the poller to determine whether the CPU has a task to perform for a device external to the CPU; after clearing the local run queue (205), checking a buffer (215) shared by a group of CPUs comprising the CPU and one or more additional CPUs for threads to be executed; taking a stack of threads (220, 222, 224, 226, 228) associated with the group of CPUs from the run queue of the buffer (215); to place the stack of threads (220, 222, 224, 226, 228) into the local run queue (205); and to execute, by the processing circuitry, a respective thread in the stack of threads (220, 222, 224, 226, 228) from the local run queue (205).

This disclosure provides a task scheduling method, apparatus, and electronic device for a Completely Fair Scheduler (CFS), relating to the fields of artificial intelligence and cloud computing technologies, and particularly to operating system scheduling technology. The implementation scheme includes: determining the waiting time of ready tasks in the processor's run queue; updating the processor's latency information based on the waiting time of a ready task leaving the run queue when the ready task leaves; comparing the maximum waiting time of ready tasks in the run queue with a latency threshold, wherein the latency threshold is based on the processor's latency information; and scheduling the ready task corresponding to the maximum waiting time in response to the maximum waiting time exceeding the latency threshold. Using embodiments of this disclosure, the latency threshold for prioritizing task scheduling is dynamically determined by utilizing the processor's latency, thus achieving dynamic scheduling based on system operating conditions on top of CFS.

The present disclosure provides a video memory management method and device, equipment and medium, relates to the technical field of computers, and particularly to the field of artificial intelligence and intelligent chips. The implementation scheme is as follows: an operation queue of a target task is obtained; for each group of video memory management events, the group of video memory management events and all events between the group of video memory management events in the operation queue are encapsulated as an event function; for each event function in a plurality of event functions corresponding to a plurality of groups of video memory management events, first information of a video memory space corresponding to the event function is determined; and a plurality of sub-tasks are sequentially issued to a processing unit.

An electronic device according to an embodiment of the present disclosure may: select at least one process for memory recovery from among processes being executed in response to an out-of-memory event having occurred; add the at least one process to a first run queue managed by a first scheduler; and arrange processes to be terminated in the first run queue in the order of memory usage and sequentially terminate said processes.

The embodiment of the invention provides a thread processing method and device, relates to the technical field of computers, and can solve the problem of priority inversion in a read-write lock scene and improve the execution efficiency of threads. The method is applied to the electronic equipment, the electronic equipment comprises a target processor, the target processor comprises a target operation queue, the target operation queue comprises a first thread, a second thread and a third thread, and the method comprises the steps that under the condition that the first thread holds a read lock of a target resource, the second thread enters a sleep state and is not cleared from the target operation queue; under the condition that the first thread releases the read lock of the target resource, the second thread is awakened and holds the write lock of the target resource; after the second thread runs, the third thread runs, and the priority of the third thread is lower than that of the second thread and higher than that of the first thread.

The present invention relates to the field of artificial intelligence technology, and discloses a high-concurrency reasoning method and system for a large language model. The method comprises the following steps: utilizing an executor to calculate the size of a memory block and allocate memory space; utilizing a scheduler to convert a request sequence and place it into a waiting queue of the scheduler; the scheduler allocates a corresponding memory block to each request sequence until each request sequence is capable of performing the next reasoning; the scheduler calculates the memory requirements of the request sequences in the waiting queue according to priority order, and transfers the request sequences in the waiting queue to a running queue; and allocating the number of memory blocks for executing pre-filled reasoning or for executing decoding reasoning based on the number of pre-filled types and the number of decoded types of the request sequence. Therefore, the present invention adopts continuous batch processing, a dynamic space allocation mechanism, and a task scheduling framework, fully utilizing the parallel reasoning capability of continuous batch processing, improving the concurrency and throughput of large model reasoning, and overcoming the limitation of traditional continuous batch processing requiring pre-allocation of space.

The application discloses a processor task scheduling method and device and a storage medium. The application relates to the field of computer operating systems, and the method comprises the following steps: acquiring a centralized queue, wherein the centralized queue is used for storing a plurality of to-be-allocated tasks and priorities corresponding to the to-be-allocated tasks; when a running queue of a target processor is empty, selecting a to-be-allocated task stored in the centralized queue to obtain a first task, and reserving the first task for the target processor to add the first task to the running queue of the target processor; and determining a target task to be executed by the target processor according to the first task to control the target processor to execute the target task. The application solves the technical problem that the performance overhead of processor task scheduling is high.

The invention relates to a method, a device and a computer program product for backup. The method comprises the following steps: determining a plurality of parameters of a plurality of copy sessions in a waiting queue in response to the fact that a backup execution command is triggered; the method further includes determining a plurality of weights for the plurality of replication sessions based on the plurality of parameters for the plurality of replication sessions. The method further includes selecting at least one replicated session from the plurality of replicated sessions into a run queue based on the plurality of weights for the plurality of replicated sessions. The method further includes performing a backup task for at least one replication session in the run queue. In this way, each copy session is subjected to priority ranking, and a plurality of parameters are considered during ranking, so that the backup efficiency is improved, the number of waiting and failed tasks is reduced, data are better protected, the copy capacity of the system is enhanced, the copy limit number is increased, and the memory of the system is prevented from being used up.

The invention discloses a task scheduling method and device of a processor and a storage medium. Relates to the field of computer operating systems, and the method comprises the steps that a centralized queue is obtained, and the centralized queue is used for storing a plurality of to-be-distributed tasks and priorities corresponding to the to-be-distributed tasks; when the operation queue of the target processor is empty, selecting the to-be-allocated tasks stored in the centralized queue to obtain a first task, and reserving the first task for the target processor so as to add the first task into the operation queue of the target processor; and determining a target task to be executed by the target processor according to the first task to control the target processor to execute the target task. The technical problem that the task scheduling performance overhead of the processor is high is solved.

The invention provides a task scheduling method and device for a completely fair scheduler and electronic equipment, and relates to the technical field of artificial intelligence and cloud computing, in particular to an operating system scheduling technology. According to the implementation scheme, the waiting duration of a ready task in a running queue of a processor in the running queue is determined; when the ready task leaves the running queue, delay information of the processor is updated according to waiting duration of the task; comparing the maximum waiting time length of the ready task in the running queue with a delay threshold value, wherein the delay threshold value is based on delay information of the processor; and in response to the fact that the maximum waiting duration exceeds the delay threshold value, scheduling and executing a ready task corresponding to the maximum waiting duration. By utilizing the embodiment of the invention, the delay threshold value for preferentially scheduling the task is dynamically determined by utilizing the delay condition of the processor, so that dynamic scheduling based on the running condition of the system is realized on the basis of CFS.

The invention provides a task migration method and a related device, which are applied to electronic equipment, the electronic equipment comprises one or more processors, and when the electronic equipment monitors that a task with the highest priority in a running queue on a second processor is executed, the task is migrated. And migrating the task with the highest priority in the waiting queue on the first processor to the second processor for execution. Wherein the first processor and the second processor both have tasks with the highest priority, the tasks with the highest priority on the first processor wait in a queue, and the tasks with the highest priority on the second processor are being executed. When the task with the highest priority in the operation queue on the second processor is executed, it can be indicated that the task with the highest priority does not exist on the second processor at the moment, balancing is carried out at the moment, and it can be guaranteed that the task with the highest priority migrated to the second processor is executed in time. Compared with the existing load balancing, the migration timeliness and success rate can be improved.

Techniques for backup involve determining parameters of replication sessions in a waiting queue in response to a backup execution command being triggered; and such techniques further involve determining weights of the replication sessions based on the parameters of the replication sessions. Such techniques further involve selecting at least one replication session from the replication sessions to enter a running queue based on the weights of the replication sessions. Such techniques further involve performing a backup task with respect to the at least one replication session in the running queue. Accordingly, each replication session is ranked based on priority, and parameters are taken into account in the ranking to improve backup efficiency, thereby reducing the number of waiting and failed tasks and better protecting data, which enhances the replication capability of the system and increases the replication limit number to prevent the system from running out of memory.

Provided are a resource scheduling method and apparatus, where the resource scheduling method is applied to a physical machine, including: a kernel scheduler of the physical machine determining, based on a preset resource scheduling cycle, a quantity of virtual CPUs of a virtual machine running in a run queue of a target physical CPU; determining, based on the quantity of the virtual CPUs, a switch state of a non-exit function of a virtual CPU, and determining, based on the switch state, a target scheduling strategy of the virtual CPU; regulating, based on the target scheduling strategy, a running state of the virtual CPU of the virtual machine, thereby achieving timely shutdown of the non-exit function(s) of the virtual CPU(s), ensuring that unnecessary physical CPU resources are not consumed, as well as enabling the non-exit function(s) of the virtual CPU(s) to improve application performance of the virtual CPU(s).

The present disclosure provides a task scheduling method and device, equipment and a storage medium, relates to the technical field of computers, in particular to the technical fields of cloud computing, data processing, task scheduling and the like. The specific implementation scheme is as follows: obtaining at least two tasks in a running queue; scheduling the at least two tasks in a central processing unit of cloud computing resources based on the task types of the tasks; wherein the task types include a virtual machine thread type, a system thread type and a best effort thread type; wherein the priority of the virtual machine thread type is higher than that of the system thread type, and the priority of the system thread type is higher than that of the best effort thread type. The present disclosure can improve the execution stability of the virtual machine thread type task.