16 results about "Scheduling complexity" patented technology

The invention provides a computing power resource partitioning method and device based on hypergraph clustering, equipment and a medium. The method comprises the following steps: constructing a computing power resource hypergraph in a computing power network resource side scene; determining an equipment degree matrix and a hyperedge degree matrix of the computing power resource hypergraph, and converting the computing power resource hypergraph into an equipment Laplacian matrix required by hypergraph clustering by using the equipment degree matrix and the hyperedge degree matrix; carrying out eigendecomposition on the equipment Laplacian matrix to obtain an equipment eigenvector; and in combination with the graph cutting target function, performing hypergraph clustering processing on the computing power resource hypergraph based on the feature vector of the device Laplacian matrix to obtain a target resource partitioning result corresponding to the computing power resource hypergraph. According to the method, the problems of low resource utilization rate, high scheduling complexity, poor resource dynamics, lack of effective partitioning strategies and the like existing in the existing computing power resource partitioning can be obviously improved.

The invention discloses a task hierarchical scheduling method for an intelligent computing fusion network, belongs to the technical field of the intelligent computing fusion network, and designs a double-layer task scheduling mechanism (ICCN-DSF) for the intelligent computing fusion network to effectively cope with task scheduling challenges in a large-scale heterogeneous resource environment. Wherein the global task scheduling layer is based on a CIEK-Means algorithm, integrates geographic position constraint and group intelligent optimization, and realizes efficient clustering and dynamic matching of resource ethnic groups; and the local task scheduling layer constructs a three-dimensional state action space and a multi-target reward function through a reinforcement learning algorithm to complete fine-grained resource allocation. According to the method, the scheduling complexity is remarkably reduced, meanwhile, the cross-domain connection requirement is reduced through a layering mechanism, global resource coordination can be achieved only through a small number of WAN links, the operation and maintenance cost and the safety risk are reduced, meanwhile, the cooperation potential of the computing power and the network is fully excavated, and efficient and robust scheduling support is provided for intelligent services in complex scenes.

The invention relates to the technical field of computer science and industrial engineering, in particular to a tool workshop scheduling method, which comprises the following steps of: acquiring production characteristics of a current tool workshop, generating an optimal batch scheduling strategy according to the production characteristics of the current tool workshop based on a preset evolutionary strategy, and automatically designing an algorithm by utilizing a preset large model, and calculating the minimum completion time and the maximum completion time of the optimal batch scheduling strategy, and completing tool workshop scheduling according to the minimum completion time and the maximum completion time based on the optimal batch scheduling strategy. Therefore, the problems of high workshop scheduling complexity and the like caused by strong coupling of batch division and working procedure sorting, strict constraint between working procedures and high resource networking degree in workshop scheduling are solved, an automatic algorithm design framework based on a large model is introduced, the expert algorithm design time is shortened, and the workshop scheduling efficiency is improved. Therefore, the efficiency and feasibility of aircraft tooling production scheduling are improved.

The invention discloses a large model cross-domain training scheduling method, system, device and equipment, a medium and a product. The method is applied to a scheduling system comprising a scheduling layer and a management and control layer, and comprises the following steps: obtaining job requirements and resource states of large-model cross-domain training jobs through the scheduling layer; the operation requirements comprise a computing resource requirement, a storage resource requirement and a network resource requirement; determining a corresponding target scheduling scheme through the scheduling layer according to the job demand and the resource state; and based on the target scheduling scheme, controlling the management and control layer to construct a data transmission path and a task communication path through the scheduling layer, and starting large-model cross-domain training based on the data transmission path and the task communication path. According to the scheme, through cooperation of the scheduling layer and the management and control layer, while the requirements of large-model cross-domain training operation for calculation, storage and network resources are met, a high-cost special network is replaced by constructing a data transmission path and a task communication path, and the training cost and the scheduling complexity are effectively reduced.

The application provides a TSN gate scheduling table generation method for frame merging pre-processing, and mainly solves the problems of low real-time data transmission and low network resource utilization efficiency caused by the inability to effectively manage and schedule a large number of data streams in the prior art. The method of the application adopts a scheme of pre-processing the flow, and according to the pre-processing result, the flows with the same characteristics are merged, so as to improve the real-time transmission and provide the utilization efficiency of network resources. The method adopted by the application reduces the number of flows in the network through the merging processing, reduces the scheduling complexity, combines the advantages of the determinacy of SMT solving and improves the solving efficiency on this basis, and is suitable for highly time-sensitive industrial automation and other real-time data processing applications.

This application belongs to the field of computer technology and provides a task execution method, compilation method, chip, and device for an artificial intelligence model. The application first requires compiling the tasks to be processed by the artificial intelligence model according to a preset compilation method, resulting in multiple sub-tasks. This allows each task to be processed to receive loading and scheduling from the NPU at a moderately granular scheduling unit. Specifically, this application uses the sub-tasks as the basic scheduling unit of the NPU, thereby achieving resource scheduling at a moderate granularity that balances scheduling complexity and task switching flexibility. When encountering urgent tasks, the task switching cost of the NPU is significantly reduced; thus, in the event of sudden changes in the target application running on the computer device, the response efficiency and resource utilization of the NPU can be effectively improved.

The invention provides a distributed large model arrangement-oriented multi-agent cooperation platform and a method thereof. The platform comprises a forward task execution circulation system, a multi-agent layer, a task matching module, an evaluation alternative module and a backward model learning circulation system, the forward task execution circulation system is composed of a distributed environment reasoning unit, a task stage sequence unit and an edge server; the backward model learning circulation system is composed of an experience playback buffer area, a self-adaptive cooperation module and a synchronous coordination module; the multi-agent layer is composed of a decision network and an evaluation network, the problems of unbalanced resource allocation and complex scheduling caused by dispersed user requirements and various task stages in distributed edge computing are solved, and intelligent arrangement and adaptive decision are implemented in the task stage granularity, so that the distributed edge computing efficiency is improved. The cooperative relationship between server selection and model selection is optimized, the overall response speed of the system is improved, and the end-to-end reasoning time delay is reduced, so that the high-standard requirement for service quality is met.

The application discloses a network flow scheduling method and a storage medium, and belongs to the technical field of communication. The method comprises the following steps: constructing a current flow relationship graph corresponding to to-be-scheduled flows of a target network according to flow relationships among the to-be-scheduled flows; regarding each node in the current flow relationship graph as a community respectively, and determining a current modularity of the current flow relationship graph; performing grouping processing on the current flow relationship graph according to the current modularity, determining a key flow set and a plurality of grouped flow sets; preferentially scheduling the key flow set based on a preset model, and after the scheduling of the key flow set is completed, scheduling the plurality of grouped flow sets in parallel based on the preset model, and obtaining a target scheduling result. The technical scheme provided by the application can reduce scheduling conflicts among flows, reduce scheduling complexity, and significantly improve scheduling efficiency.

The application relates to the technical field of computing power scheduling, in particular to a heterogeneous computing power unified scheduling method and system based on global planning. A scheduling model is created, computing power demand information is collected, pretreatment information and the computing power demand information are input into the scheduling model, the computing power of each target device is calculated through the scheduling model, the computing power levels of the target devices are divided based on the computing power, then the coupling of the computing power demand is completed in combination with the computing power levels, the standard device of each computing power demand is obtained, finally, real-time information is input into the trained scheduling model, the real-time demand is distributed to the corresponding target device through the trained scheduling model, and the demand distribution of the real-time demand is completed. The computing power of the target device is calculated, the computing power levels of the target devices are divided based on the computing power, the computing power level division and the demand-device coupling are realized, the fuzzy computing power demand is converted into definite device matching rules, and the scheduling complexity is reduced.

Stacked-die memory components are programmably configurable to emulate memory architectures having fewer logical / virtual memory banks, bank groups and / or integrated circuit dies than the resident physical quantities of those resources, substantially reducing command scheduling complexity in counterpart host component and thereby enabling deeper scheduling queue implementation and correspondingly higher bus utilization efficiency.

The invention relates to the technical field of computing power scheduling, in particular to a heterogeneous computing power unified scheduling method and system based on global planning. The method comprises the steps of creating a scheduling model, collecting computing power demand information, inputting preprocessing information and the computing power demand information into the scheduling model, calculating the computing power of each target device through the scheduling model, dividing the computing power levels of the target devices based on the computing power, and then completing coupling of computing power demands in combination with the computing power levels. The method comprises the following steps: calculating the computing power of a target device to obtain a standard device of each computing power demand, finally inputting real-time information into a trained scheduling model, and distributing the real-time demand to a corresponding target device through the trained scheduling model to complete demand distribution of the real-time demand. And the computing power level of the target equipment is divided based on the computing power, so that computing power level division and demand-equipment coupling are realized, a fuzzy computing power demand is converted into a clear equipment matching rule, and the scheduling complexity is reduced.

The present invention discloses a cloud platform GPU virtual machine scheduling method and device, which relates to the field of computer technology. The method obtains the target number of GPU resources required for a new task. If the total number of GPU resources remaining in all channels is greater than the target number of GPU resources, and no single channel meets the target number of GPU resources, all channels are sorted from most to least by the number of remaining GPU resources of each channel, and then a preset number of channels are obtained to obtain an effective channel set. The resources in the effective channel set are dynamically migrated by a preset algorithm to obtain a migration result. If the migration result is to generate a target migration channel, the target number of GPU resources of the new task is all scheduled to the target migration channel. By dynamically migrating idle resources across channels, the fragmentation of single channel resources is avoided, and migration from channels with more remaining resources is prioritized, thereby improving the utilization rate of GPU resources, reducing scheduling complexity, and meeting the needs of high-performance computing tasks.

The invention belongs to the field of power system communication, and discloses a routing resource allocation method and system based on joint optimization of working and protection paths, and the method comprises the steps: obtaining a network state and a service request, and determining a source node and a destination node according to the network state and the service request, so that routing and spectrum allocation are established on the basis of real-time network resources; the introduction of extra scheduling complexity due to mismatching of network states is avoided; a plurality of disjoint paths are obtained between a source node and a destination node and are sorted according to path lengths to form a path set, so that subsequent routing is carried out in a controllable candidate range, and the combination complexity in a multi-working-path scene is reduced; by analyzing the path set and selecting a plurality of groups of working paths and protection paths corresponding to the working paths, unified modeling and joint consideration are performed on the multiple working paths and the protection paths, and spectrum conflicts and resource waste caused by working path and protection path separation decisions are avoided.

Embodiments of the present disclosure provide an event queue scheduling method, device, equipment and medium. The method comprises: in response to an event enqueue request, determining a target bucket matching a target event to be enqueued in an event queue; the event queue is composed of one active bucket, no more than a preset number of alternative buckets, and one default bucket; inserting the target event into the target bucket, and performing heap sorting on events in the target bucket; and in response to an event dequeue request, taking a heap top event from the active bucket of the event queue. Through the above technical solution, the embodiments of the present disclosure reduce the event scheduling complexity based on the total number of events, and improve the scheduling efficiency of the event queue.

Embodiments of the present application provide a scheduling method, device and equipment. The method comprises: obtaining a to-be-scheduled request of an object; putting the to-be-scheduled request into a corresponding position of a corresponding request flow in a queue, to serve as a to-be-scheduled request of the corresponding request flow, the request flow being obtained according to the object, and a weight of the request flow being a weight of the corresponding object; selecting a target request flow from the request flows with to-be-scheduled requests according to the weights of the request flows, selecting a target to-be-scheduled request of the target request flow from the queue, and taking the target to-be-scheduled request out of the queue for executing the target to-be-scheduled request. The present application not only considers the weight of the object (for example, a user) to which the to-be-scheduled request belongs when scheduling, to meet the scheduling requirement, but also has low scheduling complexity, and is a scheduling mode that can balance the weight fairness and the scheduling efficiency.

Embodiments of the present application disclose a heterogeneous GPU scheduling system, method, device and readable storage medium, the system comprising a heterogeneous GPU management layer, a GPU scheduling rendering layer and a GPU scheduler; the heterogeneous GPU management layer is integrated with device plugins corresponding to GPUs of different brands and different models, for automatically discovering and reporting heterogeneous GPU resources to a Kubernetes cluster; the GPU scheduling rendering layer is used for rendering configuration information of GPUs of different brands and different models into unified standard scheduling parameters; the GPU scheduler determines corresponding GPU configuration information through the GPU scheduling rendering layer based on the received GPU scheduling parameters, so that the Kubernetes performs intelligent scheduling of target GPUs based on the GPU configuration information. The present application solves the technical problems in the prior art that GPUs of different manufacturers, brands and models cannot be scheduled, and that the scheduling complexity is high due to the need to pay excessive attention to scheduling details during the scheduling process.