188 results about "Resource contention" patented technology

The invention provides a multi-agent cooperation method, system and device and a storage medium, and relates to the technical field of multi-agent collaboration.The method comprises the steps that initial role allocation is conducted on multiple agents, one agent is an observer, the other agent is a coordinator, and the other agents are all executors; a strategy network based on deep reinforcement learning is introduced according to the running state of the multiple agents to dynamically adjust role allocation of the multiple agents, and a role allocation strategy is dynamically adjusted according to task completion rewards, role conflict punishment and resource conflict rewards; the coordinator constructs a task priority and a dependency relationship based on the task graph or the task dependency tree, and dynamically allocates tasks according to the state, the capability vector and the task adaptation degree of the executor; conflicts are recognized through resource contention detection, task overlapping detection and behavior conflict detection, and the conflicts are coordinated. According to the invention, multi-agent responsibilities are layered, and the task completion efficiency is improved through task allocation and conflict detection and coordination.

The invention discloses a data processing system, method and device, a medium and a program product in the technical field of computers. In the application, the plurality of acceleration cards are directly connected with the plurality of memory devices through the management device, and the actual operation information of each acceleration card is collected through the management device, so that the memory space is allocated to each acceleration card as required according to the actual operation information of the acceleration cards. In other words, each memory device is not perceived by the host and is directly used as the extended memory of each acceleration card, so that the host does not need to distribute the memory for each acceleration card, thereby avoiding the condition of scrambling resources such as bandwidth easily occurring when multiple acceleration cards apply for the memory at the same time, and being not easy to cause data transmission delay. The purpose of expanding the memory for the accelerator card on the premise of not influencing the system operation performance is achieved.

The invention relates to a traffic anomaly detection and network security reinforcement method for a full data center, and the method comprises the steps: achieving the precise recognition and active blocking of an attack chain through the real-time collection of full network traffic, the construction of a multi-dimensional time sequence model based on historical attack chain features, and the combination of a dynamic judgment threshold value and spatial and temporal distribution feature analysis. The method specifically comprises the steps of analyzing traffic in real time to generate an attack chain feature sequence; performing clustering and correlation analysis based on the multi-dimensional time sequence model; dynamically adjusting the monitoring density to focus the high-risk area; and generating a threat blocking instruction and executing network isolation. Aiming at the problems of high attack chain omission ratio, resource contention conflict and insufficient dynamic defense capability caused by dependence on a static rule and a fixed threshold value in the traditional scheme, the method solves the problem of weak perception capability of the traditional technology on a complex attack chain, remarkably reduces the omission rate and the error blocking rate, guarantees the service continuity through dynamic resource scheduling, and improves the service performance of the system. The method is suitable for real-time security protection of a large-scale data center.

The invention discloses a cloud side-end cooperative task scheduling and efficiency optimization method and system for heterogeneous patrol resources, and relates to the technical field of intelligent scheduling and resource optimization. According to the method, accurate perception of a resource state is realized by constructing a digital twinborn and federated learning mechanism, resource contention conflicts are solved by adopting a space-time diagram attention network and multi-agent reinforcement learning, and multi-target optimization and trusted execution are realized in combination with a quantum genetic algorithm and a block chain smart contract. Finally, the stability of the system is verified through Lyapunov optimization, a complete scheduling system from resource perception and conflict resolution to steady state maintenance is formed, and the task scheduling efficiency and the system stability in the heterogeneous resource environment are remarkably improved.

The invention relates to the technical field of automatic control, in particular to an intelligent equipment management method and system, and the method comprises the following steps: obtaining a control cabinet log, extracting an equipment temperature load pressure sequence, comparing a set offset to generate a deviation label, collecting an operation time length and load change mapping early warning table, and generating an abnormal record; and extracting task configuration and equipment state re-marking alternating points to generate load configuration, resetting an instruction sequence to generate a scheduling sequence, extracting an instruction time-consuming marking interference section, and switching a standby link to generate an evasion list. According to the invention, by collecting the temperature, the load rate and the pressure sequence in the operation cycle of the equipment, the dynamic state identification capability is improved, the scheduling path is optimized, the resource contention is relieved, the delay path is marked in combination with the original response data, and the standby link is switched to avoid interference, so that the flexibility and stability of linkage scheduling are improved, and the task execution beat accuracy is enhanced; the instruction blocking and interruption risks are effectively reduced, and the operation toughness and response efficiency of a production line are improved.

Provided in the present disclosure are an RDMA-based data transmission method, and a device, a storage medium and a program product. During an RDMA connection establishment process, a local work queue used for same-host RDMA communication and a remote work queue used for cross-host RDMA communication are created at the same time, and a same-host recognition function is added. When same-host communication is recognized, the local work queue is used to directly execute a data copy operation in registered memories of two applications to replace a traditional RDMA transmission operation. Therefore, the compatibility with an RDMA programming interface and transparency or unawareness with regard to applications are achieved, the transmission delay between different RDMA applications on the same host can also be reduced, thereby reducing the consumption of a PCIe bandwidth, and ensuring the fairness in resource contention between local traffic and Internet traffic.

The invention discloses a cross-cloud task arrangement optimization method based on multi-cloud resource scheduling, and the method comprises the steps: constructing a task topological weighted graph, and recognizing a parallelizable node set; constructing a task node preliminary scheduling feasibility table according to task topological weighted hierarchy information; constructing a granularity feature vector, and updating the task structure matrix; identifying granularity collaborative attenuation abnormal nodes, and performing cross-cloud task path repair and structure update; generating a cross-cloud scheduling time matrix based on each cloud platform resource supply capability index; predicting resource competition conflicts in combination with task execution index historical data, and generating a conflict prediction table; identifying granularity strategy deviation, and determining a repair trigger point; and generating a cross-cloud task granularity adaptive arrangement configuration file based on the repair trigger point information. According to the method, high reliability, high flexibility and high adaptability of task scheduling can be realized in a complex multi-cloud heterogeneous environment, and the stability, the resource utilization rate and the overall cooperation efficiency of cross-cloud task execution are remarkably improved.

The invention provides a containerized resource elastic scheduling method and system in a cloud computing environment, and relates to the technical field of containerized resource elastic scheduling, and the method comprises the steps: obtaining various resource contention events, and generating data which records a resource contention relationship between tenants in detail; identifying a preset high-frequency conflict mode by deeply analyzing the data, and determining a specific tenant group with a strong resource mutual exclusion effect; dynamically estimating a probability level of resource conflict between the new container and the specific tenant group in combination with historical contention information and a real-time resource load state; generating a set of dynamic isolation scheduling strategies according to the conflict risk indicated by the probability level; an independent conflict buffer resource pool is configured and adjusted for a specified key tenant, a tenant resource conflict mode can be identified, dynamic isolation scheduling and buffer resource configuration can be implemented, and the resource utilization efficiency and the system stability of a multi-tenant container in a cloud computing environment are improved.

The invention discloses a Presto task scheduling method and a Presto task scheduling device. According to the method, after a task request is received, tasks are allocated to matched clusters according to user information and task information in the request, then task types are further analyzed, execution modes of the tasks are determined, whether the tasks belong to task types needing to exclusively share resources or not is judged, and if the tasks belong to resource exclusive tasks, the tasks are allocated to the clusters. Tasks are allocated to reserved single tasks and special resource nodes, and when the tasks exclusively share working node resources, no other types of tasks are executed on the nodes, so that the tasks exclusively share the resources on specified cluster nodes, physical isolation of the working nodes in a cluster is realized, the performance problem caused by resource contention is avoided, and the service life of the cluster is prolonged. Therefore, the method is suitable for a complex query environment shared by multiple users, efficient execution of the important tasks can be ensured even when a large number of tasks are processed, and the overall reliability and stability of the system are improved.

The invention discloses a hybrid multiple access method which comprises the following steps: constructing an unmanned aerial vehicle cluster networking system which comprises an unmanned aerial vehicle cluster network; an unmanned aerial vehicle sending node in an unmanned aerial vehicle cluster network divides channel resource contention and a data packet receiving and transmitting process according to a unified clock, and channel time domain resources are divided into a control channel time interval (CCHI) stage and a service channel time interval (SCHI) stage through time division multiplexing; wherein in the CCHI stage, service channel resource negotiation is performed according to a preset MAC protocol frame structure, time-frequency multiplexing access is performed on a shared service channel, and preliminary hybrid multiple access is completed; the SCHI stage is used for sending a data packet; the unmanned aerial vehicle sending node increases the length of an avoidance window of the unmanned aerial vehicle sending node in the next round according to a preset backoff process in the service channel access process on the premise that a to-be-sent data packet does not suddenly change if the unmanned aerial vehicle sending node encounters the collision conflict of the data packet on the premise that the to-be-sent data packet does not suddenly change; and if the sending is smooth, reducing the length of the receding window.

The invention relates to the technical field of computing resource scheduling, in particular to a GPU server resource dynamic allocation management method, device and equipment and a medium, the GPU server resource dynamic allocation management method comprises the steps of extracting a jump path to divide task migration segments, matching path repeated tasks in a thread, screening overlapped tasks in combination with scheduling and residence time, and performing dynamic allocation management on GPU server resources. And analyzing a resource access trend to identify scrambling stable combinations, and associating address frequencies to generate a dynamic allocation list. According to the method, the behavior similarity among multiple tasks is identified by extracting the jump path among the tasks and performing repetition analysis, the dense overlapping interval of task execution is accurately delimited by means of cross mapping of task scheduling time and path residence time, and the task performance of the multiple tasks is determined in combination with the derivative judgment mode of the change trend of the access frequency of the shared resources. And identifying a task combination with stable resource scrambling behavior distribution, and establishing a dynamic binding strategy between the task and the GPU address resource based on the joint sorting of the address access frequency and the residence duration.

The invention discloses a management system for an integrated circuit test, and relates to the technical field of integrated circuit test management, and the system comprises a task scheduling module which is used for receiving a plurality of test task requests, and generating a test task scheduling sequence based on the priority parameter of each test task, the current state of a test resource, and a historical execution result; and the resource coordination module is used for acquiring the running states of the plurality of test devices, executing resource allocation operation according to the test task scheduling sequence, and generating a coordination abnormal event if resource conflict or scheduling failure is detected in the execution process. According to the invention, a dependency graph between test equipment and bottom hardware resources is constructed through the resource coordination module, and accurate modeling of a resource mutual exclusion relationship, concurrent capacity and conflict risk is supported, so that resource conflicts are pre-judged before scheduling, and the problems of resource scrambling and deadlock are effectively avoided.

The invention relates to the technical field of electric digital data processing, and discloses a data processing system for message management, which comprises a look-ahead buffer unit for maintaining a sliding window with variable window size parameters and pre-reading a message sequence; the mapping logic circuit is used for converting the resource identifier into a resource request mask; the state register is used for storing a global occupancy bitmap for representing resource occupancy; the scheduling controller is used for distributing messages according to bit operation results, counting bitmap setting quantity based on the population counting instruction to generate a resource saturation index, and then adjusting window size parameters in a negative feedback mode. The resource contention entropy value is directly sensed by utilizing a hardware instruction, nanosecond-level self-adaptive breathing control of a scheduling window is realized, and the service life of the scheduling window is prolonged. The problem of physical computing power collapse caused by bitmap scanning jitter in a high-concurrency scene is solved.

The invention provides a complex scene-oriented multi-modal AI Agent integrated development method and system, and the method comprises the steps: generating a layout graph G in an integrated development environment, and carrying out the consistency verification to obtain a constraint set C; a graph neural network is adopted to represent the arrangement graph to obtain graph embedding Z, and an environment state s is formed by combining key performance indexes in a sliding window; and in an action space AC limited by the C, jointly deciding computing power node selection, concurrent resource quota adjustment and semantic caching or semantic routing parameter configuration according to a reinforcement learning strategy pi, and executing actions to optimize throughput, tail delay, failure rate and resource contention degree. Through a closed loop of verification-constraint-learning-execution-monitoring-updating, cross-modal consistency and computing power scheduling are integrally optimized, training convergence and online stability are improved, long-tail time delay and the failure rate are remarkably reduced, the resource utilization rate is improved, and the method is suitable for large-scale popularization and application. The method is suitable for the fields of medical diagnosis, industrial quality inspection and equipment maintenance, retail recommendation and shopping guide and the like.

The invention discloses a large model mixed load-oriented self-adaptive low-delay reasoning configuration generation method and device, computer equipment and a storage medium, and the method comprises the steps: determining a first token generation delay and an adjacent token delay interval which are historically configured on requests with different reasoning configurations, and obtaining tuples to form a configuration performance database, generating a delay prediction model by combining a least square method with the configuration performance database; dynamically dividing the historical request into a plurality of buckets according to the input length and the output length through a self-adaptive bucket dividing strategy; a configuration generator generates reasoning configuration for each bucket according to the input length and the output length of the historical request of each bucket; under the real mixed load, the problems of remarkable resource contention, queue head blockage, KV Cache switching overhead increase and the like are avoided in concurrent execution of long and short requests, and meanwhile, the problem of tail delay amplification is avoided, so that a reasoning system gives consideration to low delay and high throughput among different requests.

The invention provides a real-time monitoring and root cause positioning method and system for software defects. The method comprises the following steps: collecting real-time monitoring data of a software processing thread, and generating a thread competition thermodynamic diagram for identifying a shared resource contention point through time sequence association processing; triggering a server memory bus embedded hardware detection device based on a contention point to dynamically capture a locking state transition signal, and converting the locking state transition signal into a locking event sequence; constructing a locking contention track diagram according to timestamp sorting, and completely displaying a process path of a locking mechanism from obtaining to releasing; constructing a dependency topological structure of the thread and the lock based on the trajectory graph, and identifying a resource contention chain; and mapping the contention chain into an unsynchronized shared variable access path in the source code, and triggering a real-time alarm signal. According to the method, the access path of the unsynchronized shared variable is accurately positioned, the lock competition diagnosis efficiency is improved, and the system deadlock risk is zeroed.

The invention relates to the field of unmanned aerial vehicle control, in particular to a multi-mode link switching method for a heterogeneous unmanned aerial vehicle cluster and related products, and the method comprises the steps: obtaining the current task mode information of the heterogeneous unmanned aerial vehicle cluster and the link state information of each node; dynamically selecting and activating a corresponding link switching strategy from a preset strategy set according to the task mode information; according to different task modes, different link switching strategy sets are dynamically selected, under the constraint condition that communication resources are strictly limited, the resource contention contradiction in the heterogeneous unmanned aerial vehicle cluster is solved, and the communication toughness and efficiency of a cluster system are improved.

The invention relates to the field of chemiluminescence immunity analyzers, and discloses a high-speed full-automatic chemiluminescence immunity analyzer multi-module parallel motion control method, which comprises the following steps: in a current execution period, pre-calculating all secondary function module action instruction sets of a next period based on real-time position information of a reaction cup and test sample parameters; according to a pre-generated instruction set, coordinating parallel actions of multiple modules through a preposition and subsequent dependency relationship, and driving a hardware execution unit to complete sample processing; and synchronizing the reaction cup position data and the test step state in real time, and switching to a next period instruction set when the period is ended. According to the method, the problems of time sequence conflict and resource contention of multi-module cooperative control are solved, the parallel efficiency of the detection process and the fault-tolerant capability of the system are improved, and the method is suitable for a high-throughput chemiluminescence immunoassay scene.

The invention discloses a computing node power consumption monitoring and predicting method, and belongs to the technical field of computers. The problem that an existing power consumption prediction method is poor in accuracy is solved. According to the method, the dynamic operation states of the nodes, the hardware static configuration attributes and the coupling influence of the communication relation and power consumption between the nodes are comprehensively considered, a graph structure based on node communication is constructed, and a power consumption coupling path between the nodes is modeled and calculated, so that the dynamic energy consumption of local nodes can be captured; and cross-node resource contention, communication bottleneck and energy consumption conduction effect can be considered. Through a structure-guided graph attention mechanism, the model can realize information sharing and feature collaboration among a plurality of nodes. By introducing an energy efficiency attention fusion module, dynamic weighted modeling is performed on multi-channel input features, key factors having high influence on energy consumption fluctuation are automatically identified, and the accuracy and interpretability of the model in the aspects of power consumption prediction and optimization suggestion output are improved. The method can be applied to power consumption prediction of the computing node.

The invention discloses an advertisement project collaborative management method based on intelligent allocation, and particularly relates to the field of advertisements, which comprises the steps of system initialization, multi-dimensional basic data acquisition, resource dispersion and time pressure analysis, collaborative conflict coefficient synthesis analysis and dynamic priority allocation decision. According to the method, the resource utilization rate is remarkably improved, the resource tension type is accurately identified and the allocation strategy is dynamically optimized by calculating the standard deviation dispersion index of the resource occupancy rate in real time, and the phenomena of resource idleness and overload are eliminated; an exponential function time pressure model is innovatively constructed, dual dimensions of task residual workload and deadline are fused, nonlinear priority evaluation of emergency tasks is realized, and priority response of high-load short-period tasks is ensured; a cooperation conflict coefficient mechanism is originally created, a resource contention factor and a time window factor are integrated, a high-conflict task combination is actively early warned, and the risk of cooperation blocking is effectively reduced.

The invention provides a task scheduling method and device for a civil aviation industry-level data service platform, equipment and a medium, and relates to the technical field of task scheduling, and the method comprises the steps: determining the task type of each task in a to-be-scheduled task set; if it is determined that the resource demand of any calculation-intensive analysis task exceeds a preset resource threshold value and the calculation-intensive analysis task is decomposable, decomposing the calculation-intensive analysis task into a plurality of sub-tasks; constructing a to-be-scheduled queue comprising all routine data processing tasks, all sub-tasks and all undecomposed compute-intensive analysis tasks, and recording attributes of each task in the to-be-scheduled queue; based on the attribute of each task and the real-time available resource state and topological information of the computational nodes in the civil aviation industry-level data service platform, each task is dynamically allocated to a plurality of computational nodes, so that the problems of resource scrambling, routine data processing task performance damage, low computational intensive analysis task efficiency and the like are solved.

A GEMM load-oriented GPU modeling method is characterized in that through a multi-stage collaborative modeling mechanism, cache behaviors, instruction overhead and calculation intensity are deeply coupled, accurate performance prediction of GPU execution GEMM operators is realized, the method can be widely applied to scheduling optimization of GPU intensive scenes such as AI training and scientific calculation, firstly, a three-stage cache weight distribution mechanism is established, and then, a three-stage cache weight distribution mechanism is established; quantifying the contribution of the L1 / L2 cache hit rate and the DRAM bandwidth degradation factor to the effective bandwidth; secondly, an instruction-level memory access overhead correction mechanism is introduced, and the mixing precision and the real calculation strength of a sparse calculation scene are captured through dynamic parameter adjustment and optimization; then combining the calculation force peak value and the bandwidth upper limit to construct a double-boundary constraint model, and generating a theoretical performance critical value; further predicting a stream multiprocessor utilization rate based on a neural network, and quantifying efficiency loss caused by hardware resource contention through a multi-layer perceptron structure; and finally, the integration module outputs task execution time to realize end-to-end performance prediction.

The invention relates to a high code coverage rate fuzzy test method based on a parallel strategy. The method comprises the following steps: deploying a plurality of fuzzy test examples running in parallel; acquiring an execution path and state data of the target program in real time through a dynamic instrumentation technology; dynamically evaluating the seed priority based on the central management system; distributing a high-priority seed to each fuzzy test instance through a feedback driving mechanism, and optimizing file operation and multi-thread synchronization; and dynamically adjusting the seed priority and the test strategy according to the real-time test data. The method has the advantages that system calling and resource contention are effectively reduced, the test efficiency and coverage rate are improved, and efficient cooperation and intelligent test strategy adjustment are achieved through the design of a multi-thread synchronization mechanism and a central management system.

The invention relates to the technical field of data processing, in particular to a material purchasing demand response system based on intelligent decision making, and the system comprises a data acquisition unit which collects environment physical data and purchasing business data, and outputs a cross-dimensional original data flow; the data processing unit fuses the heterogeneous data to generate multi-modal feature data; the prediction unit calculates a blocking risk probability and an optimal path, and generates a path optimization instruction; the reconstruction control unit allocates a high-blocking task to a low-delay optical channel, allocates a conventional task to a high-fault-tolerance channel, and outputs a task and path mapping table; the scheduling unit decomposes the mapping table into a neuromorphic scheduling instruction and a quantum optimization instruction and fuses the neuromorphic scheduling instruction and the quantum optimization instruction to generate a cross-scale scheduling command; and when resource contention is detected, rescheduling and acquisition strategy adjustment are triggered. Nanosecond instruction writing is realized through a photon and quantum direct connection architecture, and the problems of resource contention and delay accumulation of an electronic digital data processing system in a high-concurrency scene are solved in combination with a dynamic entropy reduction regulation and control and hardware resource isolation mechanism.

The invention discloses a power distribution automation terminal task scheduling method and device based on a power distribution network edge computing node, electronic equipment and a storage medium, and belongs to the field of power distribution network automation, and the method comprises the steps: obtaining the total amount of special and public containers of an edge node and the decomposition configuration and operation parameters of a target power distribution automation task; the method comprises the following steps of: decomposing a task into micro-applications, constructing a calling graph, counting the repetition degree of calling edges, combining the edges with high repetition degree into an undetachable micro-application chain, and calculating and distributing exclusive containers according to the repetition degree and the total number of the exclusive containers; and calculating weights according to task calling times, priorities and concurrency numbers, and calculating and distributing the public containers based on the weights and the total number of the public containers. Therefore, by implementing the method and the device, the problems that high-frequency fixed service flow resources are severely scrambled and the utilization rate of discrete task resources is low in the prior art can be solved.

The invention discloses a multi-disciplinary simulation task intelligent planning method based on a knowledge graph, and relates to the technical field of intelligent scheduling, and the method comprises the steps: constructing a preliminary knowledge graph comprising multi-disciplinary task description, resource limitation and execution dependence, analyzing historical task data through a pre-trained deep learning model, and obtaining a multi-disciplinary simulation task planning model; dynamically adjusting the preliminary knowledge graph through adaptive learning, and outputting an adaptive knowledge graph; carrying out resource redistribution optimization according to the updated task scheduling strategy, continuing to execute the task, starting monitoring, and collecting task progress and resource data as a new feedback data stream; the resource reallocation and the new feedback data stream are fed back to the adaptive knowledge graph, and multidisciplinary task description, resource limitation and execution dependency are updated through reasoning to generate a final optimization task execution sequence; the synchronous suppression of waiting time and resource contention is realized, the average utilization rate of resources and the time sequence stability are improved, and the adaptive adjustment capability of operation disturbance is formed.

The application discloses a kind of based on flow perception intelligent network interface card optimization system and method, it belongs to computer network technical field.The optimization system of the application includes flow perception module, flow prediction module, resource scheduling module, priority queue module and load balancing module.The application introduces resource monitoring, flow perception, resource prediction and dynamic scheduling module, real-time monitoring network flow and predicting its change trend, according to flow mode dynamically adjusts resource allocation.The application effectively solves the problem of resource contention, through priority ordering and load balancing, ensure the stable operation of high priority task, improve the throughput and stability of system, reduce resource waste.The application is suitable for resource management of intelligent network interface card in data center and cloud computing environment, improves network processing efficiency and resource utilization, guarantees service quality, reduces resource waste, and enhances the adaptive capacity of system to dynamic load and complex flow mode.

The invention provides a method for realizing an intelligent non-player character in an interactive game, which relates to the field of data processing and comprises the following steps of: packaging a decision intention into an immutable action package in a predefined decision period and placing the action package in a time domain trusteeship queue to realize time decoupling of a decision and a runtime load; the world state projection is divided into an opaque decision phase and a reconciliation phase, the decision phase only exposes the event category projection and the historical abstract, and the reconciliation phase reflows detailed trigger details in an independent window for posteriori maintenance but does not backtrack submitted actions; submission follows an atomization criterion and the submitted abstract does not contain time information; during resource contention, accumulating in response to a debt, and repaying through display shaping on the premise of not exposing a time sequence so as to keep the semantic consistency of main games; the posterior processing updates the action banks and event mappings offline based on the backflow details to improve subsequent decisions.

This disclosure provides embodiments of a method, apparatus, and electronic device for dispatching supplies for grassroots governance matters. One specific implementation of the method includes: in response to detecting a supply dispatch request for a grassroots governance matter characterizing a public emergency, acquiring corresponding matter information; performing information standardization processing on the matter information to obtain standard information; extracting target elements from the standard information to obtain a target element information set; performing resource contention conflict detection on the target element information set to obtain a resource contention index set; in response to the resource contention index set reaching a preset threshold, dynamically prioritizing the target element information set based on the resource contention index set to obtain a priority index set; generating a resource allocation instruction based on a hardware rule engine with pre-set resource arbitration rules and the priority index set; and performing supply dispatch operations according to the resource allocation instruction. This implementation achieves the technical effect of reducing the response delay of supply dispatch operations.

Systems, apparatuses, and methods are disclosed for arbitrating threads in a computing system. The computing system includes a processor having a plurality of cores that are each capable of concurrently processing instructions of a plurality of threads. When a thread throttling unit receives an indication that a shared cache has a resource contention, the throttling unit sets a cache miss threshold for the cache. If the number of cache misses exceeds the threshold, the throttling unit notifies a particular upstream computing unit to throttle the processing of instructions of the thread. After a time period elapses, if the cache continues to exceed the threshold, the throttling unit notifies the upstream computing unit to more restrictively throttle the thread by performing one or more of decreasing a selection rate and increasing the time period. Otherwise, the unit notifies the upstream computing unit to less restrictively throttle the thread.