20 results about "Multilevel queue" patented technology

The invention provides a resource management method and system for a server operating system, and the method comprises the steps: carrying out the double-layer graph modeling of process and resource data in the server operating system, and obtaining a process competition graph layer and a resource dependence graph layer; utilizing a preset graph neural network to perform feature extraction on the node information of the process competition graph layer and the node information of the resource dependence graph layer to obtain a node feature vector containing a competition relation and a distribution state; according to the node feature vectors, deadlock probability calculation is carried out through a multi-layer perceptron, and a process priority adjustment scheme is obtained; according to the node feature vector and a process priority adjustment scheme, performing resource allocation strategy learning to obtain a first resource allocation scheme; according to the first resource allocation scheme, performing multi-level queue allocation on the process to obtain a second resource allocation scheme; and performing resource management on the server according to the second resource allocation scheme.

The present invention relates to a virtual machine page table tracking method and system based on PML. The method comprises: in a virtual machine, obtaining the page table pages used by the virtual machine load by scanning processes with a resident set greater than 1GB and transmitting them back to the hypervisor to obtain a monitoring target; periodically clearing the Dirty bit in the EPT of the virtual machine page table page and the Access bit and Dirty bit of the virtual machine page table entries contained therein, utilizing Intel PML technology to efficiently obtain the page table page entries where write operations have occurred, further scanning the virtual machine page table entries therein to obtain the pages where read and write operations have occurred within this cycle; and proposing a multi-level queue algorithm to accelerate tracking efficiency. The solution proposed by the present invention makes full use of the hardware characteristics of Intel PML to track the page tables used by the client, alleviating the overhead of the traditional page table scanning method that requires traversing all the page tables of the virtual machine; using the multi-level queue algorithm to screen the tracking pages, greatly reducing the number of pages that need to be set, and further reducing the overall tracking overhead.

The invention provides an application loading optimization method and device. The method comprises the following steps: randomly distributing a plurality of resource requests sent by an application page to a plurality of initial queues; determining the processing time of each resource request and the average processing time of all resource requests; allocating the resource requests to a plurality of waiting queues with different queue levels according to the processing time of the resource requests, the average processing time and the number of the resource requests in the initial queue; sequentially adding the resource requests in the waiting queue with the highest queue level into a ready queue for processing; and updating the waiting queues, and returning to execute the step of sequentially adding the resource requests in the waiting queue with the highest queue level into the ready queue for processing until all the resource requests are processed. According to the scheme, the multiple resource requests sent by the application page are scheduled through the multi-level queues, it is guaranteed that the resource requests of each queue can be processed in time, the resource loading time is shortened, and the page rendering speed and the user experience are improved.

The invention discloses a management system compatible with an advanced cache algorithm in a content distribution network CDN. The management system comprises an Inode table, an algorithm library, a multi-stage FIFO queue, a controller and a recovery pool, the Inode table is a hash table stored in a memory and is used for quickly associating key information of each data object; the algorithm library comprises a plurality of cache strategies based on lists, and the cache strategies are processed as priority queues; the multi-level FIFO queues construct a layered storage structure, and each FIFO queue is used for managing a storage device strip; the controller is based on an in-memory cache simulator, and the working process comprises a preprocessing stage and a real-time adjustment stage; and the recycling pool is used for storing and managing the virtual stripes and the physical stripes. According to the system, the problems of write amplification and significant overhead caused when a traditional cache algorithm is applied to persistent storage can be solved, the requirement of high hit rate can be considered, and the I / O overhead can be effectively reduced.

The invention provides a computing node memory management method and device, equipment and a program product, and relates to the technical field of computers. The method comprises the steps that a multi-stage queue system is constructed, the multi-stage queue system comprises queue units of at least two stages of storage media, and the queue units are arranged in a layered mode according to the performance of the storage media; calculating a capacity threshold value of each level of queue unit in the multi-level queue system based on an available memory of a calculation node and a preset proportionality coefficient; and dynamically adjusting the capacity of each level of queue unit according to the capacity threshold, and migrating the data to the low-priority queue unit when the high-priority queue unit is fully loaded. The method is used for achieving the effect of improving the memory resource utilization rate.

The invention discloses an integrated circuit global wiring method based on a multi-level queue, and the method comprises the steps: S10, a line edge capacity reduction stage: estimating wiring congestion through employing rectangular uniform lead density, and reducing the line edge capacity based on the number of pins; s20, in a multi-stage queue wiring stage, for a network which cannot be wired in a current wiring mode, the network is directly added to a next-stage queue to wait for wiring; when overflow does not exist or wiring of the last-stage queue is completed, wiring of the multiple stages of queues is completed; s30, based on rewiring of the congestion area, adopting a labyrinth wiring strategy; and S40, outputting a global wiring result based on the rewiring result. According to the method, capacity reduction is carried out on a pin congestion area, an initial wiring result is obtained by using multi-stage queue wiring, and overflow is reduced by using labyrinth wiring based on overflow distance.

The application relates to the field of industrial internet and intelligent data processing, and discloses a cross-domain low-latency transmission device data communication method, system, medium and equipment based on semantic perception, which comprises the following steps: collecting multi-source heterogeneous original data of a comprehensive transmission device to obtain original message data sets, and obtaining semantic entities through message analysis and semantic coding; compressing and multi-stage queue scheduling the semantic entities according to link states and service priority labels to obtain edge uplink compressed semantic streams; decompressing the edge uplink compressed semantic streams, and deserializing the decompressed binary streams into standard semantic data object sets; analyzing the attribute types of each data in the standard semantic data object sets, dynamically distributing the computing tasks corresponding to the attribute types to a heterogeneous computing cluster based on preset computing affinity rules, and outputting a result data stream; and a control engine dynamically generating display control instructions according to the result data stream, and an application domain adaptively rendering and visually displaying according to the display control instructions.

Embodiments herein describe including one or more error indicating bits in a task pointer (e.g., a WQE) to tell a downstream queue or stage, such as a pipeline, that an upstream queue or stage detected a DMA error. The communication from each upstream stage or pipeline to the next is typically through posting WQEs into the next stage queue and ringing doorbells. The embodiments herein use one or more error indicator bits (e.g., color bits) in the WQE to inform a downstream queue (e.g., a downstream pipeline) that the upstream queue / pipeline detected a failed DMA operation. The downstream queue can then perform error handling where it reclaims the resources allocated for the WQE (such as an intermediate buffer).

The invention relates to the technical field of computer data storage system performance evaluation, and discloses a cloud object storage persistent layer tail delay modeling method based on a queue model, which comprises the following steps: establishing a logic processing flow of a cloud object storage persistent layer, and defining a request transfer relationship to construct a multi-stage queue model; load modulation is executed based on the collected characteristic parameters, and a simulation request load containing sudden overload and hotspot inclination is generated; executing a fault disturbance mechanism in the multi-level queue model, establishing a server-level and queue-level double-layer fault system, generating a fault disturbance factor, and correcting the actual processing time of the request by using a maximum value strategy; and tracking the queuing waiting time of the request full link and the corrected service time, accumulating to obtain time delay distribution, and extracting a tail time delay index. According to the method, the problem of long tail phenomenon prediction distortion of a traditional model is effectively solved by simulating an unsteady load and a transient fault, and the accuracy of system performance evaluation under a complex working condition is improved.

The application relates to the technical field of medical informatization, and discloses a case consultation and pathological section automatic diagnosis system based on AI analysis, which comprises a server and a client; the server generates a lesion binary mask by using a deep learning model, constructs a sparse quadtree index carrying a composite state bit mask, and generates a vector section with the background removed; the client splits resource requests to a multistage queue according to the bit mask in the index, and preferentially loads high-value area images; a three-layer heterogeneous rendering architecture is adopted, and adaptive Gaussian blur is used to process the vector section to realize smooth transition; meanwhile, a behavior audit module collects viewport tracks based on a physical rendering completion state, calculates high-risk area reading coverage, and controls diagnosis report submission permission. The application effectively reduces the transmission bandwidth of gigapixel-level sections, improves loading flicker, and reduces the risk of missed diagnosis caused by network delay or human negligence.

This application relates to the field of medical information technology and discloses an AI-based automatic diagnosis system for case consultation and pathological slides. The system includes a server and a client. The server uses a deep learning model to generate a binary mask of lesions, constructs a sparse quadtree index carrying a composite state bitmask, and generates vector slices with background removed. The client distributes resource requests to multi-level queues based on the bitmask in the index, prioritizing the loading of high-value region images. A three-layer heterogeneous rendering architecture is adopted, using adaptive Gaussian blur to process the vector slices for smooth transitions. Simultaneously, a behavior auditing module completes the state acquisition viewport trajectory based on physical rendering, calculates the coverage rate of high-risk areas, and controls the submission permissions of diagnostic reports. This invention effectively reduces the transmission bandwidth of gigapixel-level slices, improves loading flicker, and reduces the risk of missed diagnoses due to network latency or human error.

The present application relates to a kind of flow optimization method and system for ship-shore offline data transmission under restricted network, the system first real-time acquisition ship end including the multi-source data of safety alarm data, collision avoidance trajectory data and sensor data, and carry out differentiated preprocessing to generate lightweight data set, then real-time compression processing is carried out;Then construct the priority layered model including multi-stage queue scheduling mechanism, transmission scheduling is implemented according to safety level, and flow quota monitoring and fuse protection are carried out, and network state is monitored in real time, when network state is abnormal, suspend the data transmission of all queues, and record the transmission breakpoint state of each queue and the data packet sequence number of uncompleted transmission when abnormal occurs, finally, after network state recovers, based on the transmission breakpoint state of each queue and the offline data packet sequence number of uncompleted transmission recorded, data is continued according to priority order, and the data integrity is verified using hash check method, complete the flow optimization of ship-shore offline data transmission.

The present application belongs to the technical field of storage system scheduling optimization, and discloses a storage system, a multi-queue RTC coroutine scheduling method and a scheduler. The scheduling method comprises the following steps: moving a read task to a main queue; selecting tasks in a multi-level queue as ready tasks one by one, and the high-priority queue in the multi-level queue is prior to the main queue; regularly polling a storage device to call a target data block of a suspended task and then placing the suspended task in a high-priority queue; if the selected task is from the main queue, analyzing query information and searching in a cache; if the target data block is not found in the cache, constructing a storage device read request to enable the storage device to query internally and suspend the current ready task, and then executing the next ready task; if the current selected ready task is from the high-priority queue, performing a request response, and then executing the next ready task. The above scheduling method can reduce the tail latency of the storage system when performing RTC coroutine scheduling in the same execution domain.

The invention discloses a long-term violation video evidence storage method adopting hierarchical queues, and belongs to the technical field of violation detection. Defining a multi-stage queue (capacity and sampling rate), judging whether the multi-stage queue meets evidence synthesis of violation videos, and counting input violation images for each stage of queue in the multi-stage queue; continuously storing the input violation images according to the sampling rate of the corresponding level queue; judging whether the storage capacity reaches the capacity of the corresponding level queue or not; when the storage capacity reaches the capacity of the corresponding-level queue, judging whether the corresponding-level queue has a next-level queue or not; if yes, removing the frame of violation image stored earliest in the corresponding-level queue, and transmitting the frame of violation image to the next-level queue of the corresponding-level queue; and otherwise, removing the frame of corresponding-level queue image stored earliest in the corresponding-level queue. Finally, long-term storage capability and dynamic configuration are realized, and integrity and reliability of video storage are ensured.

This application discloses a method and system for segmented resource reclamation in large-scale network scenarios of a network range, relating to network ranges and cyberspace security technologies. The method includes: parsing the resource topology of the target network scenario to generate a resource list and individual dependency relationships in a specified format; formulating a segmented, structured reclamation plan based on the parsing results; breaking down the reclamation plan into independent reclamation tasks and placing these tasks into multi-level queues; dynamically generating queues based on the task types of the reclamation tasks, sorting the queues according to the weight of the task types, with higher-weighted tasks having higher priority; and executing reclamation operations according to priority. This application effectively solves the problems of excessively long reclamation processes and low resource reuse rates in existing technologies, significantly improving resource reclamation efficiency and accelerating the turnover speed of range resources.

The present application provides a resource management method and system for a server operating system, the method comprising: performing double-layer graph modeling on process and resource data in the server operating system to obtain a process competition layer and a resource dependency layer; using a preset graph neural network to perform feature extraction on node information of the process competition layer and node information of the resource dependency layer to obtain a node feature vector containing a competition relationship and an allocation state; performing deadlock probability calculation using a multi-layer perceptron based on the node feature vector to obtain a process priority adjustment scheme; performing resource allocation strategy learning based on the node feature vector and the process priority adjustment scheme to obtain a first resource allocation scheme; performing multi-level queue allocation on the process based on the first resource allocation scheme to obtain a second resource allocation scheme; and performing resource management on the server based on the second resource allocation scheme.

The invention relates to a flow optimization method and system for ship-shore off-line data transmission under a limited network, and the system comprises the steps: firstly collecting multi-source data, including safety alarm data, collision avoidance trajectory data and sensor data, of a ship end in real time, carrying out the differential preprocessing to generate a lightweight data set, and then carrying out the real-time compression processing; then, constructing a priority hierarchical model comprising a multi-stage queue scheduling mechanism, implementing transmission scheduling according to the security level, performing traffic quota monitoring and fusing protection, monitoring the network state in real time, and pausing data transmission of all queues when the network state is abnormal; the method comprises the following steps of: acquiring an abnormal state of each queue, recording a transmission breakpoint state of each queue at an abnormal moment and a serial number of a data packet which is not transmitted completely, continuously transmitting data according to a priority sequence based on the recorded transmission breakpoint state of each queue and the serial number of the offline data packet which is not transmitted completely after a network state is recovered, and verifying data integrity by adopting a Hash verification method. And flow optimization of ship-shore off-line data transmission is completed.

The embodiment of the invention provides a prize issuing control method and system, the prize issuing control system comprises a first-level queue and a second-level queue, and the method comprises the steps that a prize issuing message is received, and the prize issuing message comprises a business source and a prize type; issuing the prize issuing message to a target channel of the first-level queue according to the business source; performing integration processing on the prize issuing message in the target channel according to the prize type, and issuing the integrated prize issuing message to a target sub-queue of the secondary queue; and the prize issuing message in the target subqueue is consumed to realize prize issuing. Prize issuing is carried out in a multi-level queue design mode, the first-level queue can be used for channel isolation to solve the resource competition problem of upstream services, the second-level queue can flexibly limit the speed of the consumption process of prize issuing messages, and when multiple services are in resource competition, manual intervention can be carried out to guarantee that high-level services are carried out in advance. And the prize issuing efficiency is improved.

The application relates to a phased array radar task scheduling method, which comprises the following steps: a task registration center sends a radar task scheduling request to a scheduling module; a comprehensive priority calculation module calculates a current task comprehensive priority and generates radar task priority data; and a scheduling decision module sends the radar task to a multi-level priority queue according to the initial comprehensive priority of the corresponding radar task and inserts the radar task into the tail of the corresponding priority queue. The application has the beneficial effect that the dynamic scheduling process in the application adopts a scheduling program combining sequential execution and parallel execution, which can realize timely scheduling of phased array radar tasks, meet the scheduling priority demand of high-priority phased array radar tasks, and improve the overall scheduling efficiency of the radar scheduling system through the dynamic scheduling mode of the multi-level queue.

The disclosure herein describes using satellites and ground sinks and / or stations for routing IoT device data packets from IoT devices. A target ground sink in range of the satellite is identified and an expected reception (ER) score for the target ground sink is calculated based on ER parameter data and location data of the satellite. A data packet in a first level of a multi-level data structure of the satellite is sent to the target ground sink and, based on an ER threshold exceeding the ER score, the packet is moved to a second level of the multi-level data structure, whereby the data packet is queued to be sent to another ground sink. The disclosure further includes using cell towers as ground sinks and / or using them for backhauling with other ground sinks. The flexibility of the disclosure enables large ground sink networks to be established, reducing latency of packet routing.