259 results about "Cpu load" patented technology

The invention belongs to the technical field of computers, and provides a video display and storage method and system based on a wayland protocol and a storage medium in order to solve the problems that an existing wayland video system is high in rendering delay, high in CPU load, low in storage efficiency and poor in cross-platform compatibility. Video frames are directly written into an Overlay layer through DRM-KMS and mixed by bypassing a synthesizer, meanwhile, CPU memory amplitude overhead is eliminated through zero copy transmission, an ISP-GPU-VPU direct transmission assembly line is constructed, full-link DMA-BUF direct transmission is achieved, and delay is greatly reduced; iSP preprocessing, GPU shader scaling and VPU coding are all executed by special hardware, so that the load is reduced, and the multi-path processing capability is improved; vPU dynamic code rate compression is utilized, segmented storage is carried out according to events / time, metadata is synchronized to a database, redundant storage is avoided, and space is saved.

The invention discloses an industrial personal computer and multi-graphics card collaborative parallel computation acceleration system, which relates to the technical field of industrial resource allocation and parallel computation, and comprises a resource monitoring and predicting module, a resource management module and a prediction type resource preparation module, the task splitting and collaborative execution module comprises a task splitting module, a cross-node collaborative module and a collaborative operation engine; the intelligent scheduling and dynamic resource allocation module comprises an intelligent scheduler, a dynamic resource allocation module and a conflict avoidance module. According to the method, the GPU video memory utilization rate, the core utilization rate, the temperature, the video memory fragment rate, the available video memory total amount, the CPU core total utilization rate, the load condition and the idle core number index are collected in real time through a resource monitoring module, and the video memory capacity, the GPU core occupancy rate and the CPU load requirement are predicted in advance before a task is submitted in combination with a gradient boosting decision tree and a neural network prediction model; and resources are reserved, so that the scheduling delay is remarkably reduced, and the scheduling hit rate is improved.

The invention provides a dynamic CPU scheduling method and system for an RISC-V platform, and relates to the technical field of resource allocation scheduling, and the method comprises the steps: obtaining a hardware performance event and a CPU load of the RISC-V platform in real time; feature selection is carried out on the collected hardware performance events, and a prediction data set is constructed; training a load prediction model based on reinforcement learning by utilizing the prediction data set so as to predict the load value of each CPU; adjusting the active state of the CPU based on the load prediction result of the CPU; according to the adjusted CPU state, tasks are redistributed, and the processor load is balanced; according to the RISC-V platform dynamic CPU scheduling method and system, hardware performance events and load prediction model design are combined, the fine-grained performance monitoring capacity in a system structure is fully utilized, optimal balance between performance and energy consumption is achieved, and the problem that in a traditional method, resource allocation is excessive or insufficient due to hysteresis is solved.

The invention discloses an efficient data compression and transmission method based on a computing power host management operation platform, and relates to a large-scale data concurrent management technology. Comprising the steps of 1, integrating a lightweight data feature analyzer at a Kafka producer end, extracting feature indexes of data streams in real time through the data feature analyzer, and generating compression strategy parameters according to the feature indexes, and 2, performing feature recognition on structured data and unstructured data, extracting data features, and generating compression strategy parameters according to the compression strategy parameters. And step 3, based on data characteristics, CPU load of producers / consumers and network bandwidth prediction, constructing a multi-objective optimization model, dynamically selecting a compression algorithm and a compression level, selecting compression strategy parameters according to the compression algorithm, introducing a compression income ratio index, calculating a ratio of compression rate increase to compression time consumption, and performing compression strategy optimization. And preferentially selecting a compression algorithm with a high gain ratio to perform data compression and transmission.

The invention discloses a sintering process real-time optimization method based on edge calculation, and the method comprises the steps: obtaining original multi-index time series data collected at the edge side of a sintering process in real time, and recording sampling data and corresponding process labels for sensing points at different workshop section positions in a sintering production line; based on the real-time prediction task load, the edge node computing power and the network bandwidth monitoring result, whether the current node reaches an elastic migration triggering condition or not is judged, and if yes, subsequent task migration operation is executed; the calculation and storage quotas of edge nodes are dynamically adjusted based on the GPU / CPU load, the storage space and the prediction task fluctuation of each node; according to the invention, by capturing the deep relationship of time-varying coupling between process indexes, an elastic and dynamic resource scheduling and model adaptive mechanism is constructed, the edge-side dynamic graph modeling capability is enhanced, and a model self-learning optimization mechanism is introduced; the accuracy, the real-time performance and the edge system adaptability of multi-index collaborative prediction in the complex industrial process are effectively improved.

The invention discloses an AI-driven RDMA network card hardware unloading acceleration method and system based on an RISC-V architecture, the system comprises an RISC-V processing unit, the RISC-V processing unit comprises a control module and a driving module, and the driving module stores RDMA exclusive acceleration instruction set extension; the AI reasoning unit is connected with the driving module and is used for expanding and running a pre-trained network optimization model based on an RDMA exclusive acceleration instruction set to analyze a network state and an application requirement, generating a hardware unloading strategy optimization instruction and sending the hardware unloading strategy optimization instruction to the control module; the control module generates a corresponding management command according to the hardware unloading strategy optimization instruction; and the hardware unloading unit comprises a management module and an RDMA component module, and the management module receives the management command and performs parameter configuration on the RDMA component module according to the management command. According to the invention, the burden of the CPU is reduced, and the throughput is improved.

A technical field of motor control, and a motor rotation control method and device. The method includes conducting inverse Park transformation and inverse Clark transformation to control output results of a position loop, a speed loop and a current loop to obtain SVPWM signals of a present cycle; decoupling an interrupt control cycle from a PWM signal reference channel; acquiring present position sampling information of a motor, and predicting the position sampling information to obtain position result information of a motor rotor in predicted cycles. In the motor rotation control method, an interrupt function trigger signal is added to realize decoupling from the PWM signal reference channel, so that SVPWM signal update frequency of motor control will not be affected when CPU load is reduced by adding an interrupt function scheduling cycle.

The invention discloses a large model deployment method based on heterogeneous resource scheduling and multi-model collaborative reasoning, and relates to the field of artificial intelligence and server system deployment. Comprising the following steps of: 1, calculating a deployment score of a target model by utilizing a deployment scoring function according to a GPU video memory residual rate, a CPU load, network delay and a current task quantity of each node through a heterogeneous resource sensing scheduler, and selecting a node with the highest score of the target model to load or reuse the target model; a cache index key is generated, and whether key value KV pairs supporting multiplexing of the target model exist or not is inquired; step 3, if the cache is hit, skipping a forward propagation stage, and directly decoding based on a cache result, otherwise, executing complete forward reasoning and writing the result into the cache; 4, based on a plurality of user requests, performing fusion according to user priorities, model weights, reasoning costs and cue word lengths to generate execution sorting weights, and scheduling execution models in batches according to the weights; and 5, returning the model response to the user, and releasing or updating node state information.

The invention discloses a distributed acceleration method and system for a RocksDB storage system based on CSD master control. The method comprises the steps that CSD is deployed in the RocksDB storage system to serve as a computing node, a computing cluster is constructed, a Compaction process is stripped and migrated into the CSD, and an RPC channel is constructed based on the CSD; the compact instruction is directly written into a CSD memory, file meta-information is obtained through analysis, and a corresponding file is read; data needing to be calculated and a calculation instruction are sent to a calculation cluster of the CSD for calculation analysis, and calculation results of the far end are aggregated to the local; and migrating the data to the nand flash memory according to a storage instruction sent by the host side. According to the technical scheme, the CPU load of the host side is effectively reduced, the file data interaction efficiency is greatly improved, and the energy overhead and the time overhead of data migration interaction are greatly reduced.

The invention discloses a multi-node communication architecture based on RDMA (Remote Direct Memory Access), which comprises the following modules on a network card: a PCIe (Peripheral Component Interconnect Express) function module which supports PCIe x16, is downwards compatible, is used for data interaction between a master control on a local node and an RDMA function module, and supports high-speed transceiving; the MAC module is used for supporting high-speed data transceiving between the RDMA functional module and an external node by using a high-bandwidth MAC; and the RDMA function module is used for realizing a standard RoCEv2 protocol, supporting unicast and multicast, supporting reliable connection service and unreliable datagram service, supporting bilateral send / recv operation, and supporting unilateral write and read operation. The architecture can be used for a multi-core CPU heterogeneous airborne embedded supercomputing platform, realizes lossless Ethernet communication, has the characteristics of low delay, low CPU load and the like, and is used for meeting the application requirements of future airborne comprehensive processing platforms.

The invention provides a multi-modal data active recommendation method based on cross-modal dynamic fusion and equipment resource perception. The method is applied to the technical field of data processing. The method comprises the following steps: acquiring multi-modal data through a smart home device end; performing modal feature extraction on the multi-modal data to generate a voice feature matrix, an environment probability matrix and an equipment state vector matrix; adopting a cross-modal dynamic fusion algorithm, combining time dimension and space dimension weights, and carrying out adaptive weighted fusion on the voice feature matrix, the environment probability matrix and the equipment state vector matrix; the CPU load and the memory occupancy rate of the central control equipment are detected in real time through the equipment resource monitoring module, and a data calculation strategy is dynamically adjusted; and based on the fused multi-modal feature matrix, generating an active recommendation instruction through user portrait modeling and knowledge graph association, and driving the smart home device to execute adaptive adjustment. In this way, the recommendation efficiency can be improved.

The invention provides a zero-copy rapid transceiving device, a data sending method and a data receiving method, and relates to the technical field of network communication, the zero-copy rapid transceiving device comprises a network card in which a direct memory access controller and an associated annular cache region are configured; the kernel protocol stack is configured with a data cache structure body, and a data pointer of the data cache structure body and a packet descriptor of the annular cache region are configured to point to the same memory region; when the network card receives external data, the direct memory access controller copies the data to the memory area according to the direction of the packet descriptor so that the data can be processed by the kernel protocol stack, and when the kernel protocol stack sends the data, the data is copied to the memory area according to the direction of the data pointer so that the data can be sent by the direct memory access controller. The method has the beneficial effects that data copying is not needed in a driving layer message data processing process, so that storage resources are effectively saved, the load of a CPU (Central Processing Unit) is reduced, and the network bandwidth and the system performance are effectively improved.

The invention provides a multi-player voice game performance optimization method and device based on dynamic resource management. The method comprises the following steps: acquiring a core source code of a game voice tool and pre-compiling the core source code to generate a byte code file; analyzing a target scene demand and pre-loading a related resource file; performing real-time acquisition and analysis on memory occupation and CPU load to obtain performance state data of the current system; based on the performance state data, when it is detected that memory occupation and / or CPU load exceed a preset threshold value, a pre-compiling mode and a resource loading strategy are automatically adjusted; when it is detected that the multi-person voice function is started, resources and byte codes needed by voice processing are placed in a high-priority queue, and cooperative scheduling is carried out on memory occupation and / or CPU loads. According to the method, the loading delay can be reduced, the frame rate can be improved, the communication delay can be reduced, and real-time voice processing and high-load requirements of games can be dynamically considered.

The invention discloses an Android video self-adaptive secondary transcoding method and system and a medium, mainly relates to the technical field of secondary transcoding, and is used for solving the problem that part of videos cannot be played on some Android devices due to a transcoding mode in an existing scheme. Or the video playing experience is difficult to comprehensively optimize due to the fact that the video playing experience cannot be comprehensively optimized due to the fact that the video playing experience is not combined with the real-time feedback in the playing process. Comprising the steps that when it is determined that a playing report error occurs and the error indicates that the coding format is incompatible, the format with the highest priority in the coding formats supported by equipment is selected from a coding list contained in software parameters to serve as the final coding format, and secondary transcoding is conducted according to the final coding format; triggering a screen resolution and code rate down-regulation program when lagging occurs, a playing error occurs, the error is not incompatible in coding format or playing stagnation occurs, and the CPU load in the hardware parameters is greater than a preset value; and carrying out secondary transcoding based on the down-regulated screen resolution and code rate.

The invention discloses a method for dynamically setting dog feeding time of a watchdog circuit, which comprises the following steps of: collecting and monitoring power consumption of a CPU (Central Processing Unit) and a corresponding load, and fitting the load and power of the CPU according to monitoring data so as to reflect the running state condition of a system; power consumption of a monitoring CPU is collected in real time, the real-time CPU load is calculated according to the fitting relation between the CPU load and power, and the watchdog feeding time of a watchdog circuit is determined according to the CPU load; a square wave period is set, square waves are collected to set timeout time, and the timeout time is kept consistent with dog feeding time, determined according to CPU loads, of a watchdog circuit. According to the method, the CPU power consumption and the load are fitted, the CPU load is monitored in real time, the square wave period is adjusted, the dog feeding time is dynamically adjusted according to the CPU load, resource waste is avoided, and the method is suitable for the fields of embedded systems, industrial control and the like.

The invention relates to the field of search engines, and discloses a search optimization method and device, equipment and a medium. Performing weighted calculation according to the CPU load, the memory occupation, the data read-write speed, the network throughput and the index query frequency, and determining the load score of each node; if it is monitored that the load fraction of any node meets preset conditions, different search optimization schemes are executed according to the preset conditions met by the current node, and the preset conditions comprise the first preset condition, the second preset condition and the third preset condition; the search optimization scheme is in one-to-one correspondence with an index migration scheme of a high-load node, an index pre-distribution scheme of a new node and a load balancing scheme of a hotspot index based on different preset conditions; and in response to the current search optimization scheme, adjusting index distribution of each node to complete search optimization. By dynamically sensing load data, adaptive weight hash mapping is performed according to a search optimization scheme, so that the query performance is improved, and the storage utilization rate is optimized.

The invention relates to the technical field of converters, in particular to a three-phase converter control method and system, a medium and equipment. The method comprises the steps that a main processor transmits sampling original data of a three-phase converter to a first storage area; the coprocessor reads the sampled original data, performs operation to obtain operation result data, and transmits the operation result data to the second storage area; the coprocessor runs at a period higher than the preset control frequency of the main processor, and immediately triggers the operation of the sampled original data after completing the collection of the sampled original data; the main processor performs control loop calculation, reads the operation result data when the newest operation result data is needed, and loads the newest operation result data for updating operation; the main processor runs the control program at the preset control frequency, and the control program triggering time is set in a minimum allowable time window before the control output quantity loading moment. Through the above arrangement, real-time sampling is effectively realized, CPU load is optimized, and the control performance of the converter is improved.

The invention discloses an Android Service plug-in implementation method, an Android Service plug-in implementation system, Android Service plug-in implementation equipment and a medium. The Android Service plug-in implementation method specifically comprises the following steps: deploying proxy service in a host application, creating a standard interface IServiceInterface, and fusing host and plug-in resources by rewriting a context loading method of Android Service; constructing a service life cycle prediction model through a federated learning framework based on multi-device service behavior data collected during proxy service operation; generating a dynamic scheduling strategy based on an output result of the service life cycle prediction model in combination with the equipment electric quantity, the network type and the CPU load; and based on a dynamic scheduling strategy, writing the plug-in metadata hash, the federated model fingerprint and the device state snapshot into a block chain smart contract, and verifying the integrity of the plug-in through a light node on the chain to obtain a block chain verification result. According to the method, the stability and the reliability of the Service plug-in are improved, and powerful support is provided for the hot update function of the Android application.

The invention provides a game performance optimization method and device based on dynamic scheduling and a storage medium. The method comprises the following steps: dynamically adjusting a byte code compiling strategy based on operation data, and reducing the execution priority of a byte code compiling task when monitoring that a multi-person voice team function is started and a CPU load reaches a preset condition; performing dynamic scheduling on allocation of the CPU cores based on the operation data, allocating high-priority tasks to the CPU cores with loads lower than a preset load threshold for execution, and allocating other tasks to the remaining CPU cores; carrying out scheduling and intelligent cache management on the resource file according to the operation data; and when monitoring that the network bandwidth or the network delay reaches a preset condition, adjusting a multi-person voice data transmission strategy based on the operation data. According to the method and the device, dynamic scheduling can be performed according to the real-time state during game running, tasks and resources are flexibly allocated, CPU occupation is reduced, and game running lagging is avoided.

The invention discloses an audio test data processing method, system and device based on zero-copy NIO and a dynamic sliding window, and the method comprises the following steps: S1, building a memory mapping file channel, and directly mapping an audio file to a process address space; s2, constructing a double-buffer processing pipeline, and realizing asynchronous decoupling of a collection thread and a calculation thread by adopting a producer-consumer mode; and S3, a dynamic sliding window mechanism is adopted to maintain a real-time audio data interval, and the window size is dynamically adjusted according to a television CPU load. According to the technical scheme of the invention, the problems of high memory occupation, high detection delay and low cross-application data reading efficiency existing in audio testing of a television system in a multi-application concurrent scene are solved.

The invention provides a multi-time-scale CPU load stability evaluation method used in the testing field, relates to the technical field of CPU load stability evaluation, and aims to more comprehensively capture dynamic characteristics of a system load, and the method comprises the following steps: obtaining a multi-scale load ratio based on a plurality of time scales; obtaining a load stability coefficient according to the multi-scale load ratio, wherein the load stability coefficient is used for describing the stability degree of the load; obtaining a weighted load index according to the multi-scale load ratio, wherein the weighted load index is used for endowing different weights for different time scales and obtaining an overall load level; and performing system state evaluation according to the load stability coefficient and the weighted load index. The CPU load stability evaluation method has the advantage that a more accurate and reliable CPU load stability evaluation result can be obtained.

The invention discloses a resource allocation method and device based on bit map pre-calculation, and the method comprises the steps: generating a global resource state bitmap according to the CORESET configuration in a PDCCH, and carrying out the pre-calculation of each possible candidate PDCCH to form a candidate pre-allocation bit map set; extracting a corresponding pre-allocation bitmap from the candidate pre-allocation bitmap set as a candidate set bitmap; and comparing the candidate set bitmap with the global resource state bitmap, judging whether the resources are conflicted or not, and if the resources are not conflicted, determining that the resources are available and further occupying the resources. Complex function calculation at the scheduling moment is transferred to pre-calculation at the initialization stage, only simple bitmap retrieval and bit operation are needed during scheduling, the PDCCH resource allocation time consumption is greatly shortened, real-time intensive calculation is replaced with pre-stored bitmaps, the CPU load is remarkably reduced, a processor focuses on a core scheduling algorithm, resource redundancy consumption is reduced, and the scheduling efficiency is improved. And the overall capacity and the energy utilization efficiency of the communication system are effectively improved.

Systems and methods are disclosed for optimizing network routing and data processing through dynamic route optimization. The invention features a dynamic route optimization engine that uses real-time performance metrics such as latency, bandwidth availability, memory utilization, CPU load, and throughput capacity. It allows for rule-based configurations, enabling custom parameters for transaction processing, data retrieval, and user authorization. Integration with network monitoring tools provides continuous visibility into network performance. The system supports multi-cloud and hybrid cloud environments, dynamically selecting optimal paths based on real-time conditions and regulatory compliance. Predictive analytics anticipate network congestion, allowing proactive routing adjustments. Adaptive algorithms learn from past conditions to optimize performance continuously. Scalability is achieved through automated updates to routing tables and security groups. The invention reduces latency and ensuring efficient data transfer, providing a cost-effective and reliable solution for modern network management, addressing a long-standing need for adaptability and efficiency in network routing.

Systems and methods for anomaly detection are described and contemplated herein. An Isolation Forest algorithm is implemented for both training a plurality of machine learning models and accurately detecting multiple anomaly patterns in computer equipment time series data, such as CPU loads, temperatures, RAM usage, and other computer equipment metrics.

The invention discloses an HPC cluster benchmark energy efficiency adjusting and optimizing method based on software and hardware joint optimization, and belongs to the field of cluster energy efficiency adjusting and optimizing. The method is divided into two aspects of software tuning and hardware tuning, in the software aspect, after energy efficiency sensitive parameters under the operation of a reference real machine are obtained through parameter importance analysis, a sampling algorithm in the parameter optimization process is optimized in combination with an energy efficiency prediction model, the optimization and model training overhead is reduced, the model convergence speed is increased, and the optimization efficiency is improved. The off-line low-cost parameter optimization is realized; in the aspect of hardware, on the basis of optimal cluster operation parameters, feature extraction and training are carried out on the utilization rate of a CPU core group in a reference operation period to obtain a clustering model for identifying a CPU load mode, and the resource demand of a load can be judged from a global perspective, so that corresponding dynamic frequency modulation is carried out, and the energy efficiency operation performance of the cluster is improved.

The present invention discloses a cloud platform service load prediction method based on a time series convolutional neural network, including: S1 collecting CPU load time series data to form a training sample set; S2 preprocessing and feature extraction of the training samples, wherein the feature extraction includes a periodic attention vector and a peak attention vector, and weighted summing the training samples, the periodic attention vector and the peak attention vector to obtain a final training sample; S3 constructing a time series convolutional neural network, and taking the final training sample as an input signal; S4 training the time series convolutional neural network; S5 inputting test sample data into the trained time series convolutional neural network to obtain a CPU load prediction result of the test sample. The present invention has high prediction accuracy and is conducive to solving the problems of hysteresis and service quality degradation that may be caused by responsive automatic scaling.

The embodiment of the invention provides a time synchronization method applied to an electronic mechanical braking system and the braking system. The braking system comprises a main controller and at least one auxiliary controller. The first clock is arranged on the main controller; the at least one second clock is arranged on the at least one auxiliary controller in a one-to-one correspondence manner; the main controller synchronously sends a first level signal and a second level signal to at least one auxiliary controller at a first moment and a second moment respectively through a hard wire; and when the received second level signal is different from the first level signal, the at least one auxiliary controller sets the at least one second clock to be the same as the first clock. The main controller and the auxiliary controller are connected through the hard line, the hard line is used for replacing a CAN line to carry out time synchronization, the load overhead of a CAN bus is reduced, time synchronization is carried out through the change of a level signal, a CPU is not needed for calculation, CPU resources are not occupied, the CPU load is reduced, and the synchronization precision is improved.

The invention relates to the technical field of resource allocation, in particular to an object recognition and task processing method and system for multi-machine cooperation of competition robots, and the method comprises the following steps: pre-defining a plurality of service quality levels of a target object recognition and path planning calculation task of a cooperative robot, and obtaining CPU resource consumption and task processing performance corresponding to each grade based on the visual sensor data resolution and algorithm complexity under different grades. According to the method, a plurality of task service quality levels are defined in advance, and different visual data resolutions and algorithm complexity are mapped into CPU resource requirements and task processing performance indexes, so that quantitative association of tasks and resource requirements is realized, and the resource allocation accuracy is improved; and meanwhile, the CPU load state of the robot cluster is sensed in real time, a dynamic QoS regulation and control instruction is established by combining the load change trend and the task emergency degree, and the real-time performance and sensitivity of resource allocation are improved.

The invention discloses an FPGA (Field Programmable Gate Array)-based numerical control system, which comprises an FPGA chip and a plurality of drivers, and is characterized in that the FPGA chip comprises a DMA (Direct Memory Access) core, a descriptor management module, a first bus management module, a second bus management module and a plurality of transmission channels; the DMA core is connected with the first bus management module, the second bus management module and the descriptor management module, the first end of the transmission channel is connected with the first bus management module and the second bus management module, and the second end of the transmission channel is connected with the driver; the descriptor management module is connected with the first bus management module and is used for generating a descriptor chain table; the DMA core is used for traversing the descriptor chain table to generate a data transmission request, and the second bus management module determines a driver based on the data transmission request and gates a transmission channel corresponding to the driver. The descriptor management module is configured in the FPGA chip, so that multiple times of data interaction between the CPU and the FPGA chip is avoided, the CPU load is remarkably reduced, and the communication efficiency of the numerical control system is improved.

The invention relates to a cloud computing resource load prediction method based on TCN-PatchTST (Trusted Cryptographic Network-PatchTST). Comprising the following steps: collecting CPU load time sequence data to form a training sample set; carrying out preprocessing and feature construction on the training sample, and selecting a memory utilization rate load value which has obvious influence on a CPU load value as an input feature to predict the CPU load value; according to the method, a TCN-PatchTST neural network is constructed, a two-channel feature extraction architecture is designed, local time sequence features are extracted through causal convolution and expansion convolution of TCN, a long-term dependency relationship is captured by using partitioning processing and a channel independence mechanism of PatchTST, and a feature fusion layer is designed to integrate multi-scale information; the TCN-PatchTST neural network is trained; and inputting a test sample into the trained TCN-PatchTST neural network to obtain a CPU load value prediction result of the test sample. According to the method, load prediction is carried out through the historical load time sequence data, the prediction precision of the CPU load value of the cloud computing resources is effectively improved, and technical support is provided for intelligent scheduling of the cloud resources.