An integrated management system based on an artificial intelligence computing platform

Abstract

This invention discloses an integrated management system based on an artificial intelligence computing platform, belonging to the field of integrated management technology. The system includes the following steps: resource monitoring and acquisition to obtain heterogeneous computing resource status information; resource health assessment based on the heterogeneous computing resource status information to obtain resource availability indicators; task resource matching based on the resource availability indicators to obtain a resource allocation scheme; computing task scheduling based on the resource allocation scheme to obtain a task execution sequence; task execution monitoring based on the task execution sequence to obtain the real-time task execution status; and resource adjustment based on the real-time task execution status to obtain the resource allocation adjustment result. This invention improves matching accuracy and resource utilization, avoids resource waste caused by single resource bottlenecks, and enhances load balancing and overall operational stability across nodes.

Description

Technical Field

[0001] This invention relates to the field of integrated management technology, specifically to an integrated management system based on an artificial intelligence computing platform. Background Technology

[0002] With the rapid development of artificial intelligence computing platforms and the increasingly diverse and dynamic demands of massive tasks for computing resources, how to achieve integrated management of computing resource status, efficient utilization, and reasonable allocation has become a key technical issue.

[0003] In existing technologies, task resource allocation has shortcomings: In existing technologies, task resource matching often adopts fixed heuristic algorithms or simple resource satisfaction judgments, such as only judging whether the resources required by the task are less than the currently available resources of the node, and lacks combination with real-time resource availability indicators, making it difficult to reflect the actual needs and matching degree of the task for resources, affecting the stability of task execution and the overall load balancing of the system. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an integrated management system based on an artificial intelligence computing platform to solve the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] In a first aspect, embodiments of the present invention provide an integrated management system based on an artificial intelligence computing platform, comprising the following steps:

[0007] S1. Resource monitoring and acquisition: Obtain heterogeneous computing resource status information;

[0008] S2. Perform a resource health assessment based on the heterogeneous computing resource status information to obtain resource availability indicators;

[0009] S3. Match task resources based on resource availability indicators to obtain a resource allocation plan;

[0010] S4. Calculate task scheduling based on the resource allocation scheme to obtain the task execution sequence;

[0011] S5. Monitor task execution based on the task execution sequence to obtain the real-time execution status of the tasks;

[0012] S6. Adjust resources based on the real-time execution status of the task to obtain the resource allocation adjustment results.

[0013] To further optimize this technical solution, the resource health assessment in step S2 includes:

[0014] Based on the obtained real-time heterogeneous computing resource status information, a resource health assessment model is used to perform a nonlinear assessment of the heterogeneous computing resources by fusing the status and historical stability of each resource, thereby obtaining the resource availability index of each computing node.

[0015] To further optimize this technical solution, the resource health assessment model includes:

[0016]

[0017] in:

[0018] : Resource availability metrics for the j-th node;

[0019] : Sigmoid function;

[0020] : The weighting coefficient for the current state score;

[0021] : Weighting coefficients for historical stability scores;

[0022] : Weighting coefficients of the coupling terms;

[0023] The current state score of the j-th node;

[0024] : Historical stability score of the j-th node.

[0025] To further optimize this technical solution, the current status score includes:

[0026]

[0027] in:

[0028] : CPU idle rate of the j-th node;

[0029] : GPU idle rate of the j-th node;

[0030] : The memory free rate of the j-th node;

[0031] : Weighting coefficient for CPU idle rate;

[0032] : Weighting coefficient for GPU idle rate;

[0033] Weighting coefficient for memory free rate;

[0034] Calculate the node's current status score based on the current CPU and GPU idle rates and memory idle rates.

[0035] To further optimize this technical solution, the historical stability score includes:

[0036]

[0037] in:

[0038] The number of historical observation periods;

[0039] : The anomaly rate of node j in the k-th historical observation period;

[0040] The historical stability score is obtained by calculating the normality rate of nodes in all historical observation periods.

[0041] To further optimize this technical solution, the task resource matching in step S3 includes:

[0042] Based on the obtained resource availability index and the resource requirements of the tasks to be executed, a task matching model is used to calculate the matching score between tasks and resource nodes, and tasks are allocated according to the matching score to obtain a resource allocation scheme.

[0043] To further optimize this technical solution, the task matching model includes:

[0044]

[0045] in:

[0046] : Matching and adaptation score between task i and the j-th node;

[0047] : The degree of matching between the resource requirements of task i and the j-th node;

[0048] : Historical compatibility factor between task i and the j-th node;

[0049] : Weighting coefficients for resource demand matching degree;

[0050] Weighting coefficients for resource availability metrics;

[0051] : Weighting coefficient of historical compatibility factor.

[0052] To further optimize this technical solution, the resource requirement matching degree includes:

[0053]

[0054] in:

[0055] : Estimated CPU utilization for task i;

[0056] : Expected GPU utilization for task i;

[0057] : Estimated memory usage for task i;

[0058] The weighting coefficient of the remaining available CPU;

[0059] The weighting coefficients of the remaining available GPUs;

[0060] Weighting factor for remaining available memory;

[0061] The resource requirement matching degree between nodes and tasks is calculated based on the available resources of the nodes and the resources required by the tasks.

[0062] To further optimize this technical solution, the historical adaptation compatibility factor includes:

[0063]

[0064] in:

[0065] : The number of times task i was successfully executed on node j;

[0066] : The total number of times task i is scheduled on node j;

[0067] Small constants that prevent division by zero;

[0068] The historical compatibility factor between a task and a node is calculated based on the ratio of the number of successful executions of a task on a node to the total number of schedulings.

[0069] To further optimize this technical solution, the functional modules include:

[0070] The module includes: resource status acquisition module, resource health assessment module, task resource matching module, resource allocation and execution module, task operation monitoring module, and resource dynamic adjustment module.

[0071] In a second aspect, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program instructions, when executed by the processor, implement the steps of an integrated management system based on an artificial intelligence computing platform as described in the first aspect of the present invention.

[0072] Thirdly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, they implement the steps of an integrated management system based on an artificial intelligence computing platform as described in the first aspect of the present invention.

[0073] Compared with existing technologies, the present invention provides an integrated management system based on an artificial intelligence computing platform, which has the following beneficial effects:

[0074] This integrated management system based on an artificial intelligence computing platform improves matching accuracy and resource utilization by comprehensively considering multiple resource dimensions through a task matching model. It avoids resource waste caused by single resource bottlenecks and responds to node resource changes in real time. It also matches nodes based on their historical task execution status, thereby improving the load balance of each node and the overall operational stability. Attached Figure Description

[0075] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0076] Figure 1 This is a flowchart illustrating an integrated management system based on an artificial intelligence computing platform proposed in this invention.

[0077] Figure 2 This is a flowchart illustrating a resource health assessment model for an integrated management system based on an artificial intelligence computing platform proposed in this invention.

[0078] Figure 3 This is a flowchart illustrating the task matching model of an integrated management system based on an artificial intelligence computing platform proposed in this invention.

[0079] Figure 4 This is a flowchart illustrating the execution status scoring model of an integrated management system based on an artificial intelligence computing platform proposed in this invention.

[0080] Figure 5 This is a schematic diagram of the modules of an integrated management system based on an artificial intelligence computing platform proposed in this invention. Detailed Implementation

[0081] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0082] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0083] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0084] Example 1:

[0085] Reference Figures 1-4 This is the first embodiment of the present invention, which provides an integrated management system based on an artificial intelligence computing platform, including the following steps:

[0086] S1. Resource monitoring and acquisition: Obtain heterogeneous computing resource status information.

[0087] In this embodiment, the resource monitoring and acquisition includes:

[0088] In integrated management, computing resources are diverse and heterogeneous, commonly including various computing units such as CPUs, GPUs, and TPUs. Each type of resource has different performance, architecture, and load conditions. Since the efficiency of task scheduling, resource allocation, and management depends on an accurate grasp of the status of each computing resource, real-time monitoring of the operational status of computing resources is a prerequisite for integrated management. A lack of real-time and accurate resource status information will lead to resource scheduling decisions based on incomplete or outdated data, increasing task execution failure rates, reducing resource utilization efficiency, and even triggering system overload or failure risks. Therefore, comprehensive and real-time monitoring and acquisition of computing resources are necessary.

[0089] The main purpose of this step is to monitor and acquire heterogeneous computing resource status information, including core indicators such as load, memory usage, temperature, and power consumption of each computing node, so as to accurately reflect the resource status, understand the current status of the resources, support the subsequent steps, detect hardware anomalies in a timely manner, prevent overload and damage, ensure the stable operation of the computing platform, help identify idle or inefficiently used resources, promote the rational allocation and scheduling of resources, and improve resource utilization efficiency.

[0090] This step involves installing monitoring and acquisition components on each heterogeneous computing node to periodically access the node's performance counters and sensor data, continuously collecting status data of local hardware resources, including core indicators such as CPU usage and idle rate, GPU usage and idle rate, memory usage, device temperature, and power consumption. The collected raw indicators undergo basic cleaning and formatting, including removing outliers, converting units, and standardizing timestamps to ensure consistent standards and comparability of data across nodes. Furthermore, to ensure consistency in time and content of status data in heterogeneous environments, a time synchronization mechanism is adopted to ensure that all collected data is accompanied by a standardized timestamp, thereby improving the accuracy and timeliness of the monitoring system.

[0091] S2. Perform a resource health assessment based on the heterogeneous computing resource status information to obtain resource availability indicators.

[0092] In this embodiment, the resource health assessment includes:

[0093] Based on the obtained real-time heterogeneous computing resource status information, a resource health assessment model is used to perform nonlinear assessment of heterogeneous computing resources by integrating the status and historical stability of each resource. This achieves a unified and standardized resource health assessment, thereby obtaining the resource availability index of each computing node and providing a basis for subsequent resource allocation and task scheduling.

[0094] Furthermore, the resource health assessment model includes:

[0095]

[0096] in:

[0097] The resource availability index of the j-th node, ranging from 0 to 1;

[0098] The Sigmoid function (existing technology) is used to normalize output;

[0099] The weighting coefficients for the current status score range from 0 to 1, and the sum of the weighting coefficients is 1.

[0100] : The weighting coefficient for historical stability score, ranging from 0 to 1;

[0101] : The weighting coefficients of the coupling terms, ranging from 0 to 1;

[0102] : The current state score of the j-th node, ranging from 0 to 1, where a larger score indicates that the node currently has more available resources;

[0103] : The historical stability score of the j-th node, ranging from 0 to 1, with a larger value indicating a more stable historical node.

[0104] Furthermore, the current state score includes:

[0105]

[0106] in:

[0107] : The CPU idle rate of the j-th node, ranging from 0 to 1, is calculated by the ratio of the number of unused CPU cores to the total number of CPU cores;

[0108] : The GPU idle rate of the j-th node, ranging from 0 to 1, is calculated by the ratio of the number of unused GPU cores to the total number of GPU cores;

[0109] : The memory free rate of the j-th node, ranging from 0 to 1, is calculated as the ratio of unused memory size to total memory size;

[0110] The weighting coefficient for CPU idle rate ranges from 0 to 1, and the sum of the weighting coefficients is 1.

[0111] The weighting factor for GPU idle rate, ranging from 0 to 1;

[0112] The weighting factor for memory free time is between 0 and 1.

[0113] Calculate the node's current status score based on the current CPU and GPU idle rates and memory idle rates.

[0114] Furthermore, the historical stability score includes:

[0115]

[0116] in:

[0117] The number of historical observation periods, such as the number of hourly samples within a 24-hour period;

[0118] : The anomaly rate of node j in the k-th historical observation period, that is, the proportion of abnormal time in this period, ranging from 0 to 1. Anomalies include downtime, disconnection, etc.

[0119] The historical stability score is obtained by calculating the normality rate of nodes in all historical observation periods.

[0120] This model describes how to obtain resource availability metrics for computing nodes based on real-time status information and historical data of heterogeneous computing resources.

[0121] Traditional resource health assessment methods typically process various indicators in a linear manner, such as weighted averages, which makes it difficult to reflect the nonlinear coupling characteristics of actual load risk. This results in insufficient rationality of the obtained resource availability indicators and can easily lead to node overload during scheduling. In contrast, this model integrates the status and historical stability of various resources, which better reflects the comprehensive health level of nodes, improves the rationality of resource availability indicators, and enhances the stability of nodes during scheduling.

[0122] The steps for using this model include:

[0123] Data Acquisition: Obtain the usage status of each node's resources from step S1, including CPU idle rate. GPU idle rate Memory free rate And obtain the anomaly rate of each observation period node from historical data. ;

[0124] Score Calculation: Calculate the current status score based on the obtained data. and historical stability score ;

[0125] Resource health assessment: Based on the calculated score, the resource health is assessed to obtain the node resource availability index for each node. This index reflects the overall health level of the node and serves as a key input for subsequent resource scheduling and task allocation, ensuring that tasks are assigned to nodes with abundant and stable available resources.

[0126] S3. Match task resources based on resource availability indicators to obtain a resource allocation plan.

[0127] In this embodiment, the task resource matching includes:

[0128] This step uses the obtained resource availability index, combined with the resource requirements of the task to be executed, to calculate the matching score between the task and the resource node using a task matching model, and then allocates the task according to the matching score, thereby achieving the optimal task allocation and obtaining a resource allocation scheme.

[0129] Furthermore, the task matching model includes:

[0130]

[0131] in:

[0132] : The matching and adaptation score between task i and the j-th node, ranging from 0 to 1. The higher the score, the higher the adaptation between the task and the node, and the more likely the task is to be assigned to this node.

[0133] : The resource requirement matching degree between task i and the j-th node, usually ranging from 0 to 1. The larger the value, the more idle resources the node has compared to the resources required by the task. When it is equal to 0, it means that the resources expected to be used by the task are equal to the idle resources of the node. When it is less than 0, it means that any expected resource usage of the task is greater than the idle resources of the node, and the task cannot be executed on this node.

[0134] : Historical compatibility factor between task i and node j, reflecting the historical success rate of task execution on that node, ranging from 0 to 1;

[0135] The weighting coefficient for resource demand matching degree ranges from 0 to 1, and the sum of the weighting coefficients is 1.

[0136] The weighting coefficient for resource availability indicators, ranging from 0 to 1;

[0137] : The weighting coefficient of the historical compatibility factor, ranging from 0 to 1.

[0138] Furthermore, the resource demand matching degree includes:

[0139]

[0140] in:

[0141] : The expected CPU utilization of task i, i.e., the minimum CPU utilization;

[0142] : The expected GPU utilization of task i, i.e., the minimum GPU utilization;

[0143] The expected memory usage of task i, i.e., the minimum memory usage;

[0144] The weight coefficient of the remaining available CPUs, ranging from 0 to 1, with the sum of the weight coefficients being 1;

[0145] The weighting coefficients of the remaining available GPUs, ranging from 0 to 1;

[0146] The weighting factor for remaining available memory, ranging from 0 to 1;

[0147] The resource requirement matching degree between nodes and tasks is calculated based on the available resources of the nodes and the resources required by the tasks.

[0148] Furthermore, the historical adaptation compatibility factor includes:

[0149]

[0150] in:

[0151] : The number of times task i was successfully executed on node j;

[0152] : The total number of times task i is scheduled on node j;

[0153] Small constants that prevent division by zero;

[0154] The historical compatibility factor between a task and a node is calculated based on the ratio of the number of successful executions of a task on a node to the total number of schedulings.

[0155] This model describes how to match tasks with nodes based on a comparison of the resources required for the task with the existing resources of the node and the historical task success rate.

[0156] Traditional task resource matching methods typically use fixed rules or thresholds to match a single resource dimension, ignoring historical compatibility and leading to uneven resource utilization, node overload, and high task failure rates. In contrast, this model comprehensively considers multiple resource dimensions when matching tasks and nodes, improving matching accuracy and resource utilization, avoiding resource waste caused by single resource bottlenecks, and responding to node resource changes in real time by combining the historical task execution status of nodes for matching, thereby improving the load balance of each node and the overall operational stability.

[0157] The steps for using the above model include:

[0158] Task requirement retrieval: Obtain the resource requirements of the task to be executed and the available resources of each node, including the estimated CPU utilization. Expected GPU utilization Expected memory usage CPU idle rate GPU idle rate Memory free rate ;

[0159] Node matching: Calculate the resource requirement matching degree between the task and the node based on the resource requirements of the task to be executed and the available resources of the node. Compatibility factor with history Furthermore, if any resource requirement of the task to be executed exceeds the node's available resources, the node cannot be scheduled to execute the task, based on the calculated resource requirement matching degree. Compatibility factor with history Combined with resource availability indicators Calculate the matching and adaptation score ;

[0160] Allocation scheme acquisition: based on the calculated matching and adaptation score. Each task to be executed is assigned a node. The higher the matching and adaptation score, the more suitable the node is for executing the task. Under the premise of ensuring that the idle resources of each node are greater than the resource requirements of the assigned task, the final resource allocation scheme is obtained.

[0161] S4. Calculate task scheduling based on the resource allocation scheme to obtain the task execution sequence.

[0162] In this embodiment, the computing task scheduling includes:

[0163] The resource allocation scheme obtained from step S4 solves the problem of task allocation on nodes. In actual execution, each node may undertake multiple tasks. If scheduling and sorting are not performed, resource conflicts or disordered task queuing will occur, affecting the overall computing efficiency. Therefore, it is also necessary to reasonably arrange the execution order and timing of each task on the node according to the characteristics and resource consumption of each task to obtain the task execution sequence. This will minimize the waiting time of tasks and the idle time of nodes while ensuring that the resources of each node do not conflict, increase the number of tasks completed per unit time, reduce resource contention and queuing conflicts during task execution, and improve system stability and resource utilization.

[0164] This step calculates the sorting weight value for tasks on each node based on the obtained resource allocation scheme. The sorting weight value is calculated by combining the estimated running time of the task (estimated based on historical data), the task start latency tolerance (determined based on whether it is a low-latency task or a batch processing task), and the task resource consumption (derived from the combined CPU, GPU, and memory resource consumption). The smaller the value, the higher the scheduling priority. The tasks on the node are sorted according to their scheduling priority to form the final task execution sequence of that node. This avoids execution blocking and resource idleness caused by resource contention, thereby improving the overall throughput and computing efficiency.

[0165] S5. Monitor task execution based on the task execution sequence to obtain the real-time execution status of the tasks.

[0166] In this embodiment, the task execution monitoring includes:

[0167] Based on the obtained task execution sequence, the actual execution status of the tasks at the nodes is dynamically monitored, and the execution status scoring model is used to quantitatively evaluate the execution of the tasks, thereby accurately reflecting the resource utilization efficiency and progress of task execution and supporting the dynamic adjustment of resource allocation.

[0168] Furthermore, the execution status scoring model includes:

[0169]

[0170] in:

[0171] The score for the execution status of task i at node j at time t reflects the health of the current task execution. The closer the value is to 1, the healthier and more stable the current task execution is.

[0172] The actual CPU utilization of task i on node j at time t is very close to the expected CPU utilization, which means that the resource allocation is accurate and the task is executing normally. The CPU utilization is less than the expected CPU utilization, which means that the resource supply is insufficient and the task execution is limited. The CPU utilization is greater than the expected CPU utilization, which means that the resource allocation is greater than the task demand, which may lead to a decrease in resource utilization efficiency.

[0173] The actual GPU utilization of task i on node j at time t is compared with the expected GPU utilization in the same way as the CPU utilization mentioned above.

[0174] The actual memory usage of task i on node j at time t is compared with the expected memory usage in the same way as the CPU usage above.

[0175] : The execution progress of task i on node j at time t, ranging from 0 to 1, where 0 represents just started execution and 1 represents completion;

[0176] , , , : Weighting coefficient, ranging from 0 to 1, with the sum of the weighting coefficients being 1.

[0177] This model describes how to quantitatively evaluate the execution of a task based on its performance on a node.

[0178] Traditional task execution monitoring often relies on single resource utilization metrics (such as CPU utilization), lacking a comprehensive evaluation that combines task requirements with the actual health of the task. It also struggles to assign appropriate weights to different resource types. This model integrates matching and adaptation scores with real-time usage data, overcoming the blind spots of single resource monitoring and improving monitoring accuracy. Furthermore, it employs multi-dimensional resource utilization and task progress factors for weighting, adapting to the differences in resource sensitivity among diverse and heterogeneous tasks and enhancing the model's adaptability.

[0179] The steps for using the model include:

[0180] Data Acquisition: Based on dynamic monitoring data of the actual execution status of tasks on nodes, the actual CPU utilization of tasks on nodes is obtained. Actual GPU utilization Actual memory usage and execution progress And obtain the task and node matching and adaptation score from step S3. ;

[0181] Execution status score calculation: Based on the obtained data, calculate the execution status score of the task on the node. This score reflects the health of the current task execution.

[0182] Execution status analysis: Based on the calculated execution status score The analysis yields a preliminary execution status. A score closer to 1 indicates accurate resource scheduling, high task utilization efficiency, and a healthier and more stable execution status. A score greater than 1 indicates resource pressure, requiring monitoring of task execution trends and dynamic adjustments. A score less than 1 indicates that resource usage is less than demand, suggesting potential underutilization of resources or insufficient task execution. A more specific execution status is determined based on the actual GPU, CPU, and memory usage. For example, if CPU and memory usage are very close to demand, while GPU usage is slightly higher than demand, it indicates that the task is in a relatively healthy running state, but GPU usage is slightly excessive and warrants attention.

[0183] S6. Adjust resources based on the real-time execution status of the task to obtain the resource allocation adjustment results.

[0184] In this embodiment, the resource adjustment includes:

[0185] In artificial intelligence computing platforms, the execution environment and resource status of computing tasks are highly dynamic. As tasks progress, resource demands and usage constantly change, and a single resource matching scheme cannot guarantee that tasks will always be in an optimal execution state. Furthermore, issues such as resource contention between tasks and fluctuations in node load arise. Failure to adjust resource allocation in real time may lead to resource waste or task performance degradation. Therefore, it is necessary to dynamically adjust resource allocation based on the real-time task execution status obtained from step S5, thereby improving resource utilization, reasonably increasing or decreasing resource quotas, ensuring smooth task execution, minimizing performance degradation or task blocking caused by insufficient resources, and improving load balancing among nodes.

[0186] This step introduces a resource adjustment prediction mechanism based on the historical execution trajectory of the task, based on the real-time execution status of the task. By storing and analyzing the historical execution data of the task in multiple rounds, it predicts the changing trends of indicator requirements, anticipates future resource needs, and realizes pre-adjustment to reduce the allocation delay of subsequent resource adjustments. Then, based on the detection results, it compares the resources used by the task in real-time execution with the allocated resources to determine whether there is a mismatch in resource allocation (judgment criteria include the actual resource utilization rate of the task being significantly lower than the allocated resources, the task progress being hindered or slowed down, etc.). Combining the task priority, resource availability, and the overall load of the platform, a resource adjustment plan is formulated (the plan may include increasing CPU cores, GPU units or memory capacity, or reclaiming over-allocated resources, etc.) to ensure that the adjustment process does not affect the current stable operation of the task. The resource adjustment is carried out according to the resource adjustment plan, and the adjustment effect is monitored in real time to confirm whether the resource adjustment has achieved the expected goal. If not, the adjustment and optimization continue, thereby realizing dynamic and refined management of task resource allocation, adapting to task load fluctuations and changes in the state of the artificial intelligence computing platform.

[0187] Example 2:

[0188] Reference Figure 5 This is the second embodiment of the present invention, which provides an integrated management system based on an artificial intelligence computing platform, including the following functional modules:

[0189] Resource Status Acquisition Module: Collects resource status information of each node on the computing platform, including the number of CPU cores, the number of GPU units, memory capacity and its usage, forming complete basic resource data to provide a basis for subsequent resource health assessment;

[0190] Resource health assessment module: Based on the collected resource status information, calculate the resource availability index for each node, reflecting the health status and actual availability of node resources;

[0191] Task resource matching module: Combines task requirements and resource health assessment results to calculate the matching degree between tasks and nodes and determine the optimal resource allocation scheme;

[0192] Resource allocation and execution module: Based on the task resource matching results, it allocates resources to each computing node, realizing the specific allocation and scheduling of resources;

[0193] Task execution monitoring module: Monitors task execution status in real time, and analyzes task execution and resource usage based on resource allocation results;

[0194] Resource dynamic adjustment module: Adjusts resources based on the real-time execution status of tasks to achieve dynamic optimization of resource allocation, ensuring system load balancing and task performance.

[0195] Example 3:

[0196] This embodiment also provides a computer device applicable to an integrated management system based on an artificial intelligence computing platform, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize an integrated management system based on an artificial intelligence computing platform as proposed in the above embodiment.

[0197] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements an integrated management system based on an artificial intelligence computing platform as proposed in the above embodiments.

[0198] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0199] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0200] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0201] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0202] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0203] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An integrated management system based on an artificial intelligence computing platform, characterized in that, Includes the following steps: S1. Resource monitoring and acquisition: Obtain heterogeneous computing resource status information; S2. Perform resource health assessment based on heterogeneous computing resource status information to obtain resource availability indicators; obtain the usage status of each node's resources from step S1, including CPU idle rate, GPU idle rate, and memory idle rate, and obtain the anomaly rate of nodes in each observation period from historical data; calculate the current status score and historical stability score based on the obtained data; resource health assessment: assess resource health based on the calculated scores to obtain the node resource availability indicator for each node, which reflects the overall health level of the node; S3. Perform task resource matching based on resource availability metrics to obtain a resource allocation scheme. Obtain the resource requirements of the tasks to be executed and the available resources of each node, including estimated CPU utilization, estimated GPU utilization, estimated memory utilization, CPU idle rate, GPU idle rate, and memory idle rate. Calculate the resource requirement matching degree and historical compatibility factor between the task and the node based on the resource requirements of the tasks to be executed and the node's idle resources. If any resource requirement of the task to be executed exceeds the node's idle resources, the node cannot be scheduled to execute the task. Calculate the matching and compatibility score based on the calculated resource requirement matching degree and historical compatibility factor, combined with resource availability metrics. Allocate nodes for each task to be executed based on the calculated matching and compatibility score. A higher matching and compatibility score indicates that the node is more suitable for executing the task. Under the premise that the idle resources of each node are greater than the resource requirements of the assigned task, obtain the final resource allocation scheme. Calculate the historical compatibility factor between the task and the node based on the ratio of the number of successful executions of the task on the node to the total number of scheduling attempts. S4. Calculate task scheduling based on the resource allocation scheme to obtain the task execution sequence; S5. Monitor task execution according to the task execution sequence to obtain the real-time execution status of the tasks; based on the dynamic monitoring data of the actual execution status of the tasks on the nodes, obtain the actual CPU utilization, actual GPU utilization, actual memory utilization and execution progress of the tasks on the nodes, and obtain the matching and adaptation score between the tasks and the nodes from step S3; Execution status score calculation: Based on the obtained data, calculate the execution status score of the tasks on the nodes. This score reflects the current health of the task execution; Based on the calculated execution status score, analyze to obtain the preliminary execution status. The closer the score is to 1, the more accurate the resource scheduling, the higher the task utilization efficiency, and the healthier and more stable the execution status. A score greater than 1 indicates that there is resource pressure, and the task running trend needs to be monitored and dynamically adjusted. A score less than 1 indicates that the resource usage is less than the demand, and there is insufficient resource utilization or the task is not fully executed. S6. Adjust resources based on the real-time execution status of the task to obtain the resource allocation adjustment results.

2. The integrated management system based on an artificial intelligence computing platform according to claim 1, characterized in that, The resource health assessment in step S2 includes: Based on the obtained real-time heterogeneous computing resource status information, a resource health assessment model is used to perform a nonlinear assessment of the heterogeneous computing resources by fusing the status and historical stability of each resource, thereby obtaining the resource availability index of each computing node.

3. The integrated management system based on an artificial intelligence computing platform according to claim 2, characterized in that, The resource health assessment model includes: ， in: : Resource availability metrics for the j-th node; : Sigmoid function; : The weighting coefficient for the current state score; : Weighting coefficients for historical stability scores; : Weighting coefficients of the coupling terms; The current state score of the j-th node; : Historical stability score of the j-th node.

4. The integrated management system based on an artificial intelligence computing platform according to claim 3, characterized in that, The current status score includes: ， in: : CPU idle rate of the j-th node; : GPU idle rate of the j-th node; : The memory free rate of the j-th node; : Weighting coefficient for CPU idle rate; : Weighting coefficient for GPU idle rate; Weighting coefficient for memory free rate; Calculate the node's current status score based on the current CPU and GPU idle rates and memory idle rates.

5. The integrated management system based on an artificial intelligence computing platform according to claim 3, characterized in that, The historical stability score includes: ， in: The number of historical observation periods; : The anomaly rate of node j in the k-th historical observation period; The historical stability score is obtained by calculating the normality rate of nodes in all historical observation periods.

6. The integrated management system based on an artificial intelligence computing platform according to claim 1, characterized in that, The task resource matching in step S3 includes: Based on the obtained resource availability indicators and the resource requirements of the tasks to be executed, a task matching model is used to calculate the matching score between tasks and resource nodes, and tasks are allocated according to the matching score to obtain a resource allocation scheme.

7. The integrated management system based on an artificial intelligence computing platform according to claim 6, characterized in that, The task matching model includes: ， in: : Matching and adaptation score between task i and the j-th node; : The degree of matching between the resource requirements of task i and the j-th node; : Historical compatibility factor between task i and the j-th node; : Weighting coefficients for resource demand matching degree; Weighting coefficients for resource availability metrics; : Weighting coefficient of historical compatibility factor.

8. The integrated management system based on an artificial intelligence computing platform according to claim 7, characterized in that, The resource demand matching degree includes: ， in: : Estimated CPU utilization for task i; : Expected GPU utilization for task i; : Estimated memory usage for task i; The weighting coefficient of the remaining available CPU; The weighting coefficients of the remaining available GPUs; Weighting factor for remaining available memory; The resource requirement matching degree between nodes and tasks is calculated based on the available resources of the nodes and the resources required by the tasks.

9. The integrated management system based on an artificial intelligence computing platform according to claim 7, characterized in that, The historical adaptation compatibility factors include: ， in: : The number of times task i was successfully executed on node j; : The total number of times task i is scheduled on node j; Small constants that prevent division by zero; The historical compatibility factor between a task and a node is calculated based on the ratio of the number of successful executions of a task on a node to the total number of schedulings.

10. The integrated management system based on an artificial intelligence computing platform according to claim 1, characterized in that, Includes the following functional templates: The module includes a resource status acquisition module, a resource health assessment module, a task resource matching module, a resource allocation and execution module, a task operation monitoring module, and a resource dynamic adjustment module.

Images