Power management method and device based on heterogeneous computing power scheduling

Abstract

The invention discloses an electric power management method and device based on heterogeneous computing power scheduling, and relates to the technical field of heterogeneous computing power, and the method comprises the steps: calculating the computing power evaluation coefficient of heterogeneous computing power and the calculation weight of each heterogeneous computing power for different electric power management tasks, and calculating the conventional weight of each heterogeneous computing power for different electric power management tasks; distributing the to-be-distributed power management task to the idle heterogeneous computing power with the maximum conventional weight, if no idle heterogeneous computing power exists, calculating the real-time weight of each heterogeneous computing power to the to-be-distributed power management task, calculating the comprehensive weight of each heterogeneous computing power to the to-be-distributed power management task, and calculating the real-time weight of each heterogeneous computing power to the to-be-distributed power management task; allocating the to-be-allocated power management task to the heterogeneous computing power with the maximum comprehensive weight; and calculating a busy coefficient of each power management task, calculating a maximum difference value of the busy coefficients among the power management tasks, recording the maximum difference value as a busy coefficient difference, and sending out a heterogeneous computing power optimization instruction. And the utilization rate of overall computing power resources is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of heterogeneous computing power, and specifically to a power management method and device based on heterogeneous computing power scheduling. Background Art

[0002] With the rapid development of cloud computing, big data, and artificial intelligence technologies, heterogeneous computing power scheduling has become an increasingly important topic. Heterogeneous computing power refers to the computing capabilities provided by different types of computing resources (such as CPUs, GPUs, FPGAs, etc.) during parallel computing. These heterogeneous computing power resources have different architectures and performance characteristics and can handle different types of computing tasks. In the field of power management, traditional management methods often ignore the diversity and complexity of computing resources, resulting in problems such as unreasonable power resource allocation and low energy efficiency. Therefore, combining heterogeneous computing power scheduling technology to achieve intelligent and efficient power management has become an important direction for current research and application.

[0003] In the Chinese invention application with the application publication number CN119149230A, a distributed heterogeneous computing power inference task dynamic scheduling method and system are disclosed, including a computing power architecture adaptation module, a task scheduling module, a model update module, a capability service module, a dynamic scheduling module, a resource evaluation module, and a status perception module; the computing power architecture adaptation module is used to identify and match the local heterogeneous computing power architecture and the remote heterogeneous computing power architecture; the task scheduling module is used to select a suitable computing power terminal from a distributed heterogeneous computing power environment; the model update module is used to update the AI model in real time; the capability service module is used to provide a unified AI model service call interface; the dynamic scheduling module is used to dynamically allocate computing power resources; the resource evaluation module is used to monitor and evaluate the status of computing power resources; the status perception module is used to monitor the health status of computing power nodes in real time and respond to faults in a timely manner; improving the scheduling efficiency of computing power resource scheduling and ensuring the stable operation of inference tasks.

[0004] In the above invention application, it is considered that device failures or computing power fluctuations may also cause interruptions or delays in inference tasks, but it is still based on the existing heterogeneous computing power for allocation and does not consider optimizing the computing power according to the task situation. When the task requirements do not match the allocated computing power, some computing power resources may be idle or inefficiently utilized. For example, if high-computing power resources are allocated to low-load tasks, these resources will not be fully utilized, resulting in waste. On the contrary, if low-computing power resources are allocated to high-load tasks, it may lead to slow task processing speeds or even inability to meet real-time requirements, thereby affecting the overall system performance.

[0005] Therefore, the present invention provides a power management method and device based on heterogeneous computing power scheduling. Summary of the Invention

[0006] (1) Technical problem to be solved

[0007] In view of the deficiencies of the prior art, the present invention provides a power management method and device based on heterogeneous computing power scheduling. The present invention obtains the average waiting duration De of each power management task from the operation records of heterogeneous computing power j , processing duration Ci j , energy consumption Xh j and total number per unit time Sg j , calculates the busyness coefficient Fm of each power management task j , and calculates the maximum difference between the busyness coefficients Fm j of each power management task, denoted as the busyness coefficient difference Cz. When the busyness coefficient difference exceeds 2ΔCz, a heterogeneous computing power optimization instruction is sent out, which can comprehensively reflect the load conditions of tasks and the tightness of computing power resources, can reveal the degree of uneven load between tasks, and the issuance of the optimization instruction helps to reallocate computing power resources, enabling high-load tasks to receive more computing power support and low-load tasks to release computing power resources, thereby improving the utilization rate of overall computing power resources, and thus solving the technical problems recorded in the background art.

[0008] (2) Technical solution

[0009] To achieve the above object, the present invention is implemented through the following technical solutions: A power management method based on heterogeneous computing power scheduling, including the following steps:

[0010] Obtain the computing power Sl i , bandwidth Dk i and computing density Sm i of each heterogeneous computing power, calculate the computing power evaluation coefficient Spg i of the heterogeneous computing power, and make each heterogeneous computing power execute the same type of power management task, use a monitoring tool to record the processing duration Cs ij and energy consumption Nu ij of the same power management task on different heterogeneous computing powers, calculate the computing weight α ij of each heterogeneous computing power for different power management tasks, and calculate the conventional weight μ i of each heterogeneous computing power for different power management tasks according to the computing power evaluation coefficient Spg ij of the heterogeneous computing power and the computing weight α ij of each heterogeneous computing power for different power management tasks;

[0011] Allocate the power management tasks to be allocated to the idle heterogeneous computing power with the largest conventional weight μ ij . If there is no idle heterogeneous computing power, obtain the waiting duration Da iCalculate the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated according to the real-time requirement level Ja of the power management task to be allocated. i According to the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated. i And the conventional weight μ of each heterogeneous computing power for the power management task to be allocated. i Calculate the comprehensive weight x of each heterogeneous computing power for the power management task to be allocated. i Allocate the power management task to be allocated to the heterogeneous computing power with the largest comprehensive weight x. i The largest heterogeneous computing power;

[0012] Obtain the average waiting time De of each power management task from the operation records of heterogeneous computing power. j The processing time Ci j The energy consumption Xh j And the total number per unit time Sg j Calculate the busy coefficient Fm of each power management task. j And calculate the maximum difference between the busy coefficients Fm of each power management task. Denote it as the busy coefficient difference Cz. When the busy coefficient difference exceeds 2ΔCz, send out an instruction for optimizing heterogeneous computing power. j

[0013] Furthermore, obtain the computing power Sl i The bandwidth Dk i And the computing density Sm i Of each heterogeneous computing power, and calculate the computing power evaluation coefficient Spg of the heterogeneous computing power. i :

[0014]

[0015] Among them, i represents the sequential number of each heterogeneous computing power, i = 1, 2,..., n, and n represents the total number of all heterogeneous computing powers; max(Sl i ) represents the maximum computing power of all heterogeneous computing powers, min(Sl i ) represents the minimum computing power of all heterogeneous computing powers, max(Dk i ) represents the maximum bandwidth of all heterogeneous computing powers, min(Dk i ) represents the minimum bandwidth of all heterogeneous computing powers, max(Sm i ) represents the maximum computing density of all heterogeneous computing powers, and min(Sm i ) represents the minimum computing density of all heterogeneous computing powers.

[0016] ​Computing power refers to the number of operations that a computing unit can perform per second, usually expressed as FLOPs (Floating Point Operations Per Second). In heterogeneous computing, the computing power of different computing units can vary greatly. For example, GPUs usually have higher parallel computing capabilities than CPUs, so they exhibit higher computing power when processing a large amount of parallel data.

[0017] Bandwidth refers to the rate of data transfer between a computing unit and memory or external devices, usually expressed as the number of bytes transferred per second (Byte / s). In heterogeneous computing, bandwidth is crucial for data transfer speed and overall system performance. High bandwidth can ensure that the computing unit can efficiently access and process data, while low bandwidth may lead to performance bottlenecks.

[0018] In heterogeneous computing, computing density refers to the computing power that can be provided under a given space or power consumption. This is usually related to the ratio of computing power to power consumption, but may also involve other factors such as heat dissipation performance and physical size. A computing unit with high computing density can provide higher computing power with limited resources.

[0019] Furthermore, make each heterogeneous computing power execute the same type of power management tasks, and use monitoring tools to record the processing duration Cs of the same power management tasks on different heterogeneous computing powers ij and the energy consumption Nu ij , and calculate the computing weight α of each heterogeneous computing power for different power management tasks ij :

[0020]

[0021] where j represents the sequential number of each power management task, j = 1, 2, …, m, and m represents the total number of all power management tasks.

[0022] Furthermore, obtain the computing power evaluation coefficient Spg of the heterogeneous computing power i and the computing weight α of each heterogeneous computing power for different power management tasks ij , and calculate the conventional weight μ of each heterogeneous computing power for different power management tasks ij :

[0023]

[0024] Furthermore, allocate the power management tasks to be assigned to the idle heterogeneous computing power with the largest conventional weight μ ij , and record the real-time processing duration in real time. Denote the difference between the average processing duration of the current heterogeneous computing power for processing the same type of power management tasks and the real-time processing duration as the waiting duration Da of the current heterogeneous computing power i .

[0025] Further, if there is no idle heterogeneous computing power, obtain the waiting duration Da of each heterogeneous computing power i and the real-time requirement level Ja of the power management task to be allocated, and calculate the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i :

[0026]

[0027] The real-time requirement level of the power management task is divided according to the type of the power management task, and the specific division is as follows:

[0028] The planning task has a low real-time requirement, and the real-time requirement level is recorded as 1. The operation monitoring has a medium real-time requirement, and the real-time requirement level is recorded as 2. The real-time scheduling has a high real-time requirement, and the real-time requirement level is recorded as 3. The fault handling has an extremely high real-time requirement, and the real-time requirement level is recorded as 4.

[0029] Further, obtain the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i and the conventional weight μ of each heterogeneous computing power for the power management task to be allocated i , and calculate the comprehensive weight x of each heterogeneous computing power for the power management task to be allocated i :

[0030] x i = ρ i * μ i

[0031] When there is no idle heterogeneous computing power, allocate the power management task to be allocated to the heterogeneous computing power with the largest comprehensive weight x i , and update the waiting duration of the current heterogeneous computing power, that is, the original waiting duration of the heterogeneous computing power + the average processing duration of the power management tasks of the same type as the power management task to be allocated.

[0032] Further, obtain the average waiting duration De of each power management task, the processing duration Ci j , the energy consumption Xh j and the total number per unit time Sg j from the operation records of the heterogeneous computing power, and calculate the busy coefficient Fm of each power management task j : j :

[0033]

[0034] Among them, max(De j ) represents the maximum value of the average waiting duration of all power management tasks, max(Ci j) represents the maximum processing duration of all power management tasks, max(Xh j ) represents the maximum energy consumption of all power management tasks, max(Sg j ) represents the maximum total number of all power management tasks per unit time.

[0035] Furthermore, obtain the busy factor Fm of each power management task j , calculate the maximum difference in the busy factor Fm between each power management task j , denoted as the busy factor difference Cz. When the busy factor difference exceeds 2ΔCz, send out an instruction for heterogeneous computing power optimization;

[0036] where ΔCz represents the average value of the busy factor differences of all heterogeneous computing power systems.

[0037] A power management device based on heterogeneous computing power scheduling includes:

[0038] A conventional weight calculation module that obtains the computing power Sl of each heterogeneous computing power i , bandwidth Dk i and computing density Sm i , calculates the computing power evaluation coefficient Spg of the heterogeneous computing power i , and makes each heterogeneous computing power execute the same type of power management task. Use a monitoring tool to record the processing duration Cs of the same power management task on different heterogeneous computing powers ij and energy consumption Nu ij , calculates the computing weight α of each heterogeneous computing power for different power management tasks ij , and based on the computing power evaluation coefficient Spg of the heterogeneous computing power i and the computing weight α of each heterogeneous computing power for different power management tasks ij , calculates the conventional weight μ of each heterogeneous computing power for different power management tasks ij ;

[0039] A computing power allocation module that allocates the power management tasks to be allocated to the idle heterogeneous computing power with the largest conventional weight μ ij . If there is no idle heterogeneous computing power, obtain the waiting duration Da of each heterogeneous computing power i and the real-time requirement level Ja of the power management task to be allocated, and calculate the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i , and based on the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i and the conventional weight μ of each heterogeneous computing power for the power management task to be allocated i , calculates the comprehensive weight x of each heterogeneous computing power for the power management task to be allocated i , and allocates the power management tasks to be allocated to the comprehensive weight xi The largest heterogeneous computing power;

[0040] The heterogeneous computing power optimization module obtains the average waiting duration De of each power management task from the operation records of the heterogeneous computing power j and the processing duration Ci j and the energy consumption Xh j and the total quantity per unit time Sg j to calculate the busy coefficient Fm of each power management task j and calculate the maximum difference between the busy coefficients Fm of each power management task j which is denoted as the busy coefficient difference Cz. When the busy coefficient difference exceeds 2ΔCz, a heterogeneous computing power optimization instruction is sent outwards.

[0041] (III) Beneficial effects

[0042] The present invention provides a power management method and device based on heterogeneous computing power scheduling, having the following beneficial effects:

[0043] 1. Obtain the computing power Sl i bandwidth Dk i and computing density Sm i of each heterogeneous computing power, calculate the computing power evaluation coefficient Spg i of the heterogeneous computing power, and make each heterogeneous computing power execute the same type of power management task, use a monitoring tool to record the processing duration Cs ij and the energy consumption Nu ij of the same power management task on different heterogeneous computing powers, calculate the computing weight α ij of each heterogeneous computing power for different power management tasks, and based on the computing power evaluation coefficient Spg i of the heterogeneous computing power and the computing weight α ij of each heterogeneous computing power for different power management tasks, calculate the conventional weight μ ij of each heterogeneous computing power for different power management tasks, which can better understand the performance of each heterogeneous computing power on different tasks, thus supporting more diverse task requirements, contributing to optimizing resource allocation when deploying tasks, and ensuring that tasks are assigned to the most suitable heterogeneous computing power for execution.

[0044] 2. Allocate the power management task to be assigned to the idle heterogeneous computing power with the largest conventional weight μ ij If there is no idle heterogeneous computing power, obtain the waiting duration Da i of each heterogeneous computing power and the real-time requirement level Ja of the power management task to be assigned, calculate the real-time weight ρ i of each heterogeneous computing power for the power management task to be assigned, and based on the real-time weight ρ iand the conventional weight μ of each heterogeneous computing power for the power management task to be allocated i Calculate the comprehensive weight x of each heterogeneous computing power for the power management task to be allocated i Allocate the power management task to be allocated to the heterogeneous computing power with the largest comprehensive weight x i Considering the current load and task urgency of the heterogeneous computing power, it helps to balance resource allocation, avoid the situation where some computing power resources are overloaded while others are idle, can flexibly handle tasks with different real-time requirements, and ensure that the system can maintain efficient operation under different load conditions.

[0045] 3. Obtain the average waiting duration De of each power management task from the operation records of the heterogeneous computing power j Processing duration Ci j Energy consumption Xh j And the total number per unit time Sg j Calculate the busy coefficient Fm of each power management task j And calculate the maximum difference between the busy coefficients Fm of each power management task j Denoted as the busy coefficient difference Cz. When the busy coefficient difference exceeds 2ΔCz, send out an instruction for optimizing the heterogeneous computing power. It can comprehensively reflect the load situation of the tasks and the tension degree of the computing power resources, can reveal the degree of load imbalance between tasks, and the issuance of the optimization instruction helps to reallocate the computing power resources, so that the high-load tasks can get more computing power support, and the low-load tasks release the computing power resources, thereby improving the overall utilization rate of the computing power resources. Brief Description of the Drawings

[0046] Figure 1 is a schematic flowchart of the power management method based on heterogeneous computing power scheduling according to the present invention;

[0047] Figure 2 is a schematic structural diagram of the power management device based on heterogeneous computing power scheduling according to the present invention. Detailed Embodiment

[0048] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0049] Please refer to Figure 1 , the present invention provides a power management method based on heterogeneous computing power scheduling, including the following steps:

[0050] Step 1. Obtain the computing power Sl of each heterogeneous computing power i Bandwidth Dki and the computing density Sm i , calculate the computing power evaluation coefficient Spg of the heterogeneous computing power i , and make each type of heterogeneous computing power execute the same type of power management task, use a monitoring tool to record the processing duration Cs of the same power management task on different heterogeneous computing powers ij and the energy consumption Nu ij , calculate the computing weight α of each type of heterogeneous computing power for different power management tasks ij , based on the computing power evaluation coefficient Spg of the heterogeneous computing power i and the computing weight α of each type of heterogeneous computing power for different power management tasks ij , calculate the conventional weight μ of each type of heterogeneous computing power for different power management tasks ij .

[0051] The first step includes the following contents:

[0052] Step 101, consult its hardware specification book or the performance parameters provided by the manufacturer to obtain the computing power Sl of each heterogeneous computing power i , bandwidth Dk i and the computing density Sm i , calculate the computing power evaluation coefficient Spg of the heterogeneous computing power i :

[0053]

[0054] where i represents the sequential number of each type of heterogeneous computing power, i = 1, 2,..., n, and n represents the total number of all heterogeneous computing powers. max(Sl i ) represents the maximum computing power of all heterogeneous computing powers, min(Sl i ) represents the minimum computing power of all heterogeneous computing powers, max(Dk i ) represents the maximum bandwidth of all heterogeneous computing powers, min(Dk i ) represents the minimum bandwidth of all heterogeneous computing powers, max(Sm i ) represents the maximum computing density of all heterogeneous computing powers, and min(Sm i ) represents the minimum computing density of all heterogeneous computing powers.

[0055] Computing power refers to the number of operations that a computing unit can perform per second, usually expressed as FLOPs (Floating Point Operations Per Second, the number of floating-point operations per second). In heterogeneous computing, the computing powers of different computing units may vary greatly. For example, GPUs usually have higher parallel computing capabilities than CPUs, so they show higher computing powers when processing a large amount of parallel data.

[0056] Bandwidth refers to the rate of data transfer between a computing unit and memory or external devices, usually expressed in bytes per second (Byte / s). In heterogeneous computing, bandwidth is crucial for data transfer speed and overall system performance. High bandwidth can ensure that the computing unit can efficiently access and process data, while low bandwidth may lead to performance bottlenecks.

[0057] In heterogeneous computing, computing density refers to the computing power that can be provided under a given space or power consumption. This is usually related to the ratio of computing power to power consumption, but may also involve other factors such as heat dissipation performance, physical size, etc. A computing unit with high computing density can provide higher computing power with limited resources.

[0058] Step 102: Have each heterogeneous computing power execute the same type of power management task, and use a monitoring tool to record the processing duration Cs of the same power management task on different heterogeneous computing powers ij and the power consumption Nu ij , and calculate the computing weight α of each heterogeneous computing power for different power management tasks ij :

[0059]

[0060] where j represents the sequential number of each power management task, j = 1, 2,..., m, and m represents the total number of all power management tasks.

[0061] Step 103: Obtain the computing power evaluation coefficient Spg of the heterogeneous computing power i and the computing weight α of each heterogeneous computing power for different power management tasks ij , and calculate the conventional weight μ of each heterogeneous computing power for different power management tasks ij :

[0062]

[0063] When in use, combine the content in Steps 101 to 103:

[0064] Obtain the computing power Sl i 、bandwidth Dk i and computing density Sm i of each heterogeneous computing power, calculate the computing power evaluation coefficient Spg i of the heterogeneous computing power, and have each heterogeneous computing power execute the same type of power management task, and use a monitoring tool to record the processing duration Cs ij and the power consumption Nu ij of the same power management task on different heterogeneous computing powers, calculate the computing weight α ij of each heterogeneous computing power for different power management tasks, and based on the computing power evaluation coefficient Spg of the heterogeneous computing poweri and the computing weights α of each heterogeneous computing power for different power management tasks ij , calculate the conventional weight μ of each heterogeneous computing power for different power management tasks ij , which can better understand the performance of each heterogeneous computing power in different tasks, thus supporting more diverse task requirements, helping to optimize resource allocation when deploying tasks, and ensuring that tasks are assigned to the most suitable heterogeneous computing power for execution.

[0065] Step 2: Allocate the power management task to be allocated to the heterogeneous computing power with the largest conventional weight μ ij If there is no idle heterogeneous computing power, obtain the waiting duration Da of each heterogeneous computing power i and the real-time requirement level Ja of the power management task to be allocated, and calculate the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i , based on the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i and the conventional weight μ of each heterogeneous computing power for the power management task to be allocated i , calculate the comprehensive weight x of each heterogeneous computing power for the power management task to be allocated i , and allocate the power management task to be allocated to the heterogeneous computing power with the largest comprehensive weight x i .

[0066] The said Step 2 includes the following contents:

[0067] Step 201: Allocate the power management task to be allocated to the heterogeneous computing power with the largest conventional weight μ ij , and record the real-time processing duration in real time. Denote the difference between the average processing duration of the current heterogeneous computing power for processing the same type of power management task and the real-time processing duration as the waiting duration Da of the current heterogeneous computing power i .

[0068] Step 202: If there is no idle heterogeneous computing power, obtain the waiting duration Da of each heterogeneous computing power i and the real-time requirement level Ja of the power management task to be allocated, and calculate the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i :

[0069]

[0070] The real-time requirement level of the power management task is divided according to the type of the power management task, and the specific division is as follows:

[0071] The planning task has low real-time requirements, and the real-time requirement level is recorded as 1. The operation monitoring has medium real-time requirements, and the real-time requirement level is recorded as 2. The real-time scheduling has high real-time requirements, and the real-time requirement level is recorded as 3. The fault handling has extremely high real-time requirements, and the real-time requirement level is recorded as 4.

[0072] Step 203: Obtain the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i and the conventional weight μ of each heterogeneous computing power for the power management task to be allocated i , and calculate the comprehensive weight x of each heterogeneous computing power for the power management task to be allocated i :

[0073] x i = ρ i * μ i

[0074] When there is no idle heterogeneous computing power, allocate the power management task to be allocated to the heterogeneous computing power with the largest comprehensive weight x i , and update the waiting duration of the current heterogeneous computing power, that is, the original waiting duration of the heterogeneous computing power + the average processing duration of the same type of power management task for processing the power management task to be allocated.

[0075] When in use, combine the content in Steps 201 to 203:

[0076] Allocate the power management task to be allocated to the idle heterogeneous computing power with the largest conventional weight μ ij . If there is no idle heterogeneous computing power, obtain the waiting duration Da of each heterogeneous computing power i and the real-time requirement level Ja of the power management task to be allocated, and calculate the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i , and based on the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i and the conventional weight μ of each heterogeneous computing power for the power management task to be allocated i , calculate the comprehensive weight x of each heterogeneous computing power for the power management task to be allocated i , and allocate the power management task to be allocated to the heterogeneous computing power with the largest comprehensive weight x i . Considering the current load and task urgency of the heterogeneous computing power, it helps to balance resource allocation, avoid the situation where some computing power resources are overloaded while other resources are idle, and can flexibly handle tasks with different real-time requirements, ensuring that the system can maintain efficient operation under different load conditions.

[0077] Step 3: Obtain the average waiting duration De, processing duration Ci j , and energy consumption Xh of each power management task from the operation records of the heterogeneous computing power j jand the total quantity per unit time Sg j , calculate the busy factor Fm of each power management task j , and calculate the maximum difference between the busy factors Fm of each power management task j . Denote it as the busy factor difference Cz. When the busy factor difference exceeds 2ΔCz, send out an instruction for heterogeneous computing power optimization

[0078] The third step includes the following contents:

[0079] Step 301: Obtain the average waiting duration De of each power management task from the operation records of heterogeneous computing power j , processing duration Ci j , energy consumption Xh j and the total quantity per unit time Sg j , calculate the busy factor Fm of each power management task j :

[0080]

[0081] where, max(De j ) represents the maximum value of the average waiting durations of all power management tasks, max(Ci j ) represents the maximum value of the processing durations of all power management tasks, max(Xh j ) represents the maximum value of the energy consumptions of all power management tasks, max(Sg j ) represents the maximum value of the total quantities per unit time of all power management tasks

[0082] Step 302: Obtain the busy factor Fm of each power management task j , calculate the maximum difference between the busy factors Fm of each power management task j . Denote it as the busy factor difference Cz. When the busy factor difference exceeds 2ΔCz, send out an instruction for heterogeneous computing power optimization

[0083] where, ΔCz represents the average value of the busy factor differences of all heterogeneous computing power systems

[0084] When in use, combine the contents in steps 301 and 302:

[0085] Obtain the average waiting duration De of each power management task from the operation records of heterogeneous computing power j , processing duration Ci j , energy consumption Xh j and the total quantity per unit time Sg j , calculate the busy factor Fm of each power management task j , and calculate the maximum difference between the busy factors Fm of each power management task jThe maximum difference is denoted as the busy coefficient difference Cz. When the busy coefficient difference exceeds 2ΔCz, an instruction for optimizing heterogeneous computing power is sent outwards. It can comprehensively reflect the load situation of tasks and the tightness of computing power resources, reveal the degree of uneven load among tasks, and the issuance of the optimization instruction helps to reallocate computing power resources, enabling high-load tasks to receive more computing power support while low-load tasks release computing power resources, thus improving the overall utilization rate of computing power resources.

[0086] Please refer to Figure 2 , the present invention provides a power management device based on heterogeneous computing power scheduling, including:

[0087] A conventional weight calculation module, which obtains the computing power Sl i 、bandwidth Dk i and computing density Sm i of each heterogeneous computing power, calculates the computing power evaluation coefficient Spg i of the heterogeneous computing power, and makes each type of heterogeneous computing power execute the same type of power management task. A monitoring tool is used to record the processing duration Cs ij and energy consumption Nu ij of the same power management task on different heterogeneous computing powers, calculates the computing weight α ij of each type of heterogeneous computing power for different power management tasks, and calculates the conventional weight μ i of each type of heterogeneous computing power for different power management tasks based on the computing power evaluation coefficient Spg ij of the heterogeneous computing power and the computing weight α ij of each type of heterogeneous computing power for different power management tasks.

[0088] A computing power allocation module, which allocates the power management task to be allocated to the idle heterogeneous computing power with the largest conventional weight μ ij . If there is no idle heterogeneous computing power, the waiting duration Da i of each heterogeneous computing power and the real-time requirement level Ja i of the power management task to be allocated are obtained, and the real-time weight ρ i of each type of heterogeneous computing power for the power management task to be allocated is calculated. Based on the real-time weight ρ i of each type of heterogeneous computing power for the power management task to be allocated and the conventional weight μ i of each type of heterogeneous computing power for the power management task to be allocated, the comprehensive weight x i of each type of heterogeneous computing power for the power management task to be allocated is calculated, and the power management task to be allocated is allocated to the heterogeneous computing power with the largest comprehensive weight x.

[0089] A heterogeneous computing power optimization module, which obtains the average waiting duration De j 、processing duration Ci j 、energy consumption Xhj and the total quantity per unit time Sg j , calculate the busy factor Fm of each power management task j , and calculate the maximum difference between the busy factors Fm of each power management task j , denoted as the busy factor difference Cz. When the busy factor difference exceeds 2ΔCz, send out a heterogeneous computing power optimization instruction externally.

[0090] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution.

[0091] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0092] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application.

Claims

1. A power management method based on heterogeneous computing power scheduling, characterized in that: including the following steps: Obtain the computing power Sl of each heterogeneous computing power i , bandwidth Dk i and computing density Sm i , calculate the computing power evaluation coefficient Spg of the heterogeneous computing power i , and make each heterogeneous computing power execute the same type of power management task, use a monitoring tool to record the processing duration Cs of the same power management task on different heterogeneous computing powers ij and energy consumption Nu ij , calculate the computing weight α of each heterogeneous computing power for different power management tasks ij , based on the computing power evaluation coefficient Spg of the heterogeneous computing power i and the computing weight α of each heterogeneous computing power for different power management tasks ij , calculate the conventional weight μ of each heterogeneous computing power for different power management tasks ij ; Allocate the power management task to be allocated to the conventional weight μ ij The maximum idle heterogeneous computing power. If there is no idle heterogeneous computing power, obtain the waiting duration Da of each heterogeneous computing power i And the real-time requirement level Ja of the power management task to be allocated, calculate the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i According to the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i And the conventional weight μ of each heterogeneous computing power for the power management task to be allocated i Calculate the comprehensive weight x of each heterogeneous computing power for the power management task to be allocated i Allocate the power management task to be allocated to the comprehensive weight x i The maximum heterogeneous computing power; Obtain the average waiting duration De of each power management task from the operation records of heterogeneous computing power j , processing duration Ci j , energy consumption Xh j and the total number per unit time Sg j , calculate the busyness factor Fm of each power management task j , and calculate the maximum difference between the busyness factors Fm j between each power management task, denoted as the busyness factor difference Cz. When the busyness factor difference exceeds 2ΔCz, send out heterogeneous computing power optimization instructions outwardly.

2. The power management method based on heterogeneous computing power scheduling according to claim 1, wherein: Obtain the computing power Sl of each heterogeneous computing power i , bandwidth Dk i and computing density Sm i , and calculate the computing power evaluation coefficient Spg of heterogeneous computing power i : where i represents the sequential number of each heterogeneous computing power, i = 1, 2, …, n, and n represents the total number of all heterogeneous computing powers; max(Sl i ) represents the maximum computing power of all heterogeneous computing powers, min(Sl i ) represents the minimum computing power of all heterogeneous computing powers, max(Dk i ) represents the maximum bandwidth of all heterogeneous computing powers, min(Dk i ) represents the minimum bandwidth of all heterogeneous computing powers, max(Sm i ) represents the maximum computing density of all heterogeneous computing powers, and min(Sm i ) represents the minimum computing density of all heterogeneous computing powers.

3. The power management method based on heterogeneous computing power scheduling according to claim 1, wherein: Make each heterogeneous computing power execute the same type of power management task, and use a monitoring tool to record the processing duration Cs of the same power management task on different heterogeneous computing powers ij and the energy consumption Nu ij , and calculate the computing weight α of each heterogeneous computing power for different power management tasks ij : where j represents the sequential number of each power management task, j = 1, 2,..., m, and m represents the total number of all power management tasks.

4. The power management method based on heterogeneous computing power scheduling according to claim 1, wherein: Obtain the computing power evaluation coefficient Spg of heterogeneous computing power i and the computing weight α of each heterogeneous computing power for different power management tasks ij , and calculate the conventional weight μ of each heterogeneous computing power for different power management tasks ij :

5. The power management method based on heterogeneous computing power scheduling according to claim 1, wherein: Allocate the power management tasks to be assigned to the regular weight μ ij The maximum idle heterogeneous computing power, and record the real-time processing duration in real time. Denote the difference between the average processing duration of the current heterogeneous computing power for processing the same type of power management tasks and the real-time processing duration as the waiting duration Da of the current heterogeneous computing power i .

6. The power management method based on heterogeneous computing power scheduling according to claim 1, wherein: If there is no idle heterogeneous computing power, obtain the waiting duration Da of each heterogeneous computing power i and the real-time requirement level Ja of the power management task to be allocated, and calculate the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i : The real-time requirement levels of power management tasks are divided according to the types of power management tasks, and the specific division is as follows: The planning task has a low real-time requirement, and the real-time requirement level is denoted as 1; the operation monitoring has a medium real-time requirement, and the real-time requirement level is denoted as 2; the real-time scheduling has a high real-time requirement, and the real-time requirement level is denoted as 3; the fault handling has an extremely high real-time requirement, and the real-time requirement level is denoted as 4.

7. The power management method based on heterogeneous computing power scheduling according to claim 1, wherein: Obtain the real-time weight ρ of each heterogeneous computing power for the power management task to be allocated i and the conventional weight μ of each heterogeneous computing power for the power management task to be allocated i , and calculate the comprehensive weight x of each heterogeneous computing power for the power management task to be allocated i : x i = ρ i * μ i When there is no idle heterogeneous computing power, allocate the power management tasks to be assigned to the comprehensive weight x i the largest heterogeneous computing power, and update the current waiting duration of the heterogeneous computing power, that is, the original waiting duration of the heterogeneous computing power + the average processing duration of the same type of power management tasks for processing the power management tasks to be assigned.

8. The power management method based on heterogeneous computing power scheduling according to claim 1, wherein: Obtain the average waiting duration De of each power management task from the operation records of heterogeneous computing power j , processing duration Ci j , energy consumption Xh j and total quantity per unit time Sg j , and calculate the busy factor Fm of each power management task j :[[]]END]] Among them, max(De j ) represents the maximum average waiting duration of all power management tasks, max(Ci j ) represents the maximum processing duration of all power management tasks, max(Xh j ) represents the maximum energy consumption of all power management tasks, and max(Sg j ) represents the maximum total number per unit time of all power management tasks.

9. The power management method based on heterogeneous computing power scheduling according to claim 1, wherein: Obtain the busy factor Fm of each power management task j , calculate the maximum difference in the busy factor Fm between each power management task j , denoted as the busy factor difference Cz. When the busy factor difference exceeds 2ΔCz, send out a heterogeneous computing power optimization instruction externally; where ΔCz represents the mean value of the differences in the busy coefficients of all heterogeneous computing power systems.

10. A power management device based on heterogeneous computing power scheduling, which is used to implement the method described in any one of claims 1 to 9, and is characterized in that: including: A conventional weight calculation module obtains the computing power Sl of each heterogeneous computing power i , the bandwidth Dk i and the computing density Sm i , calculates the computing power evaluation coefficient Spg of the heterogeneous computing power i , and enables each type of heterogeneous computing power to execute the same type of power management task, uses a monitoring tool to record the processing duration Cs of the same power management task on different heterogeneous computing powers ij and the energy consumption Nu ij , calculates the computing weight α of each type of heterogeneous computing power for different power management tasks ij , based on the computing power evaluation coefficient Spg of the heterogeneous computing power i and the computing weight α of each type of heterogeneous computing power for different power management tasks ij , calculates the conventional weight μ of each type of heterogeneous computing power for different power management tasks ij ; The computing power allocation module allocates the to-be-allocated power management task to the conventional weight μ ij The maximum idle heterogeneous computing power. If there is no idle heterogeneous computing power, obtain the waiting duration Da of each heterogeneous computing power i And the real-time requirement level Ja of the to-be-allocated power management task, calculate the real-time weight ρ of each heterogeneous computing power for the to-be-allocated power management task i According to the real-time weight ρ of each heterogeneous computing power for the to-be-allocated power management task i And the conventional weight μ of each heterogeneous computing power for the to-be-allocated power management task i Calculate the comprehensive weight x of each heterogeneous computing power for the to-be-allocated power management task i Allocate the to-be-allocated power management task to the comprehensive weight x i The maximum heterogeneous computing power; The heterogeneous computing power optimization module obtains the average waiting duration De of each power management task from the operation records of heterogeneous computing power j , processing duration Ci j , energy consumption Xh j and the total quantity per unit time Sg j , calculates the busyness coefficient Fm of each power management task j , and calculates the maximum difference between the busyness coefficients Fm j of each power management task, denoted as the busyness coefficient difference Cz. When the busyness coefficient difference exceeds 2ΔCz, a heterogeneous computing power optimization instruction is sent outwards.

Images