CN-121997581-A - Power consumption temperature digital twin modeling application method and system for computing power system associated with transmission

CN121997581ACN 121997581 ACN121997581 ACN 121997581ACN-121997581-A

Abstract

The invention relates to a calculation transmission-associated power consumption temperature digital twin modeling application method and a calculation transmission-associated power consumption temperature digital twin modeling system, wherein the calculation transmission-associated power consumption temperature digital twin modeling application system is used for executing a plurality of stages including autonomous digital modeling, digital simulation verification and application prediction, and the digital twin model is used for establishing a mapping relation among a software operator running state, a data interaction state and a hardware physical state and is used for simulating, evaluating and predicting the power consumption and the temperature state of a power calculation system. According to the invention, by establishing the precise digital mapping between the software running microscopic behavior and the hardware physical macroscopic state, the soft-to-hard efficacy simulation optimization is realized in the hardware design stage, and the proactive closed-loop control is realized in the system running stage.

Inventors

JIANG CHENG
CHEN HAIRONG
CHAI JUN
He yalong

Assignees

中国船舶集团有限公司第七一五研究所

Dates

Publication Date: 20260508
Application Date: 20260119

Claims (9)

1. A computing transmission-associated power system power consumption temperature digital twin modeling application system, comprising: the computing force interface probe module is used for acquiring an operator called by the computing force software in running and the computing size data of the operator on a computing node; the hardware probe module is used for collecting physical state data of the computing node in real time, wherein the physical state data comprise power consumption, temperature, voltage, current and external environment parameters; the data transmission probe module is used for acquiring the size information of the data blocks of the data transmission between the computing nodes; The digital twin mapping module is used for receiving and fusing the data acquired by the power interface probe module, the hardware probe module and the data transmission probe module so as to construct and operate a digital twin model of the power system; The system is used for executing a plurality of stages including autonomous digital modeling, digital simulation verification and application prediction, and the digital twin model establishes a mapping relation among a software operator running state, a data interaction state and a hardware physical state and is used for simulating, evaluating and predicting the power consumption and temperature states of the computing power system.
2. The computing transmission-associated power system power consumption temperature digital twin modeling application system of claim 1, wherein the digital twin mapping module is configured to, when the system is operating in a digital simulation verification phase: Receiving simulation data acquired by the computing force interface probe module and the data transmission probe module in a simulation environment; Inputting the simulation data, preset simulation environment parameters and simulation hardware state parameters into the digital twin model; and outputting a simulation result, wherein the simulation result comprises system power consumption, temperature and hardware state of a simulation stage and is used for generating an optimization suggestion aiming at a computing power software or hardware architecture.
3. The computing transmission-associated power system power consumption temperature digital twin modeling application system of claim 1, wherein the digital twin mapping module is specifically configured to: in an autonomous digital modeling stage, based on the data collected by the power interface probe module and the hardware probe module, an operator power consumption library is established, and the operator power consumption library is associated with dynamic power consumption of different operators under different calculation sizes; based on the data collected by the data transmission probe module and the hardware probe module, a data transmission power consumption library is established, and the data transmission power consumption library is associated with dynamic power consumption required by different data sizes transmitted between different nodes; Based on the data collected by the hardware probe module in different environments, a node hardware model is established, and the node hardware model describes dynamic relations among node power consumption, temperature, voltage, current and external environment parameters.
4. The computing transmission-associated power system power consumption temperature digital twin modeling application system of claim 3, wherein the digital twin mapping module is further configured to: Constructing a calculation power consumption model according to the operator power consumption library, and calculating corresponding power consumption according to the calling condition of the operator; constructing a transmission power consumption model according to the data transmission power consumption library, and calculating corresponding power consumption according to the data interaction condition; And fusing the calculation power consumption model, the transmission power consumption model and the node hardware model to form a system-level digital twin model capable of describing the state of the whole computing system.
5. The computing transmission-associated power system power consumption temperature digital twin modeling application system of claim 1, wherein the digital twin mapping module is configured to, when the system is running in an application phase: Receiving the operation time data collected by the power calculation interface probe module, the hardware probe module and the data transmission probe module in real time; Inputting the runtime data into the digital twin model to predict a system hardware state after a specified time in the future, the hardware state including temperature, voltage, and current; based on the prediction result, outputting an instruction for adjusting external power supply and heat dissipation measures or system abnormality early warning information.
6. A method of computing transmission-related power consumption temperature digital twin modeling application for a power system, the method comprising the steps of: the method comprises the steps of 1, an autonomous digital modeling stage, namely collecting test data of a computing power system through three types of probe modules, and constructing a digital twin model comprising an operator power consumption library, a data transmission power consumption library and a node hardware model; A stage 2, a digital simulation verification stage, in which in a simulation environment, simulated software operation data and environment parameters are input into the digital twin model, and states of the computing power system under different scenes are simulated and evaluated to form optimization suggestions; And 3, in the application stage, when the computing system actually operates, the hardware and software state data acquired in real time are input into the digital twin model, the future hardware state of the system is predicted, and the operation parameter adjustment or abnormal early warning is carried out according to the future hardware state.
7. The application method for digital twin modeling of power consumption temperature of a computing transmission-related power system according to claim 6, wherein in the autonomous digital modeling stage, the specific operations are as follows: step 1.1, running an operator test program, wherein during test, a calculation force interface probe and a hardware probe acquire operators At each computing node Performing different calculation sizes thereon Dynamic power consumption during time to form operator power consumption library ; Step 1.2, running a data interaction test program, wherein during the test, a data transmission interface probe and a hardware probe acquire different data sizes transmitted between nodes The required dynamic power consumption forms a data transmission power consumption library ; Step 1.3, running power consumption test programs under different environments, and obtaining node temperatures and power consumption under different environment parameters by a hardware probe to form a node Hardware model of (2) Wherein Representing the power consumption of the node, after inputting the current temperature and power consumption, Computable node For a period of time The hardware state information of the power calculation node; step 1.4, the digital twin mapping module is used for carrying out power consumption library according to operators Power consumption library for data transmission Sum node hardware model Establishing mapping association among a software running state, a data interaction state and a hardware state to form a digital twin model of the computing system; Wherein a power consumption model is calculated And a transmission power consumption model The calculation mode is as follows Calculating power consumption model by node And a transmission power consumption model Node hardware model After combination, a digital twin model is obtained The formula is as follows Imputation force systems are common Calculating nodes, and then calculating the digital twin model of the power system The following are listed below 。
8. The application method for digital twin modeling of power consumption temperature of computing and transmission-related power system according to claim 7, wherein the digital simulation verification stage comprises the following specific operations: Step 2.1, running the force software to be measured and calculated in a digital simulation environment, wherein the operator and the size which are called are dynamically changed in the running process of the software, and the force calculation interface probe and the data transmission interface probe acquire the execution size of each operator at each moment Data block transmission size ; Step 2.2, will 、 Simulating environmental parameters And computing node Is a simulated temperature of (2) Substituting the digital twin model of the power system in the step 1.4 to obtain the power consumption of the calculation consumption in the simulation stage Power consumption for transmission consumption And the state of the computing system Step 2.3, after one test verification is completed, resetting parameters such as different simulation environment parameters, node simulation temperature and the like, running other power calculation software, repeating the steps 2.1 to 2.2, carrying out next round of simulation verification, evaluating the influence of different environments and software on a power calculation system, and forming a hardware and software optimization suggestion of the system; And 2.4, repeating the stage 1 and the stage 2 when the hardware composition or configuration of the computing system is changed, reforming a digital twin model of the computing system and carrying out digital simulation verification.
9. The application method for calculating and transmitting the associated power consumption temperature digital twin modeling of the power system according to claim 8, wherein the application stage comprises the following specific operations: step 3.1, after the power system is started, the power interface probe and the data transmission interface probe acquire the execution sizes of operators at all times of all nodes Data block transmission size Hardware probe module acquisition power consumption Self temperature And external environmental parameters ; Step 3.2, substituting the data measured in the step 3.1 into a digital twin model of the power system Any time can be predicted The hardware state of the system is obtained; And 3.3, adjusting external power supply and heat dissipation measures in advance according to the predicted system hardware information including temperature, voltage and current, so that the system always works in an optimal environment, and early warning in advance and providing power calculation software with abnormal energy consumption if the predicted result shows that the temperature, voltage or current state is abnormal.

Description

Power consumption temperature digital twin modeling application method and system for computing power system associated with transmission Technical Field The invention belongs to the technical field of computer power, and particularly relates to a power consumption temperature digital twin modeling application method and system of a power system associated with calculation and transmission. Background With the increase of business demands in the information technology field, demands of various application software, simulation software and data processing software on computing power are continuously improved. In order to meet the service requirements, computer architecture technology and chip technology continue to advance, the types of chips are increased, the composition of the computing system is increasingly complex, the required power consumption is greatly improved, the proper temperature can ensure that hardware works in an efficient state, and the performances of the computing system and running software thereof can be fully exerted. In the calculation process of the power calculation software, corresponding calculation is dynamically completed by calling various operators and data transmission, but the utilization rate of hardware resources by various power calculation software is different under different running and data interaction states, running power consumption can change in real time, so that the temperature change of the system can also change, high dynamic change power consumption and high temperature mean that the requirements of the power calculation system on heat dissipation, power supply and environmental temperature are more strict, and challenges are also presented to the architecture design of the power calculation system. If the influence of the operation of the computing power software on the hardware architecture can be simulated in advance during the design, a more reliable hardware architecture can be designed, and if the power consumption temperature change can be predicted in advance during the operation, the operation parameters of the system can be adjusted in advance during the operation of the system, so that the computing power system always operates in an optimal state. However, the mapping method between the running state of the existing less power-calculating software and the physical state of the power-calculating system cannot achieve the above effects. Disclosure of Invention The invention aims to solve the technical problem of providing a calculation transmission-related power consumption temperature digital twin modeling application method and system of a power system, which are used for realizing soft-to-hard performance simulation optimization in a hardware design stage and realizing the proactive active closed-loop control in a system operation stage by establishing accurate digital mapping between software operation microscopic behaviors and hardware physical macroscopic states. The technical solution of the invention is to provide a power consumption temperature digital twin modeling application system of a computing power system for computing transmission association, comprising: the computing force interface probe module is used for acquiring an operator called by the computing force software in running and the computing size data of the operator on a computing node; the hardware probe module is used for collecting physical state data of the computing node in real time, wherein the physical state data comprise power consumption, temperature, voltage, current and external environment parameters; the data transmission probe module is used for acquiring the size information of the data blocks of the data transmission between the computing nodes; The digital twin mapping module is used for receiving and fusing the data acquired by the power interface probe module, the hardware probe module and the data transmission probe module so as to construct and operate a digital twin model of the power system; The system is used for executing a plurality of stages including autonomous digital modeling, digital simulation verification and application prediction, and the digital twin model establishes a mapping relation among a software operator running state, a data interaction state and a hardware physical state and is used for simulating, evaluating and predicting the power consumption and temperature states of the computing power system. The invention can simulate the hardware state by digital twin technology, evaluate the influence of software on the hardware state in advance, or monitor and predict the system state in real time when the system is running, provide data basis for the dynamic adjustment of the system hardware operation parameters, adjust external power supply or heat dissipation measures, and fully exert the system performance. The main factors influencing the hardware power consumption temperature of the computing system comprise the use condition of