Cooling circulation system of multi-core processor and control method thereof

Abstract

The invention discloses a control method for a cooling circulation system of a multi-core processor, and relates to the technical field of computer hardware heat dissipation, and the control method comprises the following steps: the multi-core processor is responsible for executing a calculation task; the sensor module monitors the temperature of each core; the cooling system control module is used for receiving data from the temperature sensor and the load monitoring module, calculating the cooling requirement of each core according to the data and commanding a cooling unit to adjust; the cooling unit is used for adjusting cooling equipment to reduce the temperature of the processor according to an instruction of the control unit; and the debugging interface is used for monitoring the state of the system and displaying the temperature, the load and the cooling resource distribution condition of each core. The method has the beneficial effects that the rotating speed of the fan, the liquid cooling flow and the working state of the heat pipe system can be accurately controlled by combining the accurate analysis of the load change and the temperature data, the heat dissipation efficiency of the processor is improved, and the energy waste is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of computer hardware heat dissipation, and specifically to a cooling circulation system for a multi-core processor and its control method. Background Art

[0002] With the continuous development of computer processor technology, especially in the field of multi-core processors, the performance and computing power of processors have been gradually improved. Modern multi-core processors usually include multiple computing cores, and each core can execute tasks independently, thereby improving the parallel computing efficiency. However, when a multi-core processor performs high-load tasks, the temperature of each core will rise rapidly. If the generated heat cannot be dissipated in time and effectively, it may lead to a decline in processor performance or even damage. Therefore, how to effectively manage the temperature of the processor core, especially under high load, is a major challenge in the design of multi-core processors.

[0003] Currently, the cooling systems of multi-core processors mainly rely on traditional air-cooling systems, liquid-cooling systems or heat pipe systems. These cooling systems usually use cooling devices such as heat sinks and fans to reduce the temperature of the processor. However, these traditional cooling solutions often lack intelligent adaptability to load changes, and the allocation of cooling resources is usually preset and not dynamically adjusted according to the actual workload and temperature status of the processor core. In addition, the response speed of traditional cooling systems is relatively slow, and there are often situations of overheating or over-cooling, resulting in low system operation efficiency. Therefore, how to dynamically adjust the cooling strategy in real time according to the core load change while ensuring the stability of the processor has become an urgent problem to be solved in the current cooling technology. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are: existing multi-core processor cooling technologies face problems such as insufficient intelligent control and inaccurate cooling resource allocation.

[0006] To solve the above technical problems, the present invention provides the following technical solution: A control method for a cooling circulation system of a multi-core processor, including

[0007] A multi-core processor, which is responsible for executing computing tasks. Each core has a different load and generates different degrees of heat; the workload affects the temperature change and provides real-time monitoring data for the temperature sensor and the cooling system; it is connected to the temperature sensor through a dedicated bus to transmit the temperature information of each core; the sensor is placed in the key heat source areas of the processor;

[0008] A sensor module, including a temperature sensor and a load monitoring module; the temperature sensor includes installing temperature sensors at each core and its surrounding area to monitor the temperature of each core;

[0009] A cooling system control module, including a control unit and a scheduling module, responsible for receiving data from the temperature sensor and the load monitoring module, calculating the cooling requirements of each core according to the data, and instructing the cooling unit to make adjustments; the control unit judges the load and temperature status of each core; the scheduling module performs feedback regulation and optimizes the decision-making;

[0010] A cooling unit, connected to the control unit through a circuit, and adjusts the cooling device according to the instruction of the control unit to reduce the temperature of the processor;

[0011] A debugging interface provides status monitoring of the system, displays the temperature, load, and cooling resource allocation of each core; connects to the control unit through a display control interface, real-time displays the parameters of the system, and allows users to intervene and adjust.

[0012] As a preferred solution of the cooling cycle system of the multi-core processor described in the present invention, wherein: the cooling system control module includes that the control unit is connected to the multi-core processor, the temperature sensor, the load monitoring module, and the cooling unit, and the control unit communicates with each module through a data bus or protocol; the scheduling module predicts the heat change of different cores of the processor according to the multi-core load information and historical data, and performs intelligent scheduling to optimize the cooling resource allocation; connects to the control unit through a data transmission interface, the scheduling module analyzes the load and temperature data and feeds back the results to the control unit to optimize the cooling strategy.

[0013] As a preferred solution of the cooling cycle system of the multi-core processor described in the present invention, wherein: the monitoring of the temperature of each core includes using the built-in monitoring module of the processor to collect the computing load data of each core, and dynamically tracking the working status of each core through a software interface;

[0014] Set a fixed time interval, and trigger the reading operation regularly through an interrupt mechanism as needed; the temperature sensor monitors the temperature of each core and feeds it back to the control unit, and is connected to the control unit through a signal line to transmit real-time temperature data; the load monitoring module monitors the working load of the multi-core processor in real time, including the computing load, thread activity, and memory bandwidth of each core;

[0015] The temperature sensor is measured using a thermistor, a thermocouple, or an integrated sensor, and is installed close to the core, divided into internal installation and external installation. The internal installation includes integrating the sensor into the processor chip, and the external installation includes installing it at the heat sink part for measurement;

[0016] The temperature sensor outputs the measured temperature information as a digital signal; for an analog signal sensor, it is converted into a digital signal through an analog-to-digital converter.

[0017] As a preferred solution of the cooling cycle system of the multi-core processor described in the present invention, wherein: the judging the load and temperature status of each core includes estimating the heat generation of each core according to the core load and working frequency;

[0018] The heat generation of the core is proportional to the product of its working load and frequency, and the power consumption calculation formula is expressed as:

[0019] P core (t) = C v ·f(t) 2 ·V 2 ·L(t)

[0020] Wherein, P core (t) represents the power consumption of each core at time t; C v represents the capacitance coefficient, which is related to the process of the processor; f(t) represents the frequency of the core at time t; V represents the working voltage; L(t) represents the load factor of the processor core, with a range of 0 to 1;

[0021] Use the mathematical model between temperature and power consumption to predict the temperature change of each core in the future for a period of time; assume that there is a linear relationship between the temperature change of the core and its power consumption, and use the mathematical model of temperature and power consumption to describe the change of the core temperature:

[0022] T core (t + 1) = T core (t) + α·(P core (t) - P ambient )

[0023] Wherein, T core (t) represents the core temperature at time t; P ambient represents the influence factor of the ambient temperature; α represents the temperature change coefficient, which represents the influence of power consumption on the core temperature; use the linear coefficient α to characterize the heat dissipation efficiency and the ambient factor, and apply a data smoothing algorithm to eliminate the instantaneous error caused by load fluctuations to obtain an accurate heat prediction;

[0024] The control unit sends the analysis result and the prediction result to the scheduling module for dynamic scheduling feedback, and establishes a cooling priority according to the load and the predicted temperature data.

[0025] As a preferred solution of the control method of the cooling cycle system of the multi-core processor described in the present invention, wherein: the adjusting the cooling device to reduce the temperature of the processor includes air-cooling system adjustment, liquid-cooling system adjustment and heat pipe adjustment;

[0026] Air-cooling system regulation includes adjusting the fan speed according to the core temperature data and adjusting the air volume in real time; the fan speed is regulated by a PWM signal to balance noise and heat dissipation effect; liquid-cooling system regulation includes adjusting the flow rate of the pump in the liquid-cooling system to change the coolant flow rate; a temperature-controlled valve is used to adjust the temperature and flow direction of the coolant to ensure that the liquid-cooling flow rate matches the heat generation of the core; the heat pipe conducts heat through an evaporation-condensation cycle, and the control unit regulates the working state of the heat pipe through an electronically controlled valve or a thermoelectric cooling device to improve the heat dissipation effect.

[0027] As a preferred solution of the cooling cycle system control method for the multi-core processor described in the present invention, wherein: the feedback regulation includes data collection and feature extraction, collecting data related to the processor load, temperature, historical cooling strategy, and ambient temperature; extracting features, such as the load change trend, temperature rise rate, and fan operation state, as the model input;

[0028] Initialize the weights and biases of the neural network; use batch gradient descent to train the model to minimize the loss function; use historical data as the training set and optimize the model through multiple iterations; evaluate the accuracy of the model on the validation set to prevent overfitting;

[0029] Optimize the model through hyperparameter tuning and regularization methods, add an L2 regularization term to avoid model overfitting:

[0030] L regularized (θ) = L(θ) + λ||θ|| 2

[0031] Where λ represents the regularization coefficient, controls the regularization intensity, uses cross-validation to select optimal hyperparameters such as the network structure and learning rate; model training and optimization, uses supervised learning methods to train the model to identify the relationship between temperature increase and load increase;

[0032] Optimize the model parameters using historical data during the training process to improve the model prediction accuracy; intelligent scheduling strategy, based on the prediction results of the AI model, generate an allocation strategy for cooling resources, predict future temperature changes, and adjust the cooling strategy according to the prediction results; assume that the core temperature change trend T core (t + 1) is predicted through the model, estimate the cooling capacity required when the future temperature reaches a certain threshold, and the cooling strategy formula is expressed as:

[0033] C strategy (t) = f(T core (t + 1), L avg (t), T ambient (t))

[0034] Where C strategy(t) represents the cooling strategy at time t, and f fan (t) represents the fan speed, and the liquid cooling flow rate Q liquid (t), the working state of the heat pipe system, etc. For the core of sudden load increase or rapid temperature rise, dynamically adjust the fan speed or increase the liquid cooling flow rate;

[0035] Combined with real-time feedback and predicted data, generate the cooling scheduling decision for the next cycle;

[0036] Model update and incremental learning. The system continuously updates the model through the incremental learning mechanism to adapt to changes in environmental temperature and workload factors;

[0037] Adjust the cooling strategy according to the new operating data to ensure maximum cooling efficiency.

[0038] As a preferred solution of the control method for the cooling cycle system of the multi-core processor described in the present invention, wherein: the optimization decision includes dynamically adjusting the cooling resources, and according to the cooling strategy C strategy (t), dynamically adjust the fan speed and the liquid cooling flow rate, and real-time control the cooling resources; the fan speed f fan (t) is adjusted through a PWM signal:

[0039] f fan (t) = f base (1 + α·ΔT(t))

[0040] Wherein, f base represents the base speed, α represents the adjustment coefficient of the fan speed and the temperature difference, and ΔT(t) represents the temperature change;

[0041] If the temperature gradually drops, the fan speed will gradually decrease to avoid excessive heat dissipation and reduce noise:

[0042] f fan (t) = f base ·(1 - α·ΔT(t))

[0043] The fan speed adjustment is through a PWM signal. The adjustment of the fan speed can be achieved through the pulse width modulation PWM signal; by adjusting the duty cycle of the PWM signal, control the working cycle of the fan, so as to achieve fine adjustment of the fan speed;

[0044]

[0045] Wherein, PWM fan (t) represents the PWM duty cycle at time t, f fan (t) represents the current fan speed, f max represents the maximum speed of the fan;

[0046] Liquid cooling flow rate adjustment, liquid cooling flow rate Q liquid (t) is dynamically adjusted according to temperature difference and load changes:

[0047] Q liquid (t) = Q base ·(1 + β·Δt(t))

[0048] Among them, Q base represents the base flow rate, and β represents the adjustment coefficient of the flow rate and temperature change;

[0049] Let the threshold of the low-temperature state be T low , and the threshold of the high-temperature state be T high , and the overheat temperature threshold be T critical ; During operation, when the core temperature or load exceeds the set threshold, the system will automatically adjust the allocation of cooling resources to ensure that the processor temperature is controlled within a reasonable range; if the core temperature is lower than T low , the system will reduce the allocation of cooling resources and lower the fan speed and liquid cooling flow rate; if the core temperature is within T low ≤T core (t) < T high range, the system maintains the base cooling strategy and operates stably; if the core temperature exceeds T high , the system strengthens cooling, increases the fan speed and liquid cooling flow rate to prevent the processor from overheating; if the core temperature exceeds T critical , the system will immediately enter the emergency cooling mode to maximize the output of cooling resources and avoid hardware damage;

[0050] To ensure that the system can always adapt to new load patterns and environmental temperature changes, the system uses incremental learning to continuously update the model;

[0051] The model parameters are updated through online learning. For incremental learning, assume there is an existing model f(θ), and the model parameters θ are updated through new data points (x new , y new ), and the formula is expressed as:

[0052]

[0053] Among them, η represents the learning rate; represents the gradient of the loss function; Incremental learning helps the system adapt to environmental changes based on new operation data and adjust the cooling strategy according to the feedback.

[0054] As a preferred solution of the control method for the cooling cycle system of the multi-core processor described in the present invention, wherein: the allowing user intervention and adjustment includes real-time data visualization, and displaying information such as the temperature, load, and cooling status of each core of the processor in real time through the interface; providing data display in the form of charts and numerical values to enable the user to clearly understand the operating condition of the system; user input and adjustment, where the user manually adjusts the cooling strategy and sets the cooling mode; the interface allows the user to manually adjust the fan speed and liquid cooling flow parameters through interaction methods such as buttons, sliders, and input boxes.

[0055] A computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the control method for the cooling cycle system of the multi-core processor.

[0056] A computer-readable storage medium stores a computer program thereon. When the computer program is executed by a processor, it implements the steps of the control method for the cooling cycle system of the multi-core processor.

[0057] The beneficial effects of the present invention: The control method for the cooling cycle system of the multi-core processor provided by the present invention integrates multiple temperature sensors and load monitoring modules in the multi-core processor to monitor the temperature and load conditions of each core in real time, and dynamically adjusts the cooling resources through an intelligent scheduling strategy. Compared with traditional cooling technologies, the present invention has obvious advantages in real-time feedback and data prediction. By combining the accurate analysis of load changes and temperature data, it can accurately control the fan speed, liquid cooling flow rate, and the working state of the heat pipe system, thereby improving the heat dissipation efficiency of the processor and reducing energy waste. The present invention provides an innovative technical solution in the intelligentization and precise adjustment of the cooling system, and solves problems such as slow response, overcooling, and low energy efficiency in existing cooling technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0059] Figure 1 It is a schematic diagram of the overall system of a control method for the cooling cycle system of a multi-core processor provided by the first embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] To make the above objects, features, and advantages of the present invention more apparent and understandable, the following provides a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all 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 scope of protection of the present invention.

[0061] Embodiment 1. Refer to Figure 1 , which is an embodiment of the present invention, providing a control method for a cooling circulation system of a multi-core processor, including:

[0062] S1: The multi-core processor 100.

[0063] Furthermore, it is responsible for executing computing tasks. Each core has a different load and generates different degrees of heat.

[0064] Even further, the workload affects the temperature change, providing real-time monitoring data for the temperature sensor 201 and the cooling system; connecting the temperature sensor 201 through a dedicated bus to transmit the temperature information of each core; the sensor is placed in the key heat source areas of the processor.

[0065] S2: The sensor module 200.

[0066] Furthermore, a temperature sensor 201 and a load monitoring module 202; the temperature sensor 201 includes installing temperature sensors 201 in each core and its surrounding areas to monitor the temperature of each core.

[0067] Even further, the monitoring of the temperature of each core includes using the built-in monitoring module of the processor to collect the computing load data of each core and dynamically tracking the working state of each core through a software interface;

[0068] Even further, set a fixed time interval, and trigger the reading operation regularly through an interrupt mechanism as needed; the temperature sensor 201 monitors the temperature of each core and feeds it back to the control unit 301, connecting to the control unit 301 through a signal line to transmit real-time temperature data; the load monitoring module monitors the workload of the multi-core processor 100 in real time, including the computing load, thread activity, and memory bandwidth of each core.

[0069] Even further, the temperature sensor 201 is measured using a thermistor, a thermocouple, or an integrated sensor, installed close to the core, and is divided into internal installation and external installation. Internal installation includes integrating the sensor into the processor chip, and external installation includes installing it at the heat sink part for measurement.

[0070] It should be noted that the temperature sensor 201 outputs the measured temperature information as a digital signal; if it is an analog signal sensor, it is converted into a digital signal through an analog-to-digital converter.

[0071] S3: Cooling system control module 300.

[0072] Furthermore, it includes a control unit 301 and a scheduling module 302, which are responsible for receiving data from the temperature sensor and the load monitoring module, calculating the cooling requirements of each core based on the data, and directing the cooling unit 400 to make adjustments; the control unit 301 judges the load and temperature status of each core; the scheduling module 302 performs feedback regulation and optimizes the decision-making.

[0073] Even further, the cooling system control module 300 includes that the control unit 301 is connected to the multi-core processor 100, the temperature sensor 201, the load monitoring module 202, and the cooling unit 400, and the control unit 301 communicates with each module through a data bus or protocol; the scheduling module 302 predicts the heat changes of different cores of the processor according to the multi-core load information and historical data, and performs intelligent scheduling to optimize the cooling resource allocation; the scheduling module 302 is connected to the control unit 301 through a data transmission interface, analyzes the load and temperature data and feeds the results back to the control unit 301 to optimize the cooling strategy.

[0074] S4: Cooling unit 400.

[0075] It is connected to the control unit 301 through a circuit, and according to the instruction of the control unit 301, adjusts the cooling device 401 to lower the temperature of the processor.

[0076] Even further, the adjustment of the cooling device 401 to lower the temperature of the processor includes air-cooling system adjustment, liquid-cooling system adjustment, and heat pipe adjustment;

[0077] The air-cooling system adjustment includes adjusting the fan speed according to the core temperature data and adjusting the air volume in real time; the fan speed is adjusted through a PWM signal to balance the noise and heat dissipation effect; the liquid-cooling system adjustment includes adjusting the flow rate of the pump in the liquid-cooling system to change the coolant flow rate; a temperature control valve is used to adjust the temperature and flow direction of the coolant to ensure that the liquid-cooling flow rate matches the heat generation of the core; the heat pipe conducts heat through an evaporation-condensation cycle, and the control unit 301 adjusts the working state of the heat pipe through an electronic control valve or a thermoelectric cooling device to improve the heat dissipation effect.

[0078] S5: Debugging interface 500, which provides system status monitoring, displays the temperature, load, and cooling resource allocation of each core; it is connected to the control unit 301 through a display control interface, displays the parameters of the system in real time, and allows users to intervene and adjust.

[0079] Embodiment 2, an embodiment of the present invention, provides a cooling cycle system for a multi-core processor and its control method. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0080] First, to verify the effectiveness of the control method of the cooling cycle system of the multi-core processor, this embodiment selects an 8-core multi-core processor and evaluates the performance of the cooling cycle system of the present invention under different loads and temperatures by comparing it with a traditional cooling system (such as a processor using only an air-cooling system). The working frequency of the processor is 2.5 GHz, the memory bandwidth is 25 GB / s, the maximum rotation speed of the fan is 4000 RPM, and the pump flow rate of the liquid-cooling system is 2 L / min. During the experiment, temperature sensors are placed in each core of the processor and its surrounding area, and together with the load monitoring module, real-time data is collected, including the load, temperature, and ambient temperature information of each core. The cooling system, through the scheduling module, intelligently adjusts the working states of the air-cooling and liquid-cooling systems according to the real-time data.

[0081] The experimental steps are as follows:

[0082] Test platform construction: Select a high-performance processor that supports 8-core processing. A temperature sensor is installed on each core of the processor to ensure accurate monitoring of the temperature of each core during the operation of the processor. At the same time, a load monitoring module is set around each core to collect real-time load data (such as the computing load and thread activity of each core).

[0083] Temperature and load data acquisition: Through the built-in monitoring module, the temperature and load data of each core are periodically collected, and the data is transmitted to the cooling system control module through a dedicated bus. The temperature sensor collects data at an interval of 1 second, and the load monitoring module dynamically evaluates the load situation through the computing tasks of each core.

[0084] The control unit determines the load and temperature status of each core in real time according to the transmitted data. The scheduling module dynamically adjusts the air-cooling and liquid-cooling systems according to the data analysis results, preferentially cooling the cores with higher temperatures or larger loads. The adjustment of the fan rotation speed and the liquid-cooling pump flow rate is achieved through the PWM signal and the liquid-cooling flow control module.

[0085] Different working loads are used to simulate different usage scenarios, including high-load, multi-threaded computing tasks, video encoding, etc. scenarios to ensure that the test results have wide applicability.

[0086] According to the information real-time feedback by the system, adjust the allocation strategy of cooling resources and optimize the cooling process. For the temperature changes of each core, a linear temperature change model is used to predict the future temperature trend, and the decision-making of the scheduling module is continuously optimized through an incremental learning mechanism.

[0087] Table 1 Experimental data table

[0088]

[0089] From the above experimental data, it can be clearly seen the innovation and advantages of the cooling cycle system of the present invention in regulating cooling resources. As can be seen from Table 1, as the load of each core increases (such as from 40% to 100%), the temperature also shows an obvious upward trend (such as Core 6 and Core 5).

[0090] The cooling system of the present invention can intelligently adjust the cooling strategy according to the changes in load and temperature. For example, the fan speed of Core 6 reaches the maximum (4000 RPM) when the load is 100%, and the liquid cooling flow rate also increases to 3.0 L / min, effectively avoiding the over-high core temperature. This ability to adjust cooling resources in real time based on load and temperature greatly improves the stability and reliability of the system during actual operation.

[0091] The "Cooling Strategy Adjustment" column in Table 1 shows the change amount of the cooling strategy. It can be seen that when the core load reaches 80% and above, the adjustment range of the cooling strategy increases significantly. For example, the adjustment range of Core 5 is 30%, and the adjustment range of Core 3 is 20%. This trend indicates that the cooling cycle system of the present invention can quickly respond according to real-time data, preferentially cool the cores with higher temperatures, thereby ensuring that the temperature of the processor is maintained within a reasonable range and avoiding the risk of overheating.

[0092] Compared with the traditional cooling systems that rely only on air cooling or liquid cooling, the cooling cycle system of the present invention can not only adjust the cooling strategy according to the real-time load through intelligent scheduling and dynamic adjustment of cooling resources, but also can predict possible temperature changes in advance through a temperature prediction model, thereby avoiding drastic temperature fluctuations. For example, when the load is high, the system automatically increases the liquid cooling flow rate and fan speed to adapt to the heat increase brought by the high load, significantly improving the heat dissipation efficiency and avoiding the over-high temperature that may be caused by over-reliance on a single cooling method.

[0093] Through the feedback regulation of the intelligent scheduling module, the present invention can precisely control the allocation of cooling resources. For example, the cooling strategy adjustments of Core 4 and Core 8 are 25% and 33% respectively, indicating that the scheduling module has optimized the allocation of cooling resources according to historical load data and temperature change trends, ensuring that the temperature of each core is controlled within the optimal range. Compared with traditional methods, the present invention can not only adjust cooling resources according to the instant temperature, but also make corresponding cooling adjustments in advance by predicting the changes in load and temperature.

[0094] Embodiment 3, an embodiment of the present invention, provides a control method for a cooling cycle system of a multi-core processor, including:

[0095] Judging the load and temperature status of each core includes estimating the heat generation of each core according to the core load and operating frequency.

[0096] The heat generation of the core is proportional to the product of its operating load and frequency, and the power consumption calculation formula is expressed as:

[0097] P core (t) = C v ·f(t) 2 ·V 2 ·L(t)

[0098] Among them, P core (t) represents the power consumption of each core at time t; C v represents the capacitance coefficient, which is related to the process of the processor; f(t) represents the frequency of the core at time t; V represents the operating voltage; L(t) represents the load factor of the processor core, ranging from 0 to 1.

[0099] Using the mathematical model between temperature and power consumption to predict the temperature change of each core in the future for a period of time; assuming that there is a linear relationship between the temperature change of the core and its power consumption, and using the mathematical model of temperature and power consumption to describe the change of core temperature:

[0100] T core (t + 1) = T core (t) + α·(P core (t) - P ambient )

[0101] Among them, T core (t) represents the core temperature at time t; P ambient represents the influencing factor of the ambient temperature; α represents the temperature change coefficient, indicating the influence of power consumption on the core temperature; using the linear coefficient α to characterize the heat dissipation efficiency and environmental factors, and applying a data smoothing algorithm to eliminate the instantaneous error caused by load fluctuations to obtain an accurate heat prediction.

[0102] The control unit 301 sends the analysis result and prediction result to the scheduling module 302 for dynamic scheduling feedback, and establishes a cooling priority according to the load and predicted temperature data.

[0103] The above-mentioned feedback regulation includes data collection and feature extraction, collecting data related to processor load, temperature, historical cooling strategies, and ambient temperature; extracting features, such as load change trends, temperature rise rates, fan operating states, etc., as model inputs.

[0104] Initialize the weights and biases of the neural network; train the model using batch gradient descent to minimize the loss function; use historical data as the training set to optimize the model through multiple iterations; evaluate the accuracy of the model on the validation set to prevent overfitting.

[0105] Optimize the model through hyperparameter tuning and regularization methods, add an L2 regularization term to avoid overfitting of the model:

[0106] L regularized (θ) = L(θ) + λ||θ|| 2

[0107] Among them, λ represents the regularization coefficient, controls the regularization strength, uses cross-validation to select optimal hyperparameters such as the network structure and learning rate; model training and optimization, uses supervised learning methods to train the model to identify the relationship between temperature rise and load increase;

[0108] Optimize the model parameters using historical data during the training process to improve the model prediction accuracy; intelligent scheduling strategy, based on the prediction results of the AI model, generate the allocation strategy of cooling resources, predict the future temperature changes, and adjust the cooling strategy according to the prediction results; assume that the core temperature change trend T core (t + 1) is predicted through the model, estimate the required cooling capacity when the future temperature reaches a certain threshold, and the cooling strategy formula is expressed as:

[0109] C strategy (t) = f(T core (t + 1), L avg (t), T ambient (t))

[0110] Among them, C strategy (t) represents the cooling strategy at time t, and f fan (t) represents the fan speed, and the liquid cooling flow rate Q liquid (t) The working state of the heat pipe system, etc. For the core with sudden load increase or rapid temperature rise, dynamically adjust the fan speed or increase the liquid cooling flow rate.

[0111] Combine real-time feedback and estimated data to generate the cooling scheduling decision for the next cycle.

[0112] Model update and incremental learning, the system continuously updates the model through the incremental learning mechanism to adapt to changes in environmental temperature and workload factors.

[0113] Adjust the cooling strategy according to the new operation data to ensure the maximization of cooling efficiency.

[0114] The optimization decision includes dynamically adjusting the cooling resources. According to the cooling strategy C strategy (t), dynamically adjust the fan speed and liquid cooling flow rate, and regulate the cooling resources in real time; the fan speed f fan (t) is adjusted through the PWM signal:

[0115] f fanf(t) = f base ·(1 + α·ΔT(t))

[0116] where f base represents the base rotational speed, α represents the adjustment coefficient between the fan rotational speed and the temperature difference, and ΔT(t) represents the temperature change amount.

[0117] If the temperature gradually drops, the fan rotational speed will gradually decrease to avoid excessive heat dissipation and reduce noise:

[0118] f fan (t) = f base ·(1 - α·ΔT(t))

[0119] The adjustment of the fan rotational speed is achieved through the PWM signal. The adjustment of the fan rotational speed can be realized by the pulse width modulation PWM signal; by adjusting the duty cycle of the PWM signal, the working cycle of the fan is controlled, thereby achieving fine adjustment of the fan rotational speed.

[0120]

[0121] where PWM fan (t) represents the PWM duty cycle at time t, f fan (t) represents the current fan rotational speed, and f max represents the maximum rotational speed of the fan.

[0122] Adjustment of the liquid cooling flow rate. The liquid cooling flow rate Q liquid (t) is dynamically adjusted according to the temperature difference and load changes:

[0123] Q liquid (t) = Q base ·(1 + β·ΔT(t))

[0124] where Q base represents the base flow rate, and β represents the incremental learning of the adjustment coefficient between the flow rate and the temperature change. To ensure that the system can always adapt to new load patterns and environmental temperature changes, the system needs to use incremental learning to continuously update the model.

[0125] The low - temperature state includes that when T core (t) < T low , the system enters the low - temperature state and reduces the cooling intensity.

[0126] Fan rotational speed: f fan (t) = f fan_base

[0127] Liquid cooling flow rate: Q liquid (t) = Q liquid_base

[0128] The normal - temperature state includes that when T low ≤Tcore (t) < T high When the core temperature is within the normal range, the system maintains the conventional cooling strategy.

[0129] Fan speed: f fan (t) = f fan_base +β fan ·(T core (t) - T ambient (t))

[0130] Liquid cooling flow rate: Q liquid (t) = Q liquid_base +β liquid ·(T core (t) - T ambient (t))

[0131] The high temperature state includes when T core (t) ≥ T high the system activates the enhanced cooling mode.

[0132] Fan speed: f fan (t) = f fan_base +β fan_high ·(T core (t) - T ambient (t))

[0133] Liquid cooling flow rate: Q liquid (t) = Q liquid_base +β liquid_high ·(T core (t) - T ambient (t))

[0134] The overheat state includes when T core (t) ≥ T critical the system enters the emergency cooling mode and all cooling devices operate at full capacity.

[0135] Fan speed: f fan (t) = f fan_max

[0136] Liquid cooling flow rate: Q liquid (t) = Q liquid_max

[0137] Wherein, T low represents the threshold value of the low temperature state; T normal_min and T normal_max represent the minimum and maximum values of the normal temperature range; T high represents the threshold value of the high temperature state; T critical represents the threshold value of the overheat state; Define the threshold parameters of the core load to determine the temperature change and cooling demand caused by the load, wherein, LFlow Threshold indicating a low - load state; LF high Threshold indicating a high - load state.

[0138] During the operation of the system, the load and temperature changes of the core will affect the selection of the cooling strategy. When the core temperature or load exceeds the set threshold, the system will automatically adjust the allocation of cooling resources to ensure that the processor temperature is controlled within a reasonable range. Low - temperature state: If the core temperature is lower than T low , the system will reduce the allocation of cooling resources, lower the fan speed and liquid - cooling flow rate to save energy. Normal - temperature state: If the core temperature is within T low ≤T core (t)<T high range, the system maintains the basic cooling strategy and operates stably. High - temperature state: If the core temperature exceeds T high , the system will strengthen cooling, increase the fan speed and liquid - cooling flow rate to prevent the processor from overheating. Over - heat state: If the core temperature exceeds T critical , the system will immediately enter the emergency cooling mode to maximize the output of cooling resources to avoid hardware damage.

[0139] Update the model parameters through online learning. For incremental learning, given an existing model f(θ), update the model parameters θ with new data points (x new ,y new ), which is expressed by the formula:[[]]

[0140]

[0141] where η represents the learning rate; represents the gradient of the loss function; Incremental learning helps the system adapt to environmental changes according to new operation data and adjust the cooling strategy according to the feedback.

[0142] The described allowing user intervention and adjustment includes real - time data visualization, which displays information such as the temperature, load, and cooling status of each core of the processor in real - time through the interface; providing data display in the form of charts and numerical values to enable users to clearly understand the system operation status; user input and adjustment, where users manually adjust the cooling strategy, such as setting the cooling mode; the interface allows users to adjust parameters such as the fan speed and liquid - cooling flow rate through interactive methods such as buttons, sliders, and input boxes.

[0143] If a function is implemented in the form of 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 the present invention, in essence, or the part that contributes to the prior art, or a part of this 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 for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0144] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a predefined sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in combination with an instruction execution system, apparatus, or device.

[0145] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memories), fiber optic devices, and portable compact disc read-only memories (CDROM). Additionally, a computer-readable medium can even be paper or other suitable media on which a program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other appropriate processing as necessary, and then stored in a computer memory.

[0146] It should be understood that each part of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following techniques known in the art or a combination thereof can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

[0147] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A cooling circulation system for a multi-core processor, characterized in that: include: A multi-core processor (100) is responsible for executing computing tasks, each core has a different load and generates different levels of heat; the workload affects temperature changes, providing real-time monitoring data for the temperature sensor (201) and the cooling system; the temperature sensor (201) is connected via a dedicated bus to transmit temperature information of each core; the sensor is placed in a key heat source area of ​​the processor; The sensor module (200) comprises a temperature sensor (201) and a load monitoring module (202); the temperature sensor (201) comprises: installing the temperature sensor (201) on each core and its surrounding area to monitor the temperature of each core; The cooling system control module (300) includes a control unit (301) and a scheduling module (302), which is responsible for receiving data from the temperature sensor and the load monitoring module, calculating the cooling demand of each core according to the data, and instructing the cooling unit (400) to make adjustments; the control unit (301) determines the load and temperature status of each core; the scheduling module (302) performs feedback adjustment and optimizes the decision; A cooling unit (400) is connected to the control unit (301) via a circuit, and adjusts the cooling device (401) to reduce the temperature of the processor according to the instruction of the control unit (301); The debugging interface (500) provides system status monitoring, displays the temperature, load and cooling resource allocation of each core; it is connected to the control unit (301) through the display control interface, displays various system parameters in real time, and allows users to intervene and adjust.

2. The cooling circulation system of a multi-core processor as claimed in claim 1, characterized in that: The cooling system control module (300) comprises a control unit (301) connected to a multi-core processor (100), a temperature sensor (201), a load monitoring module (202) and a cooling unit (400), wherein the control unit (301) communicates with each module via a data bus or a protocol; a scheduling module (302) predicts the heat changes of different cores of the processor based on multi-core load information and historical data, performs intelligent scheduling, and optimizes cooling resource allocation; and the control unit (301) is connected via a data transmission interface, wherein the scheduling module (302) analyzes load and temperature data and feeds back the results to the control unit (301) to optimize the cooling strategy.

3. The cooling circulation system of a multi-core processor as claimed in claim 2, characterized in that: The monitoring of the temperature of each core includes collecting computing load data of each core using a built-in monitoring module of the processor and dynamically tracking the working status of each core through a software interface; A fixed time interval is set, and a reading operation is triggered regularly through an interrupt mechanism as needed; a temperature sensor (201) monitors the temperature of each core and feeds it back to a control unit (301), and is connected to the control unit (301) through a signal line to transmit real-time temperature data; the load monitoring module monitors the workload of the multi-core processor (100) in real time, including the computing load of each core, thread activity, and memory bandwidth; The temperature sensor (201) is a thermistor, a thermocouple, or an integrated sensor for measurement, and is installed close to the core. It is divided into internal installation and external installation. The internal installation includes integrating the sensor into the processor chip, and the external installation includes installing it at the heat sink for measurement. The temperature sensor (201) outputs the measured temperature information as a digital signal; if it is an analog signal sensor, it is converted into a digital signal through an analog-to-digital converter.

4. A cooling circulation system control method for a multi-core processor, characterized in that: Determining the load and temperature status of each core includes estimating the heat generation of each core according to the core load and operating frequency; The heat generated by the core is proportional to the product of its workload and frequency. The power consumption calculation formula is expressed as: P core (t)=C v ·f(t) 2 ·V 2 ·L(t) Among them, P core (t) represents the power consumption of each core at time t; C v represents the capacitance coefficient, which is related to the process of the processor; f(t) represents the core frequency at time t; V represents the operating voltage; L(t) represents the load factor of the processor core, ranging from 0 to 1; Use the mathematical model between temperature and power consumption to predict the temperature changes of each core in the future; assume that there is a linear relationship between the temperature change of the core and its power consumption, and use the mathematical model of temperature and power consumption to describe the change of core temperature: T core (t+1)=T core (t)+α·(P core (t)_P ambient ) Among them, T core (t) represents the core temperature at time t; P ambient Indicates the influencing factors of ambient temperature; α indicates the temperature variation coefficient, which indicates the influence of power consumption on core temperature; linear coefficient α is used to characterize heat dissipation efficiency and environmental factors, and data smoothing algorithm is applied to eliminate instantaneous errors caused by load fluctuations to obtain accurate heat prediction; The control unit (301) sends the analysis results and prediction results to the scheduling module (302) to perform dynamic scheduling feedback and establish cooling priorities based on the load and predicted temperature data.

5. The cooling circulation system control method of a multi-core processor according to claim 4, characterized in that: The adjusting of the cooling device (401) to reduce the temperature of the processor includes adjusting the air cooling system, adjusting the liquid cooling system and adjusting the heat pipe; The air cooling system is regulated, including adjusting the fan speed according to the core temperature data and adjusting the air volume in real time; the fan speed is regulated by a PWM signal to balance the noise and heat dissipation effect; the liquid cooling system is regulated, including adjusting the flow rate of the pump in the liquid cooling system to change the flow rate of the coolant; a temperature control valve is used to adjust the temperature and flow direction of the coolant to ensure that the liquid cooling flow rate matches the heat generation of the core; the heat pipe conducts heat through an evaporation-condensation cycle, and the control unit (301) adjusts the working state of the heat pipe through an electronic control valve or a thermoelectric cooling device to improve the heat dissipation effect.

6. The cooling cycle system control method of a multi-core processor according to claim 5, characterized in that: The feedback adjustment includes data collection and feature extraction, collecting data related to processor load, temperature, historical cooling strategy, and ambient temperature; Extract features, such as load change trend, temperature rise rate, and fan operation status, as model input; Initialize the weights and biases of the neural network; use batch gradient descent to train the model and minimize the loss function; use historical data as the training set and optimize the model through multiple iterations; evaluate the accuracy of the model on the validation set to prevent overfitting; The model is optimized through hyperparameter tuning and regularization methods, and L2 regularization terms are added to avoid overfitting of the model: L regularized (θ)=L(θ)+λ||θ|| 2 Among them, λ represents the regularization coefficient, which controls the regularization strength. Cross-validation is used to select the optimal network structure and learning rate and other hyperparameters. Model training and optimization uses supervised learning methods to train the model to identify the relationship between temperature increase and load increase. During the training process, historical data is used to optimize model parameters and improve model prediction accuracy; intelligent scheduling strategy, based on the prediction results of the AI ​​model, generates a cooling resource allocation strategy, predicts future temperature changes, and adjusts the cooling strategy based on the prediction results; assume that the core temperature change trend T is predicted by the model core (t+1), estimate the cooling capacity required when the temperature reaches a certain threshold in the future, and the cooling strategy formula is expressed as: C strategy (t)=f(T core (t+1),L avg (t),T ambient (t)) Among them, C strategy (t) represents the cooling strategy at time t, f fan (t) represents the fan speed, liquid cooling flow rate Q liguid (t) The working status of the heat pipe system, etc. For cores with sudden load increases or rapid temperature rises, dynamically adjust the fan speed or increase the liquid cooling flow; Combine real-time feedback and estimated data to generate cooling scheduling decisions for the next cycle; Model update and incremental learning: The system continuously updates the model through the incremental learning mechanism to adapt to changes in ambient temperature and workload factors; Adjust cooling strategy based on new operating data to ensure maximum cooling efficiency.

7. The cooling cycle system control method of a multi-core processor according to claim 6, characterized in that: The optimization decision includes dynamically adjusting cooling resources according to the cooling strategy C strategy (t) Dynamically adjust the fan speed and liquid cooling flow rate to control cooling resources in real time; fan speed f fan (t) Adjustment through PWM signal: f fan (t)=f base ·(1+α·ΔT(t)) Among them, f base represents the basic speed, α represents the adjustment coefficient of the fan speed and temperature difference, and ΔT(t) represents the temperature change; If the temperature gradually drops, the fan speed will gradually decrease to avoid overheating and reduce noise: f fan (t)=f base ·(1-α·ΔT(t)) The fan speed is adjusted through the PWM signal. The fan speed can be adjusted through the pulse width modulation PWM signal. By adjusting the duty cycle of the PWM signal, the fan's working cycle can be controlled, thereby achieving fine adjustment of the fan speed. Among them, PWM fan (t) represents the PWM duty cycle at time t, f fan (t) indicates the current fan speed, f max Indicates the maximum speed of the fan; Liquid cooling flow adjustment, liquid cooling flow Q liquid (t) Dynamic adjustment according to temperature difference and load changes: Q liquid (t)=Q base ·(1+β·ΔT(t)) Among them, Q base represents the basic flow rate, and β represents the adjustment coefficient of flow rate and temperature change; Assume the threshold value of low temperature state is T low , the threshold value of high temperature state T high , overheating temperature threshold T critical During operation, when the core temperature or load exceeds the set threshold, the system will automatically adjust the allocation of cooling resources to ensure that the processor temperature is controlled within a reasonable range; if the core temperature is lower than T low , the system will reduce the allocation of cooling resources, reduce fan speed and liquid cooling flow; if the core temperature is T low ≤T core (t) <T high Within the range, the system maintains the basic cooling strategy and keeps stable operation; if the core temperature exceeds T high , the system strengthens cooling, increases fan speed and liquid cooling flow to prevent processor overheating; if the core temperature exceeds T Critical , the system will immediately enter emergency cooling mode to maximize the output of cooling resources and avoid hardware damage; To ensure that the system can always adapt to new load patterns and ambient temperature changes, the system uses incremental learning to continuously update the model; Update model parameters through online learning, incremental learning assumes that the existing model f(θ) is updated through new data points (x new ,y new ) to update the model parameters θ, the formula is expressed as: Among them, η represents the learning rate; Represents the gradient of the loss function; incremental learning helps the system adapt to environmental changes based on new operating data and adjust the cooling strategy based on feedback.

8. The cooling cycle system control method of a multi-core processor according to claim 7, characterized in that: The method of allowing user intervention and adjustment includes real-time data visualization, displaying the temperature, load, cooling status and other information of each core of the processor in real time through the interface; providing data display in the form of charts and numerical values ​​to allow users to clearly understand the system operation status; user input and adjustment, the user manually adjusts the cooling strategy and sets the cooling mode; the interface allows the user to manually adjust the fan speed and liquid cooling flow parameters through buttons, sliders, and input boxes.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the cooling circulation system and the control method of the multi-core processor according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the cooling circulation system and the control method of the multi-core processor according to any one of claims 1 to 7 are implemented.

Images