Heat dissipation control method, device and system of external computing power module

Abstract

The embodiment of the present application relates to the field of electronic equipment, and discloses a heat dissipation control method, device and system of an external computing power module, wherein the heat dissipation device of the external computing power module is a liquid cooling device, and the method comprises the following steps: acquiring a real-time first temperature and a prediction result, wherein the first temperature comprises a computing power chip temperature of the external computing power module and / or a temperature of a liquid cooling micro-channel plate of the liquid cooling device; when the real-time first temperature is greater than or equal to an upper limit value of temperature, the liquid cooling device is controlled to work in a first running state; when the real-time first temperature is less than the upper limit value of temperature and the prediction result is that the first temperature does not exceed a preset upper limit value of temperature in a future period of time, if the real-time first temperature is greater than or equal to a lower limit value of temperature, the liquid cooling device is controlled to keep working in a current running state; and if the real-time first temperature is less than the lower limit value of temperature, the liquid cooling device is controlled to stop running, so as to reduce heat dissipation noise and power consumption, improve heat dissipation capacity, and cope with thermal load in peak operation.

Description

Technical Field

[0001] This invention relates to the field of electronic devices, and in particular to a heat dissipation control method, apparatus and system for an external computing module. Background Technology

[0002] As AI applications running on mobile devices such as smartphones and tablets become increasingly complex, external computing modules are gradually becoming important components for expanding storage and accelerating AI. These external computing modules integrate high-performance AI chips, which generate a significant amount of heat during computationally intensive tasks such as inference and training.

[0003] Existing external computing modules mainly use natural convection cooling or small fan cooling methods, but they have the following problems: 1. Limited heat dissipation capacity, unable to cope with the thermal load of AI chips during peak operation; 2. Air-cooled solutions are noisy and consume a lot of power, making them unsuitable for mobile scenarios; 3. While liquid cooling can provide efficient heat dissipation, if the pump or circulation system continues to run, it will result in significant power consumption waste and shorten the battery life. Summary of the Invention

[0004] The purpose of this invention is to provide at least one heat dissipation control method, device and system for external computing power modules, which can at least solve the problems of high heat dissipation noise, high power consumption, low heat dissipation capacity and inability to cope with the thermal load of AI chips during peak operation, and can at least reduce heat dissipation noise and power consumption, improve heat dissipation capacity and cope with the thermal load during peak operation.

[0005] To address the aforementioned technical problems, at least one embodiment of this application provides a heat dissipation control method for an external computing module, wherein the heat dissipation device for the external computing module is a liquid cooling device, and the method includes: The system obtains a real-time first temperature and a prediction of whether the first temperature will exceed the first temperature threshold in the future. The first temperature includes the temperature of the computing chip of the external computing module and / or the temperature of the liquid-cooled microfluidic plate of the liquid cooling device. The real-time first temperature is compared with a preset first temperature threshold. When the real-time first temperature is greater than or equal to the preset first temperature threshold, the liquid cooling device is controlled to operate in the first operating state. When the real-time first temperature is less than the first temperature threshold, and the prediction result is that the first temperature will not exceed the preset first temperature threshold in the future, the real-time first temperature is compared with the preset second temperature threshold. When the real-time first temperature is greater than or equal to the second temperature threshold, the liquid cooling device is controlled to maintain its current operating state; when the real-time first temperature is less than the second temperature threshold, the liquid cooling device is controlled to stop operating. Wherein, the first temperature threshold is greater than the second temperature threshold.

[0006] In some optional embodiments, the heat dissipation control method for the external computing module further includes: When the real-time first temperature is less than the first temperature threshold, and the prediction result is that the first temperature will exceed the preset first temperature threshold in the future, the liquid cooling device is controlled to operate in the second operating state. The coolant flow rate of the liquid cooling device in the second operating state is lower than the coolant flow rate of the liquid cooling device in the first operating state.

[0007] In some optional embodiments, after the step of controlling the liquid cooling device to operate in the second operating state, the method further includes: The time during which the liquid cooling device operates in the second operating state is timed to obtain a first duration; Compare the first duration with the preset cooling duration; When the first duration reaches the cooling duration, a new prediction result is obtained.

[0008] In some optional embodiments, obtaining the prediction result of whether the first temperature will exceed the first temperature threshold in the future includes: Obtain the task type and computing power requirement of the AI ​​task that the mobile device is about to unload to the external computing power module; Based on the task type and computing power requirements of the obtained AI task, a predictive analysis is performed to predict whether the first temperature will exceed the first temperature threshold in the future, and the prediction result is generated.

[0009] In some optional embodiments, obtaining the prediction result of whether the first temperature will exceed the first temperature threshold in the future includes: Obtain historical variation data of the first temperature; Based on the historical change data of the first temperature, the future trend of the first temperature is predicted by a pre-trained temperature differential model. Based on the predicted future trend of the first temperature, the analysis is performed to determine whether the first temperature will exceed the first temperature threshold in the future, and the prediction result is generated.

[0010] In some optional embodiments, obtaining the prediction result of whether the first temperature will exceed the first temperature threshold in the future includes: Monitor the power consumption of the external computing module; If the current power consumption is greater than the preset peak power consumption and the duration is greater than a predetermined value, a prediction result is generated that the first temperature will exceed the first temperature threshold in the future; otherwise, a prediction result is generated that the first temperature will not exceed the first temperature threshold in the future.

[0011] In some optional embodiments, the step of controlling the liquid cooling device to operate in a first operating state when the real-time first temperature is greater than or equal to a preset first temperature threshold includes: Get the currently used control mode; When the control mode is performance mode, the liquid cooling device is directly controlled to operate in the first operating state; When the control mode is energy-saving mode, an over-temperature reminder message is issued to remind the user to manually trigger the liquid cooling device to operate in the first operating state.

[0012] At least one embodiment of this application also provides a heat dissipation control device for an external computing power module, wherein the heat dissipation device for the external computing power module is a liquid cooling device, and the heat dissipation control device includes: A temperature acquisition module is used to acquire a first temperature in real time, the first temperature including the temperature of the computing chip of the external computing module and / or the temperature of the liquid-cooled microfluidic plate of the liquid cooling device. The prediction module is used to predict whether the first temperature will exceed the first temperature threshold in the future and generate prediction results. The control module is configured to acquire a real-time first temperature and the prediction result, and compare the real-time first temperature with a preset first temperature threshold; when the real-time first temperature is greater than or equal to the preset first temperature threshold, control the liquid cooling device to operate in a first operating state; and when the real-time first temperature is less than the first temperature threshold, and the prediction result indicates that the first temperature will not exceed the preset first temperature threshold in the future, compare the real-time first temperature with a preset second temperature threshold; when the real-time first temperature is greater than or equal to the second temperature threshold, control the liquid cooling device to maintain its current operating state; and when the real-time first temperature is less than the second temperature threshold, control the liquid cooling device to stop operating; wherein the first temperature threshold is greater than the second temperature threshold.

[0013] At least one embodiment of this application also provides a heat dissipation control system for an external computing module, including a liquid cooling device and a heat dissipation control device for the external computing module as described above, wherein the liquid cooling device is used to dissipate heat from the external computing module.

[0014] At least one embodiment of this application also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the above-described heat dissipation control method for an external computing module.

[0015] At least one embodiment of this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described heat dissipation control method for an external computing module.

[0016] The heat dissipation control method, apparatus, and system for external computing modules provided in this embodiment, compared with the current air-cooling method, uses a liquid cooling device to dissipate heat from the external computing module, resulting in lower noise, higher heat dissipation efficiency, and a more compact structure suitable for mobile scenarios such as external computing modules. Simultaneously, temperature-controlled triggering of the liquid cooling device ensures it only activates when the temperature is too high, reducing unnecessary power consumption and extending module battery life. Furthermore, by setting a first and second temperature threshold in the temperature control trigger, a hysteresis zone is formed, preventing frequent start-stop of the liquid cooling device and improving system stability. Moreover, by predicting whether the computing chip temperature will exceed the first temperature threshold, the liquid cooling device is triggered in advance when the predicted temperature exceeds the first temperature threshold, maintaining the computing chip temperature within the optimal range and avoiding frequency throttling caused by instantaneous heat peaks. This application supports temperature-controlled triggering, predictive triggering, and manual triggering for the start-up control of the liquid cooling device, and includes performance and energy-saving modes to balance performance and energy efficiency requirements. Attached Figure Description

[0017] One or more embodiments are illustrated by way of example with reference to the accompanying drawings, and these illustrative descriptions do not constitute a limitation on the embodiments.

[0018] Figure 1 This is a flowchart of a heat dissipation control method for an external computing module provided in one embodiment of this application. Figure 1 .

[0019] Figure 2 This is a flowchart of a heat dissipation control method for an external computing module provided in another embodiment of this application. Figure 2 .

[0020] Figure 3This is a structural block diagram of a heat dissipation control method device for an external computing module provided in another embodiment of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been provided in the various embodiments of this application to help readers better understand this application. However, the technical solutions claimed in this application can be implemented even without these technical details and various changes and modifications based on the following embodiments. The division of the various embodiments below is for the convenience of description and should not constitute any limitation on the specific implementation of this application. The various embodiments can be combined with and referenced by each other without contradiction.

[0022] To facilitate understanding of the embodiments of this application, the relevant content regarding the heat dissipation method of the external computing module of the mobile device will be introduced first.

[0023] Existing external computing modules mainly use natural convection cooling or small fan cooling methods, but they have the following problems: 1. Limited heat dissipation capacity, unable to cope with the thermal load of AI chips during peak operation; 2. Air-cooled solutions are noisy and consume a lot of power, making them unsuitable for mobile scenarios; 3. While liquid cooling can provide efficient heat dissipation, if the pump or circulation system continues to run, it will result in significant power consumption waste and shorten the battery life.

[0024] To address the aforementioned technical problems of high heat dissipation noise, high power consumption, low heat dissipation capacity, and inability to handle the thermal load of AI chips during peak operation, this invention proposes a heat dissipation control method for an external computing power module. The implementation details of the heat dissipation control method for the external computing power module in this embodiment are described below. The following content is only for ease of understanding and is not necessary for implementing this solution.

[0025] Example 1: The heat dissipation control method for the external computing module in this embodiment can be applied to electronic devices with communication, computing, and data storage capabilities. The heat dissipation device for the external computing module is a liquid cooling device. The specific process of the method is as follows: Figure 1 The following are included: Step S1: Obtain the real-time first temperature and the prediction result of whether the first temperature will exceed the first temperature threshold in the future.

[0026] The first temperature includes the temperature of the computing chip of the external computing module and / or the temperature of the liquid-cooled microfluidic plate of the liquid cooling device.

[0027] Step S2: Compare the real-time first temperature with a preset first temperature threshold; when the real-time first temperature is greater than or equal to the preset first temperature threshold, proceed to step S3; when the real-time first temperature is less than the first temperature threshold, and the prediction result is that the first temperature will not exceed the preset first temperature threshold in the future, proceed to step S4.

[0028] Step S3: Control the liquid cooling device to operate in the first operating state.

[0029] Step S4: Compare the real-time first temperature with a preset second temperature threshold. If the real-time first temperature is greater than or equal to the second temperature threshold, proceed to step S5; if the real-time first temperature is less than the second temperature threshold, proceed to step S6. Wherein, the first temperature threshold is greater than the second temperature threshold.

[0030] Step S5: Control the liquid cooling device to maintain its current operating state.

[0031] Step S6: Control the liquid cooling device to stop operating.

[0032] In some embodiments, step S1, "obtaining a prediction result of whether the first temperature will exceed the first temperature threshold in the future", includes: obtaining the task type and computing power requirement of the AI ​​task that the mobile device is about to unload to the external computing power module; and performing a predictive analysis on whether the first temperature will exceed the first temperature threshold in the future based on the obtained task type and computing power requirement of the AI ​​task, and generating the prediction result.

[0033] Specifically, before deciding to offload an AI task to an external computing module, the mobile device will know the task type and computing power requirements of that AI task in advance. For example, AI tasks such as Transformer inference, image diffusion model inference, high-resolution image encoding, and video segmentation inference have strong computational peak characteristics. Once offloaded to the external computing module, they will inevitably cause the computing chip of the external computing module to operate under high load, thereby generating a lot of heat and even forming instantaneous heat peaks. Therefore, based on the task type and computing power requirements of the AI ​​task that the mobile device is about to offload to the external computing module, it can predict in advance whether a heat peak exceeding the first temperature threshold will arrive. If a heat peak exceeding the first temperature threshold is predicted to arrive, the liquid cooling device can be activated in advance to prevent the real-time first temperature from rapidly exceeding the first temperature threshold.

[0034] In some embodiments, step S1, "obtaining a prediction result on whether the first temperature will exceed the first temperature threshold in the future," includes: obtaining historical change data of the first temperature; based on the obtained historical change data of the first temperature, predicting the future change trend of the first temperature using a pre-trained temperature differential model; and analyzing whether the first temperature will exceed the first temperature threshold in the future based on the predicted future change trend of the first temperature, thereby generating the prediction result.

[0035] In one example, the first temperature change data over the past 2 to 10 seconds is obtained, and the future first temperature is predicted using a temperature differential model. Based on the predicted trend of the first temperature change in the future, it is analyzed whether the future first temperature will exceed the first temperature threshold.

[0036] In some embodiments, step S1, “obtaining a prediction result of whether the first temperature will exceed the first temperature threshold in the future”, includes: monitoring the power consumption of the external computing module; if the current power consumption is greater than a preset power consumption peak and the duration is greater than a predetermined value, then generating a prediction result that the first temperature will exceed the first temperature threshold in the future; otherwise, generating a prediction result that the first temperature will not exceed the first temperature threshold in the future.

[0037] Specifically, the power consumption of the external computing module can be monitored through the power management integrated circuit (PMIC) or a current sensor.

[0038] In some embodiments, step S1, "obtaining a prediction result of whether the first temperature will exceed the first temperature threshold in the future", includes: monitoring whether the multi-channel communication of the mobile device or the external computing module exceeds the data transmission of a quantity threshold and / or the high-speed model parameter fragmentation transmission; if the multi-channel communication of the mobile device or the external computing module exceeds the data transmission of a quantity threshold and / or the high-speed model parameter fragmentation transmission, then it is determined that the first temperature will exceed the first temperature threshold in the future, and the prediction result is generated.

[0039] Specifically, before AI tasks (such as image diffusion model inference and high-resolution image encoding) can run, the raw data required for the task (such as image pixel data and video frame data) must be transmitted from the mobile device to the external computing module. This data volume typically far exceeds that of ordinary tasks (such as text processing). Therefore, in multi-channel communication, if a bandwidth request or data transmission preprocessing signal is detected indicating that a high-computing task will soon be launched, it can be directly determined that only high-load AI tasks require such a large amount of input data. This allows for the prediction that the computing chip will enter a high-load, overheating state. Furthermore, complex AI models (such as Transformer models) have a large number of parameters that cannot be transmitted in one complete transmission. They are usually transmitted in segments, with the model parameters split and transmitted at high speed to the external module before being assembled and deployed to prepare for task execution. Therefore, detecting high-speed segmented transmission of model parameters indicates that a high-load task is about to start, allowing liquid cooling to be triggered in advance without waiting for the task to officially run.

[0040] In some embodiments, the step of controlling the liquid cooling device to operate in a first operating state when the real-time first temperature is greater than or equal to a preset first temperature threshold includes: obtaining the currently adopted control mode; when the control mode is a performance mode, controlling the pump driving the liquid cooling device to make the liquid cooling device operate in the first operating state; when the control mode is an energy-saving mode, initiating an over-temperature reminder message to remind the user to manually trigger the liquid cooling device to operate in the first operating state.

[0041] Specifically, in performance mode, when the real-time temperature of the computing chip or the temperature of the liquid-cooled microfluidic plate exceeds a first temperature threshold, the micropump of the liquid cooling device is activated by the control unit to drive the coolant to flow through the microfluidic plate and remove heat from the external computing module. In some examples, the coolant used in the liquid cooling device is insulating perfluorohexanone, and the radiator of the liquid cooling device is located on the side of the external computing module's housing.

[0042] In energy-saving mode, when the real-time temperature of the computing chip or the temperature of the liquid-cooled microfluidic plate exceeds a first temperature threshold, a high-temperature warning is displayed on the mobile device or external computing module's screen. The user can then manually trigger the liquid cooling cycle of the liquid cooling device instead of pump-driven operation to reduce power consumption. The manual triggering of the liquid cooling cycle includes methods such as squeezing the liquid reservoir of the liquid cooling device, using a hand-cranked reciprocating plunger pump, or a spring / coil-spring energy-storing micro-pump. In some examples, in energy-saving mode, when the real-time temperature of the computing chip or the temperature of the liquid-cooled microfluidic plate exceeds 70°C, a high-temperature warning is displayed on the mobile device or external computing module's screen. The user can then trigger one cycle of the liquid cooling circuit by squeezing the corresponding pressing area of ​​the flexible liquid reservoir of the liquid cooling device, thus achieving low-energy passive liquid cooling.

[0043] The heat dissipation control method for the external computing module provided in this embodiment, compared with the current air cooling method, uses a liquid cooling device to dissipate heat from the external computing module, resulting in lower noise, higher heat dissipation efficiency, and a more compact structure suitable for mobile scenarios such as external computing modules. Simultaneously, temperature-controlled triggering of the liquid cooling device ensures that it only activates when the temperature is too high, reducing unnecessary power consumption and extending the module's battery life. Furthermore, by setting a first and second temperature threshold in the temperature control triggering, a hysteresis zone is formed, preventing frequent start-stop of the liquid cooling device and improving system stability. This application also supports temperature-controlled triggering, predictive triggering, and manual triggering for the start-up control of the liquid cooling device, and includes performance and energy-saving modes to balance performance and energy efficiency requirements.

[0044] Example 2: The heat dissipation control method for the external computing module in this embodiment can be applied to electronic devices with communication, computing, and data storage capabilities. The heat dissipation device for the external computing module is a liquid cooling device. The liquid cooling device includes a liquid-cooled microfluidic plate, a cooling circuit, and a coolant. The cooling circuit includes a pump, a manual pressure chamber, piping, and a radiator. The specific flow of the method can be as follows: Figure 2 The following are included: Step S1: Obtain the real-time first temperature and a prediction of whether the first temperature will exceed the first temperature threshold in the future. The first temperature includes the temperature of the computing chip in the external computing module and / or the temperature of the liquid-cooled microfluidic plate of the liquid cooling device.

[0045] Specifically, the real-time temperature of the computing chip or the liquid-cooled microfluidic plate can be monitored by a temperature sensor installed on the computing chip or the liquid-cooled microfluidic plate.

[0046] Step S2: Compare the real-time first temperature with a preset first temperature threshold; when the real-time first temperature is greater than or equal to the preset first temperature threshold, proceed to step S3; when the real-time first temperature is less than the first temperature threshold, and the prediction result is that the first temperature will not exceed the preset first temperature threshold in the future, proceed to step S4; when the real-time first temperature is less than the first temperature threshold, and the prediction result is that the first temperature will exceed the preset first temperature threshold in the future, proceed to step S7.

[0047] Step S3: Control the liquid cooling device to operate in the first operating state.

[0048] Step S4: Compare the real-time first temperature with a preset second temperature threshold. If the real-time first temperature is greater than or equal to the second temperature threshold, proceed to step S5; if the real-time first temperature is less than the second temperature threshold, proceed to step S6. The first temperature threshold is greater than the second temperature threshold. In one example, the first temperature threshold is 70°C, and the second temperature threshold is 55°C.

[0049] Step S5: Control the liquid cooling device to maintain its current operating state.

[0050] Step S6: Control the liquid cooling device to stop operating.

[0051] Step S7: Control the liquid cooling device to operate in a second operating state. The coolant flow rate in the second operating state is lower than the coolant flow rate in the first operating state.

[0052] In some embodiments, after step S7, step S8 is further included: timing the time for the liquid cooling device to operate in the second operating state to obtain a first duration; comparing the first duration with a preset cooling duration; and executing step S1 when the first duration reaches the cooling duration.

[0053] Specifically, by setting a cooling-off period, repeated predictions can be avoided within a short period of time.

[0054] In some embodiments, step S1, "obtaining a prediction result of whether the first temperature will exceed the first temperature threshold in the future", includes: obtaining the task type and computing power requirement of the AI ​​task that the mobile device is about to unload to the external computing power module; and performing a predictive analysis on whether the first temperature will exceed the first temperature threshold in the future based on the obtained task type and computing power requirement of the AI ​​task, and generating the prediction result.

[0055] Specifically, before deciding to offload an AI task to an external computing module, the mobile device will know the task type and computing power requirements of that AI task in advance. For example, AI tasks such as Transformer inference, image diffusion model inference, high-resolution image encoding, and video segmentation inference have strong computational peak characteristics. Once offloaded to the external computing module, they will inevitably cause the computing chip of the external computing module to operate under high load, thereby generating a lot of heat and even forming instantaneous heat peaks. Therefore, based on the task type and computing power requirements of the AI ​​task that the mobile device is about to offload to the external computing module, it can predict in advance whether a heat peak exceeding the first temperature threshold will arrive. If a heat peak exceeding the first temperature threshold is predicted to arrive, the liquid cooling device can be activated in advance to prevent the real-time first temperature from rapidly exceeding the first temperature threshold.

[0056] In some embodiments, step S1, "obtaining a prediction result on whether the first temperature will exceed the first temperature threshold in the future," includes: obtaining historical change data of the first temperature; based on the obtained historical change data of the first temperature, predicting the future change trend of the first temperature using a pre-trained temperature differential model; and analyzing whether the first temperature will exceed the first temperature threshold in the future based on the predicted future change trend of the first temperature, thereby generating the prediction result.

[0057] In one example, the first temperature change data over the past 2 to 10 seconds is obtained, and the future first temperature is predicted using a temperature differential model. Based on the predicted trend of the first temperature change in the future, it is analyzed whether the future first temperature will exceed the first temperature threshold.

[0058] In some embodiments, step S1, “obtaining a prediction result of whether the first temperature will exceed the first temperature threshold in the future”, includes: monitoring the power consumption of the external computing module; if the current power consumption is greater than a preset power consumption peak and the duration is greater than a predetermined value, then generating a prediction result that the first temperature will exceed the first temperature threshold in the future; otherwise, generating a prediction result that the first temperature will not exceed the first temperature threshold in the future.

[0059] Specifically, the power consumption of the external computing module can be monitored through the PMIC or current sensor of the external computing module.

[0060] In some embodiments, step S1, "obtaining a prediction result of whether the first temperature will exceed the first temperature threshold in the future", includes: monitoring whether the multi-channel communication of the mobile device or the external computing module exceeds the data transmission of a quantity threshold and / or the high-speed model parameter fragmentation transmission; if the multi-channel communication of the mobile device or the external computing module exceeds the data transmission of a quantity threshold and / or the high-speed model parameter fragmentation transmission, then it is determined that the first temperature will exceed the first temperature threshold in the future, and the prediction result is generated.

[0061] Specifically, before AI tasks (such as image diffusion model inference and high-resolution image encoding) can run, the raw data required for the task (such as image pixel data and video frame data) must be transmitted from the mobile device to the external computing module. This data volume typically far exceeds that of ordinary tasks (such as text processing). Therefore, in multi-channel communication, if a bandwidth request or data transmission preprocessing signal is detected indicating that a high-computing task will soon be launched, it can be directly determined that only high-load AI tasks require such a large amount of input data. This allows for the prediction that the computing chip will enter a high-load, overheating state. Furthermore, complex AI models (such as Transformer models) have a large number of parameters that cannot be transmitted in one complete transmission. They are usually transmitted in segments, with the model parameters split and transmitted at high speed to the external module before being assembled and deployed to prepare for task execution. Therefore, detecting high-speed segmented transmission of model parameters indicates that a high-load task is about to start, allowing liquid cooling to be triggered in advance without waiting for the task to officially run.

[0062] In some embodiments, the step of controlling the liquid cooling device to operate in a first operating state when the real-time first temperature is greater than or equal to a preset first temperature threshold includes: obtaining the currently adopted control mode; when the control mode is a performance mode, controlling the pump driving the liquid cooling device to make the liquid cooling device operate in the first operating state; when the control mode is an energy-saving mode, initiating an over-temperature reminder message to remind the user to manually trigger the liquid cooling device to operate in the first operating state.

[0063] Specifically, in performance mode, when the real-time temperature of the computing chip or the temperature of the liquid-cooled microfluidic plate exceeds a first temperature threshold, the micropump of the liquid cooling device is activated by the control unit to drive the coolant to flow through the microfluidic plate and remove heat from the external computing module. In some examples, the coolant used in the liquid cooling device is insulating perfluorohexanone, and the radiator of the liquid cooling device is located on the side of the external computing module's housing.

[0064] In energy-saving mode, when the real-time temperature of the computing chip or the temperature of the liquid-cooled microfluidic plate exceeds a first temperature threshold, a high-temperature warning is displayed on the mobile device or external computing module's screen. The user can then manually trigger the liquid cooling cycle of the liquid cooling device instead of pump-driven operation to reduce power consumption. The manual triggering of the liquid cooling cycle includes methods such as squeezing the liquid reservoir of the liquid cooling device, using a hand-cranked reciprocating plunger pump, or a spring / coil-spring energy-storing micro-pump. In some examples, in energy-saving mode, when the real-time temperature of the computing chip or the temperature of the liquid-cooled microfluidic plate exceeds 70°C, a high-temperature warning is displayed on the mobile device or external computing module's screen. The user can then trigger one cycle of the liquid cooling circuit by squeezing the corresponding pressing area of ​​the flexible liquid reservoir of the liquid cooling device, thus achieving low-energy passive liquid cooling.

[0065] The heat dissipation control method for the external computing module provided in this embodiment, compared with the current air cooling method, uses a liquid cooling device to dissipate heat from the external computing module, resulting in lower noise, higher heat dissipation efficiency, and a more compact structure suitable for mobile scenarios such as external computing modules. Simultaneously, temperature-controlled triggering of the liquid cooling device ensures it only activates when the temperature is too high, reducing unnecessary power consumption and extending the module's battery life. Furthermore, by setting a first and second temperature threshold in the temperature control trigger, a hysteresis zone is formed, preventing frequent start-stop of the liquid cooling device and improving system stability. Moreover, by predicting whether the computing chip's temperature will exceed the first temperature threshold, the liquid cooling device is triggered in advance when this is predicted, maintaining the computing chip's temperature within the optimal range and avoiding frequency throttling caused by instantaneous heat peaks. This application supports temperature-controlled triggering, predictive triggering, and manual triggering for the liquid cooling device's activation control, and includes performance and energy-saving modes to balance performance and energy efficiency requirements.

[0066] Example 3: Another embodiment of this application relates to a heat dissipation control device for an external computing module. The implementation details of the heat dissipation control device for the external computing module in this embodiment are described in detail below. The following implementation details are provided for ease of understanding only and are not necessary for implementing this solution. The heat dissipation device for the external computing module in this embodiment is a liquid cooling device. The liquid cooling device includes a liquid cooling microfluidic plate, a cooling circuit and a coolant. The cooling circuit includes a pump, a manual pressure chamber, pipelines and a radiator.

[0067] A schematic diagram of the heat dissipation control device for the external computing module in this embodiment can be seen as follows: Figure 3 As shown, it includes a temperature acquisition module 301, a prediction module 302, and a control module 803.

[0068] Temperature acquisition module 301 is used to acquire a first temperature in real time, the first temperature including the temperature of the computing chip of the external computing module and / or the temperature of the liquid-cooled microfluidic plate of the liquid cooling device.

[0069] In some examples, the temperature acquisition module 301 employs a temperature sensor, which is mounted on the computing chip of an external computing module or on the liquid-cooled microfluidic plate of the liquid cooling device.

[0070] The prediction module 302 is used to predict whether the first temperature will exceed the first temperature threshold in the future and generate a prediction result.

[0071] In some embodiments, the prediction module 302 is further configured to obtain the task type and computing power requirement of the AI ​​task that the mobile device is about to unload to the external computing power module; based on the obtained task type and computing power requirement of the AI ​​task, to perform predictive analysis on whether the first temperature will exceed the first temperature threshold in the future, and generate the prediction result.

[0072] Specifically, before deciding to offload an AI task to an external computing module, the mobile device will know the task type and computing power requirements of that AI task in advance. For example, AI tasks such as Transformer inference, image diffusion model inference, high-resolution image encoding, and video segmentation inference have strong computational peak characteristics. Once offloaded to the external computing module, they will inevitably cause the computing chip of the external computing module to operate under high load, thereby generating a lot of heat and even forming instantaneous heat peaks. Therefore, based on the task type and computing power requirements of the AI ​​task that the mobile device is about to offload to the external computing module, it can predict in advance whether a heat peak exceeding the first temperature threshold will arrive. If a heat peak exceeding the first temperature threshold is predicted to arrive, the liquid cooling device can be activated in advance to prevent the real-time first temperature from rapidly exceeding the first temperature threshold.

[0073] In some embodiments, the prediction module 302 is further configured to acquire historical change data of the first temperature; based on the acquired historical change data of the first temperature, predict the future change trend of the first temperature through a pre-trained temperature differential model; and analyze whether the first temperature will exceed the first temperature threshold in the future based on the predicted future change trend of the first temperature, thereby generating the prediction result.

[0074] In one example, the first temperature change data over the past 2 to 10 seconds is obtained, and the future first temperature is predicted using a temperature differential model. Based on the predicted trend of the first temperature change in the future, it is analyzed whether the future first temperature will exceed the first temperature threshold.

[0075] In some embodiments, the prediction module 302 is further configured to monitor the power consumption of the external computing module; if the current power consumption is greater than a preset power consumption peak and the duration is greater than a predetermined value, then a prediction result is generated that the first temperature will exceed the first temperature threshold in the future; otherwise, a prediction result is generated that the first temperature will not exceed the first temperature threshold in the future.

[0076] Specifically, the power consumption of the external computing module can be monitored through the PMIC or current sensor of the external computing module.

[0077] In some embodiments, the prediction module 302 is further configured to monitor whether the multi-channel communication of the mobile device or the external computing module exceeds the data transmission of a quantity threshold and / or the high-speed model parameter fragmentation transmission; if the multi-channel communication of the mobile device or the external computing module exceeds the data transmission of a quantity threshold and / or the high-speed model parameter fragmentation transmission, it is determined that the first temperature will exceed the first temperature threshold in the future, and the prediction result is generated.

[0078] Specifically, before AI tasks (such as image diffusion model inference and high-resolution image encoding) can run, the raw data required for the task (such as image pixel data and video frame data) must be transmitted from the mobile device to the external computing module. This data volume typically far exceeds that of ordinary tasks (such as text processing). Therefore, in multi-channel communication, if a bandwidth request or data transmission preprocessing signal is detected indicating that a high-computing task will soon be launched, it can be directly determined that only high-load AI tasks require such a large amount of input data. This allows for the prediction that the computing chip will enter a high-load, overheating state. Furthermore, complex AI models (such as Transformer models) have a large number of parameters that cannot be transmitted in one complete transmission. They are usually transmitted in segments, with the model parameters split and transmitted at high speed to the external module before being assembled and deployed to prepare for task execution. Therefore, detecting high-speed segmented transmission of model parameters indicates that a high-load task is about to start, allowing liquid cooling to be triggered in advance without waiting for the task to officially run.

[0079] The control module 303, electrically connected to the temperature acquisition module 301 and the prediction module 302, is used to acquire a first temperature collected in real time by the temperature acquisition module 301 and a prediction result generated by the prediction module 302, and compare the real-time first temperature with a preset first temperature threshold; when the real-time first temperature is greater than or equal to the preset first temperature threshold, the control module controls the liquid cooling device to operate in a first operating state; it is also used to compare the real-time first temperature with a preset second temperature threshold when the real-time first temperature is less than the first temperature threshold and the prediction result indicates that the first temperature will not exceed the preset first temperature threshold in the future; if the real-time first temperature is greater than or equal to the second temperature threshold, the control module controls the liquid cooling device to maintain its current operating state; if the real-time first temperature is less than the second temperature threshold, the control module controls the liquid cooling device to stop operating; wherein, the first temperature threshold is greater than the second temperature threshold.

[0080] In some embodiments, the control module 303 is further configured to control the liquid cooling device to operate in a second operating state when the real-time first temperature is less than the first temperature threshold and the prediction result is that the first temperature will exceed the preset first temperature threshold in the future; wherein the coolant flow rate of the liquid cooling device in the second operating state is lower than the coolant flow rate of the liquid cooling device in the first operating state.

[0081] In some embodiments, the prediction module 302 is further configured to time the time during which the liquid cooling device operates in the second operating state to obtain a first duration; compare the first duration with a preset cooling duration; and when the first duration reaches the cooling duration, re-predict a new prediction result.

[0082] Specifically, by setting a cooling-off period, repeated predictions can be avoided within a short period of time.

[0083] In some embodiments, the control module 303 is further configured to: obtain the currently used control mode when the real-time first temperature is greater than or equal to a preset first temperature threshold; directly control the liquid cooling device to operate in the first operating state when the control mode is a performance mode; and initiate an over-temperature reminder message when the control mode is an energy-saving mode to remind the user to manually trigger the liquid cooling device to operate in the first operating state.

[0084] Specifically, in performance mode, when the real-time temperature of the computing chip or the temperature of the liquid-cooled microfluidic plate exceeds a first temperature threshold, the micropump of the liquid cooling device is activated by the control unit to drive the coolant to flow through the microfluidic plate and remove heat from the external computing module. In some examples, the coolant used in the liquid cooling device is insulating perfluorohexanone, and the radiator of the liquid cooling device is located on the side of the external computing module's housing.

[0085] In energy-saving mode, when the real-time temperature of the computing chip or the temperature of the liquid-cooled microfluidic plate exceeds a first temperature threshold, a high-temperature warning is displayed on the mobile device or external computing module's screen. The user can then manually trigger the liquid cooling cycle of the liquid cooling device instead of pump-driven operation to reduce power consumption. The manual triggering of the liquid cooling cycle includes methods such as squeezing the liquid reservoir of the liquid cooling device, using a hand-cranked reciprocating plunger pump, or a spring / coil-spring energy-storing micro-pump. In some examples, in energy-saving mode, when the real-time temperature of the computing chip or the temperature of the liquid-cooled microfluidic plate exceeds 70°C, a high-temperature warning is displayed on the mobile device or external computing module's screen. The user can then trigger one cycle of the liquid cooling circuit by squeezing the corresponding pressing area of ​​the flexible liquid reservoir of the liquid cooling device, thus achieving low-energy passive liquid cooling.

[0086] The heat dissipation control device for the external computing module provided in this embodiment, compared with the current air cooling method, uses a liquid cooling device to dissipate heat from the external computing module, resulting in lower noise, higher heat dissipation efficiency, and a more compact structure suitable for mobile scenarios such as external computing modules. Simultaneously, temperature-controlled triggering of the liquid cooling device ensures it only activates when the temperature is too high, reducing unnecessary power consumption and extending the module's battery life. Furthermore, by setting a first and second temperature threshold in the temperature control trigger, a hysteresis zone is formed, preventing frequent start-stop of the liquid cooling device and improving system stability. Moreover, by predicting whether the computing chip's temperature will exceed the first temperature threshold, the liquid cooling device is triggered in advance when this is predicted, maintaining the computing chip's temperature within the optimal range and avoiding frequency throttling caused by instantaneous heat peaks. This application supports temperature-controlled triggering, predictive triggering, and manual triggering for the liquid cooling device's activation control, and includes performance and energy-saving modes to balance performance and energy efficiency requirements.

[0087] It is worth mentioning that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed in this application; however, this does not mean that other units are absent in this embodiment.

[0088] Example 4: Another embodiment of this application relates to a heat dissipation control system for an external computing module. The implementation details of the heat dissipation control device for the external computing module in this embodiment are described below. The following details are provided for ease of understanding and are not essential for implementing this solution. The heat dissipation control system for the external computing module in this embodiment includes a liquid cooling device and a heat dissipation control device for the external computing module as described in Embodiment 3. The liquid cooling device is used to dissipate heat from the external computing module.

[0089] Example 5: Another embodiment of this application relates to an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the heat dissipation control method of the external computing module in the above embodiments.

[0090] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0091] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.

[0092] Example 6: Another embodiment of this application relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the method embodiments described above.

[0093] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0094] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing this application, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of this application.

Claims

1. A heat dissipation control method for an external computing module, characterized in that, The external computing module's heat dissipation device is a liquid cooling device, and the method includes: The system obtains a real-time first temperature and a prediction of whether the first temperature will exceed the first temperature threshold in the future. The first temperature includes the temperature of the computing chip of the external computing module and / or the temperature of the liquid-cooled microfluidic plate of the liquid cooling device. The real-time first temperature is compared with a preset first temperature threshold. When the real-time first temperature is greater than or equal to the preset first temperature threshold, the liquid cooling device is controlled to operate in the first operating state. When the real-time first temperature is less than the first temperature threshold, and the prediction result is that the first temperature will not exceed the preset first temperature threshold in the future, the real-time first temperature is compared with the preset second temperature threshold. When the real-time first temperature is greater than or equal to the second temperature threshold, the liquid cooling device is controlled to maintain its current operating state; when the real-time first temperature is less than the second temperature threshold, the liquid cooling device is controlled to stop operating. Wherein, the first temperature threshold is greater than the second temperature threshold.

2. The heat dissipation control method for the external computing module according to claim 1, characterized in that, Also includes: When the real-time first temperature is less than the first temperature threshold, and the prediction result is that the first temperature will exceed the preset first temperature threshold in the future, the liquid cooling device is controlled to operate in the second operating state. The coolant flow rate of the liquid cooling device in the second operating state is lower than the coolant flow rate of the liquid cooling device in the first operating state.

3. The heat dissipation control method for the external computing module according to claim 2, characterized in that, After the step of controlling the liquid cooling device to operate in the second operating state, the method further includes: The time during which the liquid cooling device operates in the second operating state is timed to obtain a first duration; Compare the first duration with the preset cooling duration; When the first duration reaches the cooling duration, a new prediction result is obtained.

4. The heat dissipation control method for the external computing module according to any one of claims 1 to 3, characterized in that, The step of obtaining a prediction result on whether the first temperature will exceed the first temperature threshold in the future includes: Obtain the task type and computing power requirement of the AI ​​task that the mobile device is about to unload to the external computing power module; Based on the task type and computing power requirements of the obtained AI task, a predictive analysis is performed to predict whether the first temperature will exceed the first temperature threshold in the future, and the prediction result is generated.

5. The heat dissipation control method for the external computing module according to any one of claims 1 to 3, characterized in that, The step of obtaining a prediction result on whether the first temperature will exceed the first temperature threshold in the future includes: Obtain historical variation data of the first temperature; Based on the historical change data of the first temperature, the future trend of the first temperature is predicted by a pre-trained temperature differential model. Based on the predicted future trend of the first temperature, the analysis is performed to determine whether the first temperature will exceed the first temperature threshold in the future, and the prediction result is generated.

6. The heat dissipation control method for the external computing module according to any one of claims 1 to 3, characterized in that, The step of obtaining a prediction result on whether the first temperature will exceed the first temperature threshold in the future includes: Monitor the power consumption of the external computing module; If the current power consumption is greater than the preset peak power consumption and the duration is greater than a predetermined value, a prediction result is generated that the first temperature will exceed the first temperature threshold in the future; otherwise, a prediction result is generated that the first temperature will not exceed the first temperature threshold in the future.

7. The heat dissipation control method for the external computing module according to claim 1, characterized in that, The step of controlling the liquid cooling device to operate in the first operating state when the real-time first temperature is greater than or equal to a preset first temperature threshold includes: Get the currently used control mode; When the control mode is performance mode, the liquid cooling device is directly controlled to operate in the first operating state; When the control mode is energy-saving mode, an over-temperature reminder message is issued to remind the user to manually trigger the liquid cooling device to operate in the first operating state.

8. A heat dissipation control device for an external computing module, characterized in that, The external computing module's heat dissipation device is a liquid cooling device, and the heat dissipation control device includes: A temperature acquisition module is used to acquire a first temperature in real time, the first temperature including the temperature of the computing chip of the external computing module and / or the temperature of the liquid-cooled microfluidic plate of the liquid cooling device. The prediction module is used to predict whether the first temperature will exceed the first temperature threshold in the future and generate prediction results. The control module is configured to acquire a real-time first temperature and the prediction result, and compare the real-time first temperature with a preset first temperature threshold; when the real-time first temperature is greater than or equal to the preset first temperature threshold, control the liquid cooling device to operate in a first operating state; and when the real-time first temperature is less than the first temperature threshold, and the prediction result indicates that the first temperature will not exceed the preset first temperature threshold in the future, compare the real-time first temperature with a preset second temperature threshold; when the real-time first temperature is greater than or equal to the second temperature threshold, control the liquid cooling device to maintain its current operating state; and when the real-time first temperature is less than the second temperature threshold, control the liquid cooling device to stop operating; wherein the first temperature threshold is greater than the second temperature threshold.

9. A heat dissipation control device for an external computing module, characterized in that, The control module is further configured to control the liquid cooling device to operate in a second operating state when the real-time first temperature is less than the first temperature threshold and the prediction result is that the first temperature will exceed the preset first temperature threshold in the future; wherein the coolant flow rate of the liquid cooling device in the second operating state is lower than the coolant flow rate of the liquid cooling device in the first operating state.

10. A heat dissipation control system for an external computing module, characterized in that, It includes a liquid cooling device and a heat dissipation control device for the external computing module as described in claim 8 or 9, wherein the liquid cooling device is used to dissipate heat from the external computing module.

Images