Method and device for evaluating heat dissipation performance of cold plate, storage medium, and electronic device

Abstract

The embodiments of the present application provide a method and device for evaluating the heat dissipation performance of a cold plate, a storage medium, and an electronic device, which relate to the field of computers. The method comprises obtaining a plurality of first temperature values ​​of a central processing unit and a plurality of second temperature values ​​of a graphics processing unit when the cold plate is running; performing a first temperature difference judgment on the central processing unit and the graphics processing unit based on the plurality of first temperature values ​​and the plurality of second temperature values; when the result of the first temperature difference judgment indicates a pass, performing a first temperature standard deviation judgment on the central processing unit and the graphics processing unit based on the plurality of first temperature values ​​and the plurality of second temperature values; when the result of the first temperature standard deviation judgment indicates a pass, performing a first weighted judgment on the central processing unit and the graphics processing unit based on the result of the first temperature difference judgment and the result of the first temperature standard deviation judgment; when the result of the first weighted judgment indicates a pass, determining that the cold plate meets the first heat dissipation performance condition.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computers, and more specifically, to a method and device for evaluating the heat dissipation performance of a cold plate, a storage medium, and an electronic device. Background Art

[0002] Ensuring high standards of heat dissipation performance in liquid cooling systems, particularly cold plates, is crucial during server manufacturing and quality control. Traditional methods often rely on testing at a single temperature threshold, which makes it difficult to comprehensively and accurately assess the cold plate's heat dissipation capabilities, resulting in low accuracy in cold plate heat dissipation performance assessments.

[0003] Therefore, there is a technical problem in the related art that the accuracy of the heat dissipation performance evaluation of the cold plate is low. Summary of the Invention

[0004] The embodiments of the present application provide a method and device for evaluating the heat dissipation performance of a cold plate, a storage medium, and an electronic device, to at least solve the technical problem of low accuracy in evaluating the heat dissipation performance of a cold plate in the related art.

[0005] According to an embodiment of the present application, a method for evaluating the heat dissipation performance of a cold plate is provided, comprising: obtaining a plurality of first temperature values ​​of a central processing unit (CPU) and a plurality of second temperature values ​​of a graphics processing unit (GPU) when the cold plate is in operation, wherein the cold plate is used to dissipate heat for the CPU and the graphics processing unit; when each of the plurality of first temperature values ​​and the plurality of second temperature values ​​is less than a first temperature threshold, performing a first temperature difference judgment on the CPU and the graphics processing unit based on the plurality of first temperature values ​​and the plurality of second temperature values; when a result of the first temperature difference judgment indicates a pass, performing a first temperature standard deviation judgment on the CPU and the graphics processing unit based on the plurality of first temperature values ​​and the plurality of second temperature values; when a result of the first temperature standard deviation judgment indicates a pass, performing a first weighted judgment on the CPU and the graphics processing unit based on a result of the first temperature difference judgment and a result of the first temperature standard deviation judgment; when the result of the first weighted judgment indicates a pass, determining that the cold plate meets a first heat dissipation performance condition.

[0006] According to another embodiment of the present application, a device for evaluating heat dissipation performance of a cold plate is provided, comprising: an acquisition unit, configured to acquire a plurality of first temperature values ​​of a central processing unit (CPU) and a plurality of second temperature values ​​of a graphics processing unit (GPU) when the cold plate is in operation, wherein the cold plate is configured to dissipate heat for the CPU and the graphics processing unit; a first judgment unit, configured to, when each of the plurality of first temperature values ​​and the plurality of second temperature values ​​is less than a first temperature threshold, perform a first temperature difference judgment on the CPU and the graphics processing unit based on the plurality of first temperature values ​​and the plurality of second temperature values; a second judgment unit, configured to, when a result of the first temperature difference judgment indicates a pass, perform a first temperature standard deviation judgment on the CPU and the graphics processing unit based on the plurality of first temperature values ​​and the plurality of second temperature values; a third judgment unit, configured to, when a result of the first temperature standard deviation judgment indicates a pass, perform a first weighted judgment on the CPU and the graphics processing unit based on a result of the first temperature difference judgment and a result of the first temperature standard deviation judgment; and a determination unit, configured to, when a result of the first weighted judgment indicates a pass, determine that the cold plate meets a first heat dissipation performance condition.

[0007] According to another embodiment of the present application, a computer-readable storage medium is provided, in which a computer program is stored. The computer program is configured to execute the steps of any of the above method embodiments when running.

[0008] According to another embodiment of the present application, an electronic device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0009] Through the embodiments provided by the present application, the temperature data of the central processing unit and the graphics processing unit are obtained during operation, and not only is it monitored whether the temperature exceeds the standard, but temperature difference judgment, temperature standard deviation judgment and weighted judgment are further performed in sequence, thereby establishing a comprehensive and accurate multi-dimensional heat dissipation performance evaluation system, which significantly improves the diagnostic accuracy of the cold plate heat dissipation performance, thereby achieving the technical effect of improving the accuracy of the cold plate heat dissipation performance evaluation, and solving the technical problem of low accuracy of the cold plate heat dissipation performance evaluation in related technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 This is a hardware structure block diagram of a method for evaluating the heat dissipation performance of a cold plate according to an embodiment of the present application;

[0011] Figure 2 is a flow chart of a method for evaluating the heat dissipation performance of a cold plate according to an embodiment of the present application;

[0012] Figure 3This is a flowchart for implementing the heat dissipation performance diagnosis of a cold plate of a liquid-cooled server according to an embodiment of the present application;

[0013] Figure 4 This is a flowchart for implementing the heat dissipation performance diagnosis of a cold plate of a liquid-cooled server according to an embodiment of the present application;

[0014] Figure 5 4 is a structural block diagram of a device for evaluating the heat dissipation performance of a cold plate according to an embodiment of the present application. DETAILED DESCRIPTION

[0015] The embodiments of the present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0016] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0017] The method embodiments provided in the embodiments of the present application can be executed in a computer terminal or similar computing device. Taking running on a computer terminal as an example, Figure 1 This is a hardware structure block diagram of a computer terminal for a method for evaluating the heat dissipation performance of a cold plate according to an embodiment of the present application. Figure 1 As shown, the computer terminal may include one or more ( Figure 1 Only one is shown) a processor 102 (the processor 102 may include but is not limited to a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. The computer terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. For example, the computer terminal may also include Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0018] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for determining the mapping relationship in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, implementing the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal via a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0019] Transmission device 106 is used to receive or transmit data via a network. A specific example of the aforementioned network may include a wireless network provided by a computer terminal's communications provider. In one embodiment, transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0020] As an optional solution, the cold plate heat dissipation performance evaluation method includes the following steps:

[0021] S202, obtaining a plurality of first temperature values ​​of the central processing unit and a plurality of second temperature values ​​of the graphics processing unit when the cold plate is in operation, wherein the cold plate is used to dissipate heat for the central processing unit and the graphics processing unit;

[0022] S204, when each of the plurality of first temperature values ​​and the plurality of second temperature values ​​is less than the first temperature threshold, determining a first temperature difference between the central processing unit and the graphics processing unit based on the plurality of first temperature values ​​and the plurality of second temperature values;

[0023] S206 , when the result of the first temperature difference determination indicates a pass, performing a first temperature standard deviation determination on the CPU and the GPU based on the plurality of first temperature values ​​and the plurality of second temperature values;

[0024] S208, when the result of the first temperature standard deviation judgment indicates a pass, performing a first weighted judgment on the central processing unit and the graphics processing unit according to the result of the first temperature difference judgment and the result of the first temperature standard deviation judgment;

[0025] S210 , when the result of the first weighted judgment indicates a pass, determining that the cold plate meets the first heat dissipation performance condition.

[0026] Optionally, in this embodiment, the multiple first temperature values ​​and the multiple second temperature values ​​may be, but are not limited to, multiple temperature readings collected from a central processing unit (hereinafter referred to as CPU) and a graphics processing unit (hereinafter referred to as GPU) when the cold plate is operating under a first load condition (low load condition), and may be, but are not limited to, data collected multiple times continuously at a first sampling frequency.

[0027] Optionally, in this embodiment, the first temperature threshold is a safe temperature upper limit set for the CPU and the GPU. Exceeding this upper limit will be regarded as insufficient heat dissipation performance.

[0028] Optionally, in this embodiment, the first temperature difference judgment is used to evaluate whether the maximum difference between the surface temperatures of the CPU and GPU under low load conditions meets a preset limit, and is used to check the uniformity of temperature distribution.

[0029] Optionally, in this embodiment, the first temperature standard deviation is used to calculate the standard deviation of the CPU and GPU temperature data to evaluate the stability of the heat dissipation performance, that is, the degree of temperature fluctuation.

[0030] Optionally, in this embodiment, the first weighted judgment is used to combine the evaluation results of the temperature difference and the temperature standard deviation, and perform a comprehensive analysis using predefined weights to determine whether the overall heat dissipation performance of the cold plate meets the first heat dissipation performance standard.

[0031] Optionally, in this embodiment, multiple first temperature values ​​(CPU temperature) and multiple second temperature values ​​(GPU temperature) obtained under current load conditions (e.g., low load conditions) are analyzed to determine whether each reading is below a preset first temperature threshold. If all readings do not exceed the first temperature threshold, the cold plate has preliminarily passed the temperature safety check, and the next step can be determined.

[0032] For example, the preset temperature threshold is 90°C. During the low-load test, the temperature readings of the CPU and GPU remained below 85°C and did not exceed the limit.

[0033] Further analysis is performed on temperature data that passes the temperature threshold check, calculating the maximum difference between the CPU and GPU surface temperatures and determining whether this difference is less than the temperature difference limit defined in the configuration file. For example, the temperature difference between each CPU measurement point and each GPU measurement point is calculated, and by comparing these differences with the limit, the uniformity of the cold plate heat dissipation is evaluated.

[0034] For example, if the CPU and GPU temperature difference limits are set to 5°C and 7°C, respectively, under low load conditions, the maximum temperature differences between the CPU and GPU are 3°C and 6°C, respectively, which are within the limits, and the temperature difference judgment passes.

[0035] Calculate the standard deviation of the CPU and GPU temperature data to check the stability of the thermal performance, that is, the degree of fluctuation in the temperature readings. If the standard deviation of the temperature data is less than the standard deviation limit defined in the configuration file, the cold plate meets the requirements for temperature stability.

[0036] For example, the standard deviation limit for CPU temperature data is set to 2°C, while that for GPU is 3°C. Under low load conditions, the standard deviations for CPU and GPU temperatures are 1.7°C and 2.5°C, respectively, both within the limits and passing the standard deviation judgment.

[0037] The results of the temperature difference and standard deviation are weighted and combined to form a comprehensive evaluation metric. For example, if the temperature difference and standard deviation are weighted 0.6 and 0.4, respectively, the cold plate's overall thermal performance score will be calculated based on the evaluation results of these two metrics. If the combined score meets the preset multi-metric fusion limit, the cold plate will be deemed to meet the thermal performance standard under low load conditions.

[0038] For example, assuming the temperature difference judgment score (based on comparison with the limit) is 80 and the temperature standard deviation judgment score (based on comparison with the limit) is 90, with weights of 0.6 and 0.4, respectively, the combined score is calculated as: 80*0.6 + 90*0.4 = 84. If the multi-indicator fusion limit is set at 85, since the combined score of 84 is slightly below the limit, the cold plate design needs to be reassessed or adjusted.

[0039] Finally, if the cold plate performs well in each step of the temperature threshold, temperature difference, temperature standard deviation, and the first weighted judgment and meets all set standards, it can be determined that the heat dissipation performance of the cold plate under the current load conditions meets the first heat dissipation performance standard.

[0040] The embodiments provided herein utilize multi-stage temperature data evaluation to ensure that the cold plate not only operates within safe temperature thresholds but also maintains good temperature distribution uniformity and stability, comprehensively considering multiple aspects of heat dissipation performance. This process makes cold plate evaluation more scientific and comprehensive, avoiding the potential bias caused by single-metric evaluation. This ensures that when the server is operating at low load, the cold plate can effectively manage heat, prevent overheating, and maintain the normal operation of the CPU and GPU, thereby ensuring server stability and extending hardware life.

[0041] As an optional solution, determining a first temperature difference between the CPU and the GPU based on a plurality of first temperature values ​​and a plurality of second temperature values ​​includes:

[0042] Determining a first temperature value of each of the central processing units from the plurality of first temperature values, and determining a second temperature value of each of the graphics processing units from the plurality of second temperature values, wherein the first temperature value of each of the central processing units is used to determine an average temperature value of each of the central processing units, and the second temperature value of each of the graphics processing units is used to determine an average temperature value of each of the graphics processing units;

[0043] Determine a first temperature difference parameter corresponding to the central processing unit according to a first average temperature value having the highest average temperature value and a second average temperature value having the lowest average temperature value among the average temperature values ​​of the central processing units;

[0044] Determining a second temperature difference parameter corresponding to the graphics processor according to a third average temperature value having the highest average temperature value and a fourth average temperature value having the lowest average temperature value among the average temperature values ​​of the graphics processors;

[0045] In a case where the first temperature difference parameter is smaller than the temperature difference threshold and the second temperature difference parameter is smaller than the temperature difference threshold, it is determined that the result of the first temperature difference judgment indicates a pass.

[0046] Optionally, in this embodiment, the first temperature difference parameter and the second temperature difference parameter represent the ratio of the difference between the highest and lowest average temperatures of the CPU and GPU, respectively, under a first load condition, and are used to evaluate the uniformity of the cold plate's heat dissipation performance. The temperature difference threshold is used to determine whether the temperature difference parameter is within an acceptable range, thereby determining whether the cold plate's heat dissipation performance meets standards.

[0047] Optionally, in this embodiment, temperature data belonging to each CPU is extracted from multiple first temperature values, and the average temperature value of each CPU is calculated; similarly, temperature data belonging to each GPU is extracted from multiple second temperature values, and the average temperature value of each GPU is calculated.

[0048] For example, under low load conditions, the system monitored continuous temperature data for four CPUs and two GPUs, collecting data every minute for a total of 60 times. For each CPU, we obtained 60 temperature readings and calculated the average temperature value. For example, the average temperature of the first CPU was 85°C, the second was 84°C, and so on. Similarly, for the GPUs, the average temperature of the first GPU was 80°C, and the second was 82°C.

[0049] Find the highest and lowest average temperature values ​​for the CPU and GPU, and record them as the first and second average temperature values ​​for the CPU, and the third and fourth average temperature values ​​for the GPU. Then, calculate the first and second temperature difference parameters, which are the ratios of the differences between the highest and lowest average temperature values.

[0050] For example, suppose the average temperature of CPU 1 is 85°C (the first average temperature value), while the average temperature of CPU 2 is 84°C (the second average temperature value). The first temperature difference parameter is calculated as (85-84) / 84.5, which is approximately 1.18%. Similarly, suppose the average temperature of GPU 1 is 82°C (the third average temperature value), while the average temperature of GPU 2 is 80°C (the fourth average temperature value). The second temperature difference parameter is calculated as (82-80) / 81, which is approximately 2.47%.

[0051] The calculated first and second temperature difference parameters are compared with the temperature difference thresholds set in the configuration file. If both parameters are less than the thresholds, the cold plate has uniform heat dissipation under the current load conditions and there is no severe local overheating.

[0052] Through the embodiments provided in the present application, by comparing the first temperature difference parameter, the second temperature difference parameter and the temperature difference threshold, design or manufacturing deficiencies can be accurately identified, ensuring that the cold plate heat dissipation performance of the liquid-cooled server reaches the expected consistency and stability standards before actual deployment.

[0053] As an optional solution, determining the first temperature difference parameter corresponding to the central processor according to the first average temperature value with the highest average temperature value and the second average temperature value with the lowest average temperature value among the average temperature values ​​of the central processors includes:

[0054] Accumulating average temperature values ​​of each central processing unit to obtain a first accumulated temperature value;

[0055] Divide the first difference between the first average temperature value and the second average temperature value by the first accumulated temperature value to obtain a first temperature difference parameter;

[0056] Determining a second temperature difference parameter corresponding to the graphics processor according to a third average temperature value having the highest average temperature value and a fourth average temperature value having the lowest average temperature value among the average temperature values ​​of the graphics processors includes:

[0057] Accumulating average temperature values ​​of each graphics processor to obtain a second accumulated temperature value;

[0058] A second temperature difference parameter is obtained by dividing a second difference between the third average temperature value and the fourth average temperature value by the second accumulated temperature value.

[0059] Optionally, in this embodiment, the first accumulated temperature value is the sum of the average temperature values ​​of all CPUs under low load conditions, which is used to calculate the base of the first temperature difference parameter. The second accumulated temperature value is the sum of the average temperature values ​​of all GPUs under low load conditions, which is used to calculate the base of the second temperature difference parameter.

[0060] Optionally, in this embodiment, the first difference and the second difference respectively represent the difference between the highest average temperature value and the lowest average temperature value in the CPU (the difference between the first average temperature value and the second average temperature value), and the difference between the highest average temperature value and the lowest average temperature value in the GPU (the difference between the third average temperature value and the fourth average temperature value).

[0061] Optionally, in this embodiment, the first temperature difference parameter and the second temperature difference parameter are calculated by calculating the ratio of the first difference and the second difference to their corresponding accumulated temperature values, so as to quantitatively evaluate the consistency of the surface temperature distribution of the CPU and the GPU.

[0062] Optionally, in this embodiment, the average temperature value of each CPU is added to obtain a first accumulated temperature value; similarly, the average temperature value of each GPU is added to obtain a second accumulated temperature value.

[0063] For example, assuming the average temperatures of the four CPUs under low load are 85°C, 84°C, 86°C, and 83°C, respectively, the first cumulative temperature value is 85+84+86+83=338°C. Meanwhile, if the average temperatures of the two GPUs are 82°C and 80°C, respectively, the second cumulative temperature value is 82+80=162°C.

[0064] Calculate the first difference (i.e., the difference between the first and second average temperature values, here, the difference between the highest and lowest average temperature values) and divide it by the first accumulated temperature value to obtain the first temperature difference parameter. For GPUs, calculate the second difference (i.e., the difference between the third and fourth average temperature values) and divide it by the second accumulated temperature value to obtain the second temperature difference parameter.

[0065] For example, based on the above steps, the CPU's highest average temperature is 86°C (the first average temperature value), and its lowest average temperature is 83°C (the second average temperature value). The first difference is 86 - 83, which is 3°C. The first temperature difference parameter is calculated as (first difference / first accumulated temperature value) × 100% = (3 / 338) × 100% ≈ 0.89%. For the GPU, the highest average temperature is 82°C (the third average temperature value), and the lowest average temperature is 80°C (the fourth average temperature value). The second difference is 82 - 80, which is 2°C. The second temperature difference parameter is calculated as (second difference / second accumulated temperature value) × 100% = (2 / 162) × 100% ≈ 1.23%.

[0066] The examples provided in this application further refine the calculation process for the temperature difference parameter. By accumulating average temperature values ​​to form a reference baseline, the ratio of the difference to the accumulated value is calculated to quantitatively assess the uniformity of the CPU and GPU surface temperature distribution. This parameter calculation method not only considers the influence of the maximum temperature difference but also places it under the sum of all average temperature values ​​for evaluation. This more accurately reflects the consistency of the cold plate's heat dissipation performance and avoids the one-sidedness that can result from relying solely on the maximum temperature difference.

[0067] As an optional solution, judging the first temperature standard deviation of the central processing unit and the graphics processing unit based on the multiple first temperature values ​​and the multiple second temperature values ​​includes:

[0068] When the average temperature values ​​of the respective CPUs in the CPU are obtained, a standard deviation operation is performed on the average temperature values ​​of the respective CPUs to obtain a first temperature standard deviation parameter;

[0069] When the average temperature values ​​of the respective graphics processors in the graphics processors are obtained, a standard deviation operation is performed on the average temperature values ​​of the respective graphics processors to obtain a second temperature standard deviation parameter;

[0070] When the first temperature standard deviation parameter is less than the temperature standard deviation threshold, and the second temperature standard deviation parameter is less than the temperature standard deviation threshold, it is determined that the result of the first temperature standard deviation judgment indicates a pass.

[0071] Optionally, in this embodiment, the first temperature standard deviation parameter is the standard deviation of the average temperature values ​​of all CPUs under the current load conditions, which is used to assess the volatility of the CPU surface temperature and the consistency of the cooling system. The second temperature standard deviation parameter is the standard deviation of the average temperature values ​​of all GPUs under the same load conditions, which is used to assess the stability of the GPU surface temperature and the uniformity of the cooling system. The temperature standard deviation threshold is a maximum allowable standard deviation value defined in the configuration file and is used to determine whether the temperature fluctuations of the CPU and GPU are within an acceptable range.

[0072] Optionally, in this embodiment, after obtaining the average temperature values ​​of the CPU and the GPU, standard deviations of these values ​​are calculated respectively to obtain a first temperature standard deviation parameter and a second temperature standard deviation parameter.

[0073] For example, assume that under low load conditions, the average temperatures of four CPUs are 85°C, 84°C, 86°C, and 83°C, respectively. The standard deviation of these values ​​is a measure of their volatility. Assume the calculated first temperature standard deviation parameter is 1.5°C. For GPUs, assume that the average temperatures of two GPUs are 82°C and 80°C, respectively, and the calculated second temperature standard deviation parameter is 1.4°C.

[0074] Compare the calculated first and second temperature standard deviation parameters to the temperature standard deviation thresholds in the configuration file. If both parameters are less than or equal to the thresholds, the CPU and GPU temperature fluctuations are within normal ranges, and the cold plate's cooling performance is stable and uniform.

[0075] For example, assume that the temperature standard deviation threshold set in the configuration file is 2° C. Since the first temperature standard deviation parameter of 1.5° C. is less than the threshold, and the second temperature standard deviation parameter of 1.4° C. is also less than the threshold, the result of the temperature standard deviation judgment indicates a pass.

[0076] Through the embodiments provided in this application, by calculating the first and second temperature standard deviation parameters, the fluctuation of the CPU and GPU surface temperatures can be quantitatively evaluated to ensure that the cold plate can provide a stable heat dissipation effect and avoid potential problems caused by excessive temperature fluctuations, such as local overheating or unstable heat dissipation efficiency.

[0077] As an optional solution, a first weighted judgment is performed on the CPU and the GPU based on the temperature difference judgment result and the temperature standard deviation judgment result, including:

[0078] Obtaining a first temperature difference parameter of the central processing unit and a second temperature difference parameter of the graphics processing unit based on a result of the first temperature difference judgment, and obtaining a first temperature standard deviation parameter of the central processing unit and a second temperature standard deviation parameter of the graphics processing unit based on a result of the first temperature standard deviation judgment;

[0079] Performing a first weighted fusion operation on the first temperature difference parameter and the first temperature standard deviation parameter to obtain a first comprehensive parameter of the central processing unit;

[0080] performing a second weighted fusion operation on the second temperature difference parameter and the second temperature standard deviation parameter to obtain a second comprehensive parameter of the graphics processor;

[0081] When the first comprehensive parameter is less than the comprehensive parameter threshold and the second comprehensive parameter is less than the comprehensive parameter threshold, it is determined that the result of the first weighted judgment indicates a pass.

[0082] Optionally, in this embodiment, the first comprehensive parameter and the second comprehensive parameter are results obtained by weighted fusion of the first and second temperature difference parameters and the first and second temperature standard deviation parameters, and are used to comprehensively evaluate the heat dissipation performance of the CPU and GPU under low load conditions.

[0083] Optionally, in this embodiment, the comprehensive parameter threshold is a numerical standard used to determine whether the first and second comprehensive parameters indicate that the heat dissipation performance of the cold plate is qualified.

[0084] Optionally, in this embodiment, first, based on the temperature difference determination result, a first temperature difference parameter for each CPU and a second temperature difference parameter for each GPU are obtained. Then, based on the temperature standard deviation determination result, a first temperature standard deviation parameter for each CPU and a second temperature standard deviation parameter for each GPU are obtained.

[0085] For example, assume that after temperature difference judgment, the first temperature difference parameter of the CPU is obtained to be 1.0% and the second temperature difference parameter of the GPU is obtained to be 1.2%; after temperature standard deviation judgment, the first temperature standard deviation parameter of the CPU is obtained to be 1.5°C and the second temperature standard deviation parameter of the GPU is obtained to be 1.4°C.

[0086] The first temperature difference parameter and the first temperature standard deviation parameter are weightedly fused to form a first comprehensive parameter. The second temperature difference parameter and the second temperature standard deviation parameter are weightedly fused to form a second comprehensive parameter. The weighted fusion can assign different weights to different parameters based on their importance.

[0087] For example, assume that when calculating the first comprehensive parameter, the weight of the first temperature difference parameter is 0.7, and the weight of the first temperature standard deviation parameter is 0.3. When calculating the second comprehensive parameter, the weight of the second temperature difference parameter is also 0.7, and the weight of the second temperature standard deviation parameter is 0.3. Therefore, the first comprehensive parameter = 1.0% × 0.7 + 1.5°C × 0.3; the second comprehensive parameter = 1.2% × 0.7 + 1.4°C × 0.3.

[0088] The first and second comprehensive parameters are respectively compared with the comprehensive parameter thresholds set in the configuration file. If both comprehensive parameters are less than the thresholds, the result of the first weighted judgment indicates a pass, indicating that the comprehensive evaluation of the heat dissipation performance of the CPU and GPU under low load conditions meets the standards.

[0089] For example, assuming the comprehensive parameter thresholds set in the configuration file are 2.0% and 2.5°C, the first comprehensive parameter calculated in step 2 is approximately 1.35%, and the second comprehensive parameter is approximately 1.18% + 0.42°C. Considering the nature of the weighted fusion parameters, if they are treated as a comprehensive percentage, approximate calculations or direct comparisons of the values ​​are possible. In this case, both the first and second comprehensive parameters are less than their corresponding thresholds, so the first weighted judgment result indicates a pass.

[0090] Through the embodiments provided in this application, a comprehensive and quantitative heat dissipation performance evaluation index is provided by fusing the temperature difference parameter and the temperature standard deviation parameter. The temperature difference parameter reflects the consistency of temperature at different locations, while the temperature standard deviation parameter measures the stability of the temperature data. By weightedly fusing these two parameters, the overall level of cold plate heat dissipation performance can be more accurately assessed, ensuring that under low-load conditions, the heat dissipation of the CPU and GPU is both uniform and stable, achieving the expected heat dissipation performance requirements.

[0091] As an optional solution, a first weighted fusion operation is performed on the first temperature difference parameter and the first temperature standard deviation parameter to obtain a first comprehensive parameter of the central processing unit, including:

[0092] Obtaining a first weighting parameter corresponding to the central processing unit and a second weighting parameter corresponding to the graphics processing unit, and obtaining a weighted parameter and value of the first weighting parameter and the second weighting parameter;

[0093] Obtaining a first cumulative parameter obtained by multiplying the first weighted parameter by the first temperature difference parameter, and obtaining a second cumulative parameter obtained by multiplying the second weighted parameter by the first temperature standard deviation parameter;

[0094] Determine the result of dividing the accumulated value of the first accumulated parameter and the second accumulated parameter by the sum of the weighted parameters as the first comprehensive parameter;

[0095] Performing a second weighted fusion operation on the second temperature difference parameter and the second temperature standard deviation parameter to obtain a second comprehensive parameter of the graphics processor includes:

[0096] Obtaining a third cumulative parameter obtained by multiplying the first weighted parameter by the second temperature difference parameter, and obtaining a fourth cumulative parameter obtained by multiplying the second weighted parameter by the second temperature standard deviation parameter;

[0097] The result of dividing the accumulated value of the third accumulated parameter and the fourth accumulated parameter by the weighted parameter sum is determined as the second comprehensive parameter.

[0098] Optionally, in this embodiment, the first weighting parameter * the second weighting parameter is a pre-set weighting value when calculating the first comprehensive parameter and the second comprehensive parameter, used to highlight the importance of one parameter relative to the other. The weighted parameter sum is the sum of the first weighting parameter and the second weighting parameter, and is used to normalize the calculation results during the weighted fusion process.

[0099] Optionally, in this embodiment, a first weighting parameter (for the first temperature difference parameter) and a second weighting parameter (for the first temperature standard deviation parameter) are preset for the CPU. For example, the first weighting parameter is set to 0.7 and the second weighting parameter is set to 0.3, ensuring that the temperature difference is more important than the temperature standard deviation.

[0100] Obtain a weighted parameter sum value, add the first weighted parameter to the second weighted parameter, and obtain a weighted parameter sum value. In this example, the weighted parameter sum value = 0.7 + 0.3 = 1.

[0101] The first cumulative parameter is obtained by multiplying the first weighted parameter by the first temperature difference parameter, and the second cumulative parameter is obtained by multiplying the second weighted parameter by the first temperature standard deviation parameter. For example, if the first temperature difference parameter is 1.0%, the first cumulative parameter = 0.7 × 1.0% = 0.7%; if the first temperature standard deviation parameter is 1.5°C, the second cumulative parameter = 0.3 × 1.5°C = 0.45°C.

[0102] Divide the sum of the first and second cumulative parameters by the weighted parameter sum to obtain the first comprehensive parameter. In this example, the first comprehensive parameter = (0.7% + 0.45°C) / 1. Because the parameters are in different units, it is usually necessary to convert them to the same unit (such as percentage) for calculation, or take other appropriate standardization measures.

[0103] Similarly, for the GPU, set the first weighting parameter (for the second temperature difference parameter) and the second weighting parameter (for the second temperature standard deviation parameter). For example, set the first weighting parameter to 0.7 and the second weighting parameter to 0.3. The sum of the weighted parameters is also 1 (0.7 + 0.3 = 1), which is consistent with the CPU calculation. The third cumulative parameter is obtained by multiplying the first weighting parameter by the second temperature difference parameter, and the fourth cumulative parameter is obtained by multiplying the second weighting parameter by the second temperature standard deviation parameter. For example, if the second temperature difference parameter is 1.2%, the third cumulative parameter = 0.7 × 1.2% = 0.84%; if the second temperature standard deviation parameter is 1.4°C, the fourth cumulative parameter = 0.3 × 1.4°C = 0.42°C. Divide the sum of the third and fourth cumulative parameters by the weighted parameter sum to obtain the second comprehensive parameter. Similar to the calculation of the first comprehensive parameter, ensure that the parameter units are consistent or implement appropriate standardization.

[0104] Through the embodiments provided in this application, by calculating the first and second comprehensive parameters and comparing them with the comprehensive parameter thresholds, it is possible to determine whether the heat dissipation performance of the cold plate under low-load conditions meets high standards, ensuring that the server can maintain a good level of thermal management in actual applications, especially in environments with strict requirements on heat dissipation such as high-performance computing and data centers. The introduction of this evaluation mechanism greatly improves the effectiveness and accuracy of the diagnostic system.

[0105] As an optional solution, before determining the first temperature difference between the CPU and the GPU based on the multiple first temperature values ​​and the multiple second temperature values, the method further includes:

[0106] Get the target material type of the cold plate;

[0107] The temperature difference threshold corresponding to the target material type, the temperature standard deviation threshold corresponding to the target material type, and the comprehensive parameter threshold corresponding to the target material type are determined from the configuration file, wherein the temperature difference threshold is used for the first temperature difference judgment, the temperature standard deviation threshold is used for the first temperature standard deviation judgment, and the comprehensive parameter threshold is used for the first weighted judgment. The configuration file stores the correspondence between multiple groups of material types and thresholds.

[0108] Optionally, in this embodiment, the target material type refers to the material type of the liquid-cooled server cold plate being tested, and different material types correspond to different heat dissipation performance requirements and standards.

[0109] Optionally, in this embodiment, the configuration file stores a file of the correspondence between multiple cold plate material types and various thresholds, including temperature difference threshold, temperature standard deviation threshold and comprehensive parameter threshold, which is used to guide parameter acquisition and comparison in the heat dissipation performance judgment process.

[0110] Optionally, in this embodiment, the temperature difference threshold, the temperature standard deviation threshold and the comprehensive parameter threshold are heat dissipation performance standards for specific material types defined in the configuration file, and are used for the first temperature difference judgment, the first temperature standard deviation judgment and the first weighted judgment, respectively, to ensure that the heat dissipation performance of the liquid-cooled server cold plate meets the design requirements.

[0111] Optionally, in this embodiment, before starting to judge the heat dissipation performance, it is first necessary to determine the material type of the cold plate to be tested. This step is crucial for subsequently calling the correct threshold value from the configuration file.

[0112] For example, assuming that the cold plate model is V0020FZ, the target material type is V0020FZ.

[0113] Once the target material type is determined, the temperature difference threshold, temperature standard deviation threshold and comprehensive parameter threshold related to the material type are retrieved from the configuration file to provide a benchmark for the upcoming heat dissipation performance judgment.

[0114] After obtaining the threshold corresponding to the target material type, the first heat dissipation performance of the cold plate can be judged based on the collected first temperature data and second temperature data, and the heat dissipation performance of the cold plate can be evaluated by comparing with the threshold.

[0115] Through the embodiments provided in this application, the importance of obtaining the target material type of the cold plate and its corresponding threshold value before making a judgment on the heat dissipation performance is demonstrated. This process ensures the pertinence and accuracy of the heat dissipation performance test and avoids the deviation in test results that may be caused by the use of universal thresholds. Through the configuration file, a series of heat dissipation performance standards that match the target material type can be quickly located and called before the test begins, including temperature difference thresholds, temperature standard deviation thresholds, and comprehensive parameter thresholds. This not only simplifies the test process and reduces manual errors, but also can adapt to the specific heat dissipation requirements of cold plates of different material types, thereby improving the efficiency and accuracy of the test.

[0116] As an optional solution, the method further includes:

[0117] When there is a temperature value greater than or equal to a preset threshold value among the multiple first temperature values ​​and the multiple second temperature values, or when the result of the first temperature difference judgment indicates failure, or when the result of the first temperature standard deviation judgment indicates failure, or when the result of the first weighted judgment indicates failure, it is determined that the result of the first heat dissipation performance judgment indicates that the cold plate does not meet the first heat dissipation performance standard.

[0118] Optionally, in this embodiment, the collected plurality of first temperature values ​​(average CPU temperature) and second temperature values ​​(average GPU temperature) are checked to see whether any temperature value exceeds a preset threshold. If so, it is preliminarily determined that there may be a serious problem with the heat dissipation performance of the cold plate, failing to meet the heat dissipation requirements, and the test result will indicate that it has failed the first heat dissipation performance standard.

[0119] After the calculation of the temperature difference parameter is completed, if the first temperature difference parameter exceeds the temperature difference threshold defined for the target material type in the configuration file, the result of the first temperature difference judgment also indicates failure, indicating that the surface temperature distribution of the cold plate is uneven and the heat dissipation performance is poor.

[0120] After calculating the temperature standard deviation parameters, if the first temperature standard deviation parameter or the second temperature standard deviation parameter exceeds the corresponding temperature standard deviation threshold, this indicates that the temperature fluctuation of the CPU or GPU is large and the stability of the cooling system is insufficient. The result of the temperature standard deviation judgment also indicates failure.

[0121] When the first comprehensive parameter or the second comprehensive parameter is compared with the comprehensive parameter threshold in the configuration file, if any comprehensive parameter exceeds the threshold, the result of the first weighted judgment indicates failure, which means that after comprehensively considering the temperature difference and the temperature standard deviation, the overall heat dissipation performance of the cold plate fails to reach the qualified level.

[0122] The embodiments provided in this application utilize a series of conditional assessments to ensure that cold plates can be accurately evaluated and effectively identified when facing various cooling challenges, including high temperatures, uneven temperature distribution, large temperature fluctuations, and poor overall performance. This design not only enhances the reliability of test results and prevents potential cooling performance issues from being overlooked, but also provides a multi-level assessment mechanism, making the testing process more rigorous and comprehensive.

[0123] As an optional solution, obtaining multiple first temperature values ​​of the central processing unit and multiple second temperature values ​​of the graphics processing unit when the cold plate is running includes:

[0124] acquiring, at a first sampling frequency, a plurality of first temperature values ​​of the central processing unit and a plurality of second temperature values ​​of the graphics processing unit when the cold plate operates under a first load condition;

[0125] After determining that the cold plate meets the first heat dissipation performance condition, the method further includes:

[0126] According to a second sampling frequency, a plurality of third temperature values ​​of the central processing unit and a plurality of fourth temperature values ​​of the graphics processing unit are obtained when the cold plate operates under the second load condition, wherein the second sampling frequency is higher than the first sampling frequency, and the plurality of third temperature values ​​and the plurality of fourth temperature values ​​are used to determine whether the cold plate meets the second heat dissipation performance condition.

[0127] Optionally, in this embodiment, the first sampling frequency and the second sampling frequency are used to indicate the speed of data collection, expressed as the number of readings per unit time. In this embodiment, the first sampling frequency is used for temperature data collection under low load conditions, while the second sampling frequency is used under high load conditions. Its frequency is higher than the first sampling frequency to obtain more dense temperature data points and more detailed monitoring of temperature changes under heavy load.

[0128] Optionally, in this embodiment, the plurality of first temperature values ​​and the plurality of second temperature values ​​are temperature readings collected for the CPU and GPU under low-load conditions, typically data collected multiple times continuously at a first sampling frequency. The plurality of third temperature values ​​and the plurality of fourth temperature values ​​are temperature readings collected for the CPU and GPU under high-load conditions, but collected at a higher second sampling frequency to ensure more timely and comprehensive monitoring of temperature changes at performance limits.

[0129] Optionally, in this embodiment, when the server is running at low load, a lower first sampling frequency (e.g., once every minute) is used to read the temperature data of the CPU and GPU. This can provide a stable temperature baseline for judging the cooling performance of the cold plate under light load.

[0130] For example, the server is powered on and runs a lightweight task, and the CPU and GPU temperatures are read through the IPMI interface every minute for one hour. In this way, 60 sets of first and second temperature values ​​are obtained, including each reading of the CPU temperature and each reading of the GPU temperature.

[0131] Once the cold plate passes the initial low-load test, the system enters the high-load test phase. During this phase, a higher second sampling frequency (e.g., every 30 seconds) is used to collect CPU and GPU temperature data to more accurately monitor and capture temperature changes under high power consumption.

[0132] For example, the server enters high-load mode and starts executing high-performance computing tasks. At the same time, the system reads the temperature data of the CPU and GPU every 30 seconds for 10 hours. In this way, 1,200 sets of third and fourth temperature values ​​are obtained, including each CPU and GPU temperature reading during the high-load test.

[0133] Through the embodiments provided in this application, different sampling frequencies are set to evaluate the heat dissipation performance of the cold plate under low-load and high-load conditions. This method can more comprehensively reflect the heat dissipation efficiency of the cold plate. Under low-load conditions, since temperature changes are relatively gradual, using a relatively low first sampling frequency to collect temperature data can not only meet the needs of performance evaluation, but also reduce the amount of data processing and improve test efficiency. Under high-load conditions, since temperature changes may be more drastic and rapid, using a higher second sampling frequency can more accurately capture instantaneous temperature changes, thereby more accurately evaluating the heat dissipation capacity and stability of the cold plate in high-power consumption scenarios.

[0134] As an optional solution, after obtaining, at the second sampling frequency, a plurality of third temperature values ​​of the central processing unit and a plurality of fourth temperature values ​​of the graphics processing unit when the cold plate operates under the second load condition, the method further includes:

[0135] When each of the plurality of third temperature values ​​and the plurality of fourth temperature values ​​is less than the second temperature threshold, performing a second temperature difference determination on the central processing unit and the graphics processing unit according to the plurality of third temperature values ​​and the plurality of fourth temperature values;

[0136] If the result of the second temperature difference determination indicates a pass, performing a second temperature standard deviation determination on the central processing unit and the graphics processing unit based on the plurality of third temperature values ​​and the plurality of fourth temperature values;

[0137] If the result of the second temperature standard deviation judgment indicates a pass, performing a second weighted judgment on the central processing unit and the graphics processing unit according to the result of the second temperature difference judgment and the result of the second temperature standard deviation judgment;

[0138] If the result of the second weighted determination indicates a pass, the cold plate is determined to meet the second heat dissipation performance condition. Optionally, in this embodiment, the plurality of third temperature values ​​and the plurality of fourth temperature values ​​may be, but are not limited to, a plurality of temperature readings collected from the central processing unit and the graphics processing unit when the cold plate is operating under the second load condition (high load condition), and may be, but are not limited to, data collected multiple times continuously at the second sampling frequency.

[0139] Optionally, in this embodiment, the second temperature threshold is a safety temperature upper limit set for the CPU and GPU. Exceeding this upper limit will be regarded as insufficient heat dissipation performance. The second temperature threshold may be, but is not limited to, higher than the first temperature threshold.

[0140] Optionally, in this embodiment, the second temperature difference judgment is used to evaluate whether the maximum difference between the surface temperatures of the CPU and GPU under high load conditions meets a preset limit, and is used to check the uniformity of temperature distribution.

[0141] Optionally, in this embodiment, the second temperature standard deviation is used to calculate the standard deviation of the CPU and GPU temperature data to evaluate the stability of the heat dissipation performance, that is, the degree of temperature fluctuation.

[0142] Optionally, in this embodiment, the second weighted judgment is used to combine the evaluation results of the temperature difference and the temperature standard deviation, and perform a comprehensive analysis using predefined weights to determine whether the overall heat dissipation performance of the cold plate meets the second heat dissipation performance standard.

[0143] Optionally, in this embodiment, multiple third temperature values ​​(CPU temperature) and multiple fourth temperature values ​​(GPU temperature) obtained under the current load condition (high load condition) are analyzed to determine whether each reading is below a preset second temperature threshold. If all readings do not exceed the second temperature threshold, the cold plate has preliminarily passed the temperature safety check and the next step can be determined.

[0144] Further analysis is performed on temperature data that passes the temperature threshold check, calculating the maximum difference between the CPU and GPU surface temperatures and determining whether this difference is less than the temperature difference limit defined in the configuration file. For example, the temperature difference between each CPU measurement point and each GPU measurement point is calculated, and by comparing these differences with the limit, the uniformity of the cold plate heat dissipation is evaluated.

[0145] Calculate the standard deviation of the CPU and GPU temperature data to check the stability of the thermal performance, that is, the degree of fluctuation in the temperature readings. If the standard deviation of the temperature data is less than the standard deviation limit defined in the configuration file, the cold plate meets the requirements for temperature stability.

[0146] The results of the temperature difference and standard deviation are weighted and combined to form a comprehensive evaluation metric. For example, if the temperature difference and standard deviation are weighted 0.6 and 0.4, respectively, the cold plate's overall thermal performance score will be calculated based on the evaluation results of these two metrics. If the combined score meets the preset multi-metric fusion limit, the cold plate will be deemed to meet the thermal performance standard under low load conditions.

[0147] Finally, if the cold plate performs well in each step of the temperature threshold, temperature difference, temperature standard deviation, and the second weighted judgment and meets all set standards, it can be determined that the heat dissipation performance of the cold plate under the current load conditions meets the first heat dissipation performance standard.

[0148] Through the embodiments provided in the present application, a comprehensive evaluation of the heat dissipation performance of the cold plate is achieved by collecting temperature data of the central processing unit (CPU) and the graphics processing unit (GPU) under low load and high load conditions respectively. First, through the temperature data under the first load condition (for example, a lower load), it is possible to preliminarily judge whether the basic heat dissipation performance of the cold plate meets the standard. If the heat dissipation performance of the cold plate is good under this light load condition, then further testing is carried out under higher load conditions (i.e., the second load condition) to verify its heat dissipation capacity and stability under heavy load conditions. By performing a more rigorous second heat dissipation performance judgment through the third temperature data and the fourth temperature data obtained under high load conditions, the heat dissipation efficiency of the cold plate under extreme working conditions can be checked.

[0149] In summary, the embodiments provided in this application realize dynamic evaluation of the heat dissipation performance of the cold plate under different load conditions, ensuring that the cold plate can provide optimal heat dissipation effect in various working environments, thereby achieving the technical effect of improving the accuracy of the heat dissipation performance evaluation of the cold plate.

[0150] As an optional solution, the method further includes:

[0151] When the cold plate does not meet the first heat dissipation performance condition or does not meet the second heat dissipation performance condition, determining that the cold plate does not meet the expected heat dissipation performance condition;

[0152] Perform correction operations on the cold plate;

[0153] When the cold plate meets the first heat dissipation performance condition and the second heat dissipation performance condition, it is determined that the cold plate meets the expected heat dissipation performance condition.

[0154] Optionally, in this embodiment, the correction operation is used to indicate the problem location, cause analysis, and design or manufacturing correction process performed when the cold plate fails to meet the expected heat dissipation performance conditions, aiming to improve the heat dissipation performance of the cold plate and ensure that it meets the design requirements.

[0155] Optionally, in this embodiment, after completing the heat dissipation performance judgment under low load and high load conditions, if the result under any condition indicates that the cold plate does not meet the corresponding heat dissipation performance standard, it can be determined that the cold plate as a whole does not meet the expected heat dissipation performance conditions.

[0156] For example, suppose the cold plate fails the temperature difference judgment under the low-load test (i.e., it does not meet the first heat dissipation performance condition), even if all indicators are qualified in the high-load test; or if the comprehensive parameters of the cold plate exceed the threshold in the high-load test (i.e., it does not meet the second heat dissipation performance condition), even if it performs well in the low-load test; the cold plate will be judged as not meeting the expected heat dissipation performance conditions as a whole. Regardless of whether a single problem occurs under low-load or high-load conditions, it will trigger a judgment of overall performance failure.

[0157] Once it is determined that the cold plate does not meet the expected thermal performance conditions, the next step is to conduct an in-depth review of the cold plate to analyze the specific reasons for failure, such as design flaws, material issues, improper assembly, etc., and then modify or adjust it based on the review results to improve its thermal performance.

[0158] Based on the review results, it may be necessary to optimize the cold plate structure, adjust the distribution of the thermal media, or improve the assembly process to ensure that the cold plate's heat dissipation performance meets standards under low load conditions. It may also be necessary to adjust the liquid cooling system's parameter settings, optimize the cold plate material selection, or redesign the heat dissipation path to ensure that the cold plate's heat dissipation performance meets expectations when the server is under high computing load.

[0159] The embodiments provided herein provide a closed-loop mechanism for diagnosing the heat dissipation performance of liquid-cooled server cold plates. During the diagnostic process, if a cold plate fails to meet the heat dissipation performance standards under low or high load conditions, not only will it be determined to be non-compliant with the expected heat dissipation performance conditions, but it will also trigger a review and modification process for the cold plate. This mechanism ensures that manufacturers can promptly identify and resolve issues, preventing substandard cold plates from entering the market, thereby ensuring stable server operation and safe user experience.

[0160] As an alternative solution, the aforementioned cold plate heat dissipation performance evaluation method is applied to the heat dissipation performance diagnosis scenario of liquid-cooled server cold plates. Under both high and low load conditions in a Linux system, CPU and GPU sensor information is automatically read via IPMI. Using a sensor network and algorithms, automatic parameter matching is used to monitor the server's heat dissipation status in real time. This establishes a multi-dimensional quantitative evaluation system, rather than simply monitoring a single temperature threshold, to determine whether the design and assembly of the server's CPU and GPU liquid cooling plates meet heat dissipation performance requirements.

[0161] Optionally, a flow chart for implementing the diagnosis of the heat dissipation performance of the cold plate of a liquid-cooled server is as follows: Figure 3 Shown, including:

[0162] S301, during the normal operation test after startup, read the CPU temperature and GPU temperature sensor information (through IPMI commands), calculate each temperature to determine whether the heat dissipation performance requirements are met;

[0163] S302: After the machine is powered on, a continuous stress test is performed on the machine. The CPU temperature and GPU temperature sensor information are read (through IPMI commands). The respective temperatures are calculated to determine whether the heat dissipation performance requirements are met.

[0164] Specifically, a flow chart for implementing the diagnosis of the heat dissipation performance of the cold plate of a liquid-cooled server is as follows: Figure 4Shown, including:

[0165] S401, start test;

[0166] S402, collecting CPU and GPU temperature information in a stress-free state;

[0167] S403, determining whether the data collected in S402 exceeds a temperature threshold;

[0168] S404, if the data collected in S402 does not exceed the temperature threshold, calculate whether the temperature difference between each CPU and GPU meets the heat dissipation performance requirement;

[0169] S405, collecting CPU and GPU temperature information under stress if the temperature difference between the CPU and GPU in S404 meets the heat dissipation performance requirement;

[0170] S406, determining whether the data collected in S405 exceeds a temperature threshold;

[0171] S407, if the data collected in S406 does not exceed the temperature threshold, calculate whether the temperature difference between each CPU and GPU meets the heat dissipation performance requirement;

[0172] S408 , when the temperature difference between each CPU and GPU meets the heat dissipation performance requirement in S407 , it is determined that the test has passed.

[0173] It should be noted that other situations in the above judgment process are determined to have failed the test.

[0174] For further example, the temperature difference limit under low load and high load conditions ( , ), standard deviation limit ( , ), multi-index fusion limit ( , ) defined in the configuration file.

[0175] Perform low-load cold plate cooling performance diagnosis to evaluate the cold plate cooling performance of the server when the entire machine is running at low load. During normal operation, read the CPU temperature and GPU temperature through the ipmi command every 1 minute and output them to the log file. The data is collected 60 times in 1 hour. If the real-time temperature value exceeds the threshold during the data collection process, The test fails.

[0176] If the real-time collected temperature does not exceed the threshold , then further temperature difference data analysis is performed. If a large local temperature difference is measured at different positions on the surface of the heating element or the radiator (for example, the temperature at one point is significantly higher than that at other points), it indicates that the heat distribution is uneven, which may lead to local overheating or low heat dissipation efficiency. Therefore, the design evaluates whether the heat dissipation performance requirements are met by the temperature difference of the measurement points of each component.

[0177] During the above collection process, the temperature difference calculation is performed on the collected data every 5 minutes, the temperature of each CPU is read from the log file, and the average value of the collected temperature of each CPU is calculated. 、 、 ..., calculate the sum of the average temperatures of each CPU , sort the average values ​​and find the maximum value and minimum value , calculate the difference between the maximum and minimum values , using the difference and sum value Divide to calculate percentage , get the material code of the current CPU cold plate from the whole server production order BOM, and match the current material code corresponding to the configuration file , Should be less than Otherwise, the heat dissipation performance requirement is not met and the test fails. The specific calculation formula is as follows:

[0178]

[0179]

[0180]

[0181] Similarly, read the temperature of each GPU from the log file and calculate the average temperature of each GPU 、 、 ..., calculate the sum of the average temperature of each GPU , sort the average values ​​and find the maximum value and minimum value , calculate the difference between the maximum and minimum values , using the difference and sum value Divide to calculate percentage , get the material code of the current GPU cold plate from the whole server production order BOM, and match the current material code to the corresponding one from the configuration file , Should be less than Otherwise, the heat dissipation performance requirements are not met and the test fails. The specific calculation formula is as follows:

[0182]

[0183]

[0184]

[0185] If the temperature differences between the CPUs and GPUs mentioned above do not exceed the limits, further standard deviation data analysis is performed. The temperature standard deviation (σ) is a statistical indicator that measures the degree of dispersion of the temperature distribution. It can reflect the uniformity and thermal management efficiency of the cooling system. If there are design flaws in the cooling system (such as uneven heat conduction or poor airflow distribution), the temperature in some areas will be significantly higher than that in other areas, resulting in an increase in the standard deviation of the temperature distribution. A larger standard deviation indicates a more dispersed data distribution, while a smaller standard deviation indicates a more concentrated data distribution. Therefore, the standard deviation is used to evaluate whether the liquid cold plate meets the cooling performance requirements.

[0186] During the above collection process, the standard deviation of the collected data is calculated every 5 minutes to evaluate the standard deviation of each CPU temperature, and the temperature of each CPU is read from the log file to calculate the standard deviation of all collected CPU temperatures. average value and variance , get the material code of the current CPU cold plate from the BOM, and read the corresponding material code from the configuration file , Should be less than Otherwise, the heat dissipation performance requirement is not met and the test fails. The specific calculation formula is as follows:

[0187]

[0188] Similarly, to evaluate the standard deviation of GPU temperature, we calculated the standard deviation of the collected data every 5 minutes to evaluate the standard deviation of each GPU temperature, read the temperature of each GPU from the log file, and calculated the temperature of all collected GPUs. average value and variance , get the material code of the current GPU cold plate from the BOM, and read the corresponding material code from the configuration file , Should be less than Otherwise, the heat dissipation performance requirement is not met and the test fails. The specific calculation formula is as follows:

[0189]

[0190] Perform linear weighted fusion on the temperature difference and standard deviation calculated every 5 minutes, and match the corresponding value of the current material code read in the configuration file. , use comprehensive indicators to evaluate whether the heat dissipation performance requirements are met. The specific calculation formula is as follows:

[0191]

[0192] Perform high-load cold plate cooling performance diagnosis to evaluate the server's performance under high load and high heat generation. Stress testing is performed on the CPU and GPU. This increases the server's load, and the temperatures of each CPU and GPU increase accordingly. The ipmi command is used to read the CPU and GPU temperatures every 30 seconds and output them to a file. The stress test and data collection process lasts for 10 hours, or 1200 times. If any temperature value in the collected data exceeds the threshold, the test fails. If it does not exceed the threshold, the next step in the cooling performance analysis is performed.

[0193] If the real-time collected temperatures do not exceed the threshold, further temperature difference data analysis is performed to evaluate whether the temperature difference at each component measurement point under high load meets the heat dissipation performance requirements.

[0194] During the above collection process, the temperature difference calculation is performed on the collected data every 5 minutes, the temperature of each CPU is read from the log file, and the average value of the collected temperature of each CPU is calculated. 、 、 ..., calculate the sum of the average temperatures of each CPU , sort the average values ​​and find the maximum value and minimum value , calculate the difference between the maximum and minimum values , using the difference and sum value Divide to calculate percentage , get the material code of the current CPU cold plate from the whole server production order BOM, and match the current material code corresponding to the configuration file , Should be less than Otherwise, the heat dissipation performance requirement is not met and the test fails. The specific calculation formula is as follows:

[0195]

[0196]

[0197]

[0198] Similarly, read the temperature of each GPU from the log file and calculate the average temperature of each GPU 、 、 ..., calculate the sum of the average temperature of each GPU , sort the average values ​​and find the maximum value and minimum value , calculate the difference between the maximum and minimum values , using the difference and sum value Divide to calculate percentage , get the material code of the current GPU cold plate from the whole server production order BOM, and match the current material code to the corresponding one from the configuration file , Should be less than Otherwise, the heat dissipation performance requirements are not met and the test fails. The specific calculation formula is as follows:

[0199]

[0200]

[0201]

[0202] If the temperature differences between the CPUs and GPUs mentioned above do not exceed the limits, further standard deviation data analysis is performed. A larger standard deviation indicates a more dispersed data distribution, while a smaller standard deviation indicates a more concentrated data distribution. The standard deviation under high load is used to evaluate whether the liquid cooling plate meets the heat dissipation performance requirements.

[0203] During the above collection process, the standard deviation of the collected data is calculated every 5 minutes to evaluate the standard deviation of each CPU temperature, and the temperature of each CPU is read from the log file to calculate the standard deviation of all collected CPU temperatures. average value and variance , get the material code of the current CPU cold plate from the BOM, and read the corresponding material code from the configuration file , Should be less than Otherwise, the heat dissipation performance requirement is not met and the test fails. The specific calculation formula is as follows:

[0204]

[0205] Similarly, to evaluate the GPU temperature standard deviation, we calculated the standard deviation of the collected data every 5 minutes to evaluate the standard deviation of each GPU temperature, read the temperature of each GPU from the log file, and calculated the temperature of all collected GPUs. average value and variance , get the material code of the current GPU cold plate from the BOM, and read the corresponding material code from the configuration file , Should be less than Otherwise, the heat dissipation performance requirement is not met and the test fails. The specific calculation formula is as follows:

[0206]

[0207] Perform a linear weighted fusion on the temperature difference and standard deviation calculated every 5 minutes, and use comprehensive indicators to evaluate whether the heat dissipation performance requirements are met. The specific calculation formula is as follows:

[0208]

[0209] It should be noted that this embodiment uses the temperature difference Δ algorithm, standard deviation σ algorithm, linear fusion value algorithm, and sensor data (CPU / GPU temperature, etc.) under high / low load scenarios of the server under Linux system. Algorithmic analysis evaluates heat dissipation performance and establishes a multi-dimensional quantitative evaluation system, rather than just monitoring a single temperature threshold, to determine whether the server CPU and GPU liquid cooling plates meet heat dissipation performance requirements.

[0210] It can be understood that this embodiment designs a cold plate liquid-cooled AI server heat dissipation performance diagnosis method and system, which is used in the server production test and inspection process, effectively intercepting liquid cold plate body defects and poor assembly, ensuring product quality, and improving test efficiency, test coverage and test automation level.

[0211] According to the cold plate material code and performance requirements, the temperature difference limit under low load and high load conditions ( , ), standard deviation limit ( , ), multi-index fusion limit ( , ) is defined in the configuration file. An example table of the relationship between material codes and limit parameters for a configuration file is shown in Table 1:

[0212] Table 1 Example of the relationship between material codes and limit parameters in the configuration file

[0213]

[0214] Through the embodiments provided in this application, multi-parameter monitoring and data analysis are used to achieve real-time monitoring, fault diagnosis, and performance optimization of the entire server cold plate liquid cooling system, significantly improving the stability and reliability of the liquid-cooled server and having important implications for heat dissipation management in high-performance computing environments. Cold plate temperature uniformity testing and analysis are performed to evaluate the uniformity of the temperature distribution on the cold plate surface to ensure that the cold plate can effectively cool the entire cooled object. This private system can fully and automatically perform diagnostic tests under the Linux system to determine whether the server cold plate meets heat dissipation performance requirements, establishing a multi-dimensional quantitative evaluation system rather than simply monitoring a single temperature threshold. This diagnostic approach improves testing efficiency and test automation levels in server mass production, ensuring the quality of server shipments and the reliable operation of the server in client high-performance computing tasks.

[0215] Through the description of the above embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus the necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of the various embodiments of the present application.

[0216] This embodiment also provides a device for evaluating the heat dissipation performance of a cold plate. This device is used to implement the above-mentioned embodiments and preferred implementations, and details already described will not be repeated. As used below, the term "module" may refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0217] Figure 5 FIG. 1 is a structural block diagram of a device for evaluating the heat dissipation performance of a cold plate according to an embodiment of the present application. Figure 5 As shown, the device includes:

[0218] An acquiring unit 502 is configured to acquire a plurality of first temperature values ​​of the central processing unit and a plurality of second temperature values ​​of the graphics processing unit when the cold plate is in operation, wherein the cold plate is configured to dissipate heat for the central processing unit and the graphics processing unit;

[0219] A first determining unit 504 is configured to determine a first temperature difference between the CPU and the GPU based on the plurality of first temperature values ​​and the plurality of second temperature values ​​when each of the plurality of first temperature values ​​and the plurality of second temperature values ​​is less than a first temperature threshold;

[0220] The second judgment unit 506 is configured to, when the result of the first temperature difference judgment indicates a pass, perform a first temperature standard deviation judgment on the CPU and the GPU based on the plurality of first temperature values ​​and the plurality of second temperature values;

[0221] The third judgment unit 508 is configured to perform a first weighted judgment on the CPU and the GPU based on the result of the first temperature difference judgment and the result of the first temperature standard deviation judgment when the result of the first temperature standard deviation judgment indicates a pass;

[0222] The determining unit 510 is configured to determine that the cold plate meets the first heat dissipation performance condition if the result of the first weighted judgment indicates a pass.

[0223] As an optional solution, the first judging unit 504 includes:

[0224] a first determining module configured to determine a first temperature value of each of the central processing units (CPUs) from the plurality of first temperature values, and to determine a second temperature value of each of the graphics processing units (GPUs) from the plurality of second temperature values, wherein the first temperature value of each of the CPUs is used to determine an average temperature value of each of the CPUs, and the second temperature value of each of the graphics processing units is used to determine an average temperature value of each of the graphics processing units;

[0225] a second determining module, configured to determine a first temperature difference parameter corresponding to the central processing unit according to a first average temperature value having the highest average temperature value and a second average temperature value having the lowest average temperature value among the average temperature values ​​of the central processing units;

[0226] a third determining module, configured to determine a second temperature difference parameter corresponding to the graphics processor according to a third average temperature value having the highest average temperature value and a fourth average temperature value having the lowest average temperature value among the average temperature values ​​of the graphics processors;

[0227] The fourth determining module is configured to determine that a result of the first temperature difference judgment indicates a pass when the first temperature difference parameter is less than the temperature difference threshold and the second temperature difference parameter is less than the temperature difference threshold.

[0228] As an optional solution, the second determining module includes:

[0229] A first calculation submodule is configured to accumulate the average temperature values ​​of the respective CPUs to obtain a first accumulated temperature value;

[0230] a second calculation submodule, configured to obtain a first temperature difference parameter by dividing a first difference between the first average temperature value and the second average temperature value by the first accumulated temperature value;

[0231] The third determination module includes:

[0232] A third calculation submodule is configured to accumulate the average temperature values ​​of each graphics processor to obtain a second accumulated temperature value;

[0233] The fourth calculation submodule is configured to divide the second difference between the third average temperature value and the fourth average temperature value by the second accumulated temperature value to obtain a second temperature difference parameter.

[0234] As an optional solution, the second judging unit 506 includes:

[0235] a first calculation module configured to, upon obtaining an average temperature value of each CPU in the CPU, calculate a standard deviation of the average temperature value of each CPU to obtain a first temperature standard deviation parameter;

[0236] a second calculation module configured to, upon obtaining an average temperature value of each graphics processor in the graphics processor, calculate a standard deviation of the average temperature value of each graphics processor to obtain a second temperature standard deviation parameter;

[0237] The fifth determining module is configured to determine that a result of the first temperature standard deviation judgment indicates a pass when the first temperature standard deviation parameter is less than the temperature standard deviation threshold and the second temperature standard deviation parameter is less than the temperature standard deviation threshold.

[0238] As an optional solution, the third judgment unit 508 includes:

[0239] a first acquisition module, configured to acquire a first temperature difference parameter of the central processing unit and a second temperature difference parameter of the graphics processing unit based on a result of the first temperature difference determination, and to acquire a first temperature standard deviation parameter of the central processing unit and a second temperature standard deviation parameter of the graphics processing unit based on a result of the first temperature standard deviation determination;

[0240] a first weighting module, configured to perform a first weighted fusion operation on the first temperature difference parameter and the first temperature standard deviation parameter to obtain a first comprehensive parameter of the central processing unit;

[0241] a second weighting module, configured to perform a second weighted fusion operation on the second temperature difference parameter and the second temperature standard deviation parameter to obtain a second comprehensive parameter of the graphics processor;

[0242] The sixth determination module is configured to determine that a result of the first weighted judgment indicates a pass when the first comprehensive parameter is less than the comprehensive parameter threshold and the second comprehensive parameter is less than the comprehensive parameter threshold.

[0243] As an optional solution, the first weighting module includes:

[0244] A first acquisition submodule, configured to acquire a first weighting parameter corresponding to the central processing unit and a second weighting parameter corresponding to the graphics processing unit, and to acquire a weighted parameter and value of the first weighting parameter and the second weighting parameter;

[0245] a second acquisition submodule, configured to acquire a first cumulative parameter obtained by multiplying the first weighted parameter by the first temperature difference parameter, and to acquire a second cumulative parameter obtained by multiplying the second weighted parameter by the first temperature standard deviation parameter;

[0246] A fifth calculation submodule, configured to determine a result of dividing the accumulated value of the first accumulated parameter and the second accumulated parameter by the sum of the weighted parameters as a first comprehensive parameter;

[0247] The second weighting module includes:

[0248] a third acquisition submodule, configured to acquire a third cumulative parameter obtained by multiplying the first weighted parameter by the second temperature difference parameter, and to acquire a fourth cumulative parameter obtained by multiplying the second weighted parameter by the second temperature standard deviation parameter;

[0249] The sixth calculation submodule is configured to determine a result of dividing the accumulated value of the third accumulated parameter and the fourth accumulated parameter by the weighted parameter sum as the second comprehensive parameter.

[0250] As an optional solution, the device further includes:

[0251] a second acquisition module, configured to acquire a target material type of the cold plate before determining a first temperature difference between the central processing unit and the graphics processing unit based on the plurality of first temperature values ​​and the plurality of second temperature values;

[0252] The seventh determination module is used to determine the temperature difference threshold corresponding to the target material type, the temperature standard deviation threshold corresponding to the target material type, and the comprehensive parameter threshold corresponding to the target material type from the configuration file before performing the first temperature difference judgment on the central processing unit and the graphics processing unit based on multiple first temperature values ​​and multiple second temperature values, wherein the temperature difference threshold is used for the first temperature difference judgment, the temperature standard deviation threshold is used for the first temperature standard deviation judgment, and the comprehensive parameter threshold is used for the first weighted judgment, and the configuration file stores the correspondence between multiple groups of material types and thresholds.

[0253] As an optional solution, the device further includes:

[0254] an eighth determination module, configured to determine that the cold plate does not meet the first heat dissipation performance condition when there is a temperature value greater than or equal to a preset threshold value among the multiple first temperature values ​​and the multiple second temperature values, or when the result of the first temperature difference judgment indicates failure, or when the result of the first temperature standard deviation judgment indicates failure, or when the result of the first weighted judgment indicates failure.

[0255] As an optional solution, obtain the unit, including:

[0256] a third acquisition module, configured to acquire, at a first sampling frequency, a plurality of first temperature values ​​of the central processing unit and a plurality of second temperature values ​​of the graphics processing unit when the cold plate operates under a first load condition;

[0257] The device also includes:

[0258] a fourth acquisition module, configured to, after determining that the cold plate meets the first heat dissipation performance condition, obtain, according to a second sampling frequency, a plurality of third temperature values ​​of the central processing unit and a plurality of fourth temperature values ​​of the graphics processing unit when the cold plate operates under a second load condition, wherein the second sampling frequency is higher than the first sampling frequency, and the plurality of third temperature values ​​and the plurality of fourth temperature values ​​are used to determine whether the cold plate meets the second heat dissipation performance condition.

[0259] As an optional solution, the device further includes:

[0260] a first determination module configured to, after acquiring, at a second sampling frequency, a plurality of third temperature values ​​of the central processing unit and a plurality of fourth temperature values ​​of the graphics processing unit when the cold plate operates under a second load condition, perform a second temperature difference determination on the central processing unit and the graphics processing unit based on the plurality of third temperature values ​​and the plurality of fourth temperature values ​​if each of the plurality of third temperature values ​​and the plurality of fourth temperature values ​​is less than a second temperature threshold;

[0261] a second determination module configured to, after acquiring, at a second sampling frequency, a plurality of third temperature values ​​of the central processing unit and a plurality of fourth temperature values ​​of the graphics processing unit when the cold plate operates under a second load condition, and, if a result of the second temperature difference determination indicates a pass, perform a second temperature standard deviation determination on the central processing unit and the graphics processing unit based on the plurality of third temperature values ​​and the plurality of fourth temperature values;

[0262] a third judgment module, configured to, after acquiring, at a second sampling frequency, a plurality of third temperature values ​​of the central processing unit and a plurality of fourth temperature values ​​of the graphics processing unit when the cold plate operates under a second load condition, and if a result of a second temperature standard deviation judgment indicates a pass, perform a second weighted judgment on the central processing unit and the graphics processing unit based on a result of the second temperature difference judgment and a result of the second temperature standard deviation judgment;

[0263] The ninth determining module is configured to determine that the cold plate meets the second heat dissipation performance condition if the result of the second weighted judgment indicates a pass.

[0264] As an optional solution, the device further includes:

[0265] a tenth determining module, configured to determine that the cold plate does not meet the expected heat dissipation performance condition when the cold plate does not meet the first heat dissipation performance condition or does not meet the second heat dissipation performance condition;

[0266] Correction module, used for performing correction operations on the cold plate;

[0267] The eleventh determining module is configured to determine that the cold plate meets the expected heat dissipation performance condition when the cold plate meets the first heat dissipation performance condition and the second heat dissipation performance condition.

[0268] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail here.

[0269] Through the description of the above embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus the necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the existing technology, can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0270] It should be noted that the above modules can be implemented through software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.

[0271] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored. The computer program is configured to execute the steps of any of the above method embodiments when run.

[0272] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk, or an optical disk.

[0273] An embodiment of the present application further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0274] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0275] An embodiment of the present application further provides a computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores the computer program product, and when the computer program is executed by a processor, the steps of the method in each embodiment of the present application are implemented.

[0276] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail here.

[0277] Obviously, those skilled in the art should understand that the modules or steps of the present application described above can be implemented using a general-purpose computing device, they can be concentrated on a single computing device, or distributed across a network composed of multiple computing devices, they can be implemented using program code executable by the computing device, and thus, they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be performed in a different order than herein, or they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0278] The above are only preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, various modifications and variations of the present application are possible. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for evaluating the heat dissipation performance of a cold plate, characterized in that: include: Acquiring a plurality of first temperature values ​​of a central processing unit and a plurality of second temperature values ​​of a graphics processing unit when a cold plate is in operation, wherein the cold plate is used to dissipate heat for the central processing unit and the graphics processing unit; In a case where each temperature value among the plurality of first temperature values ​​and the plurality of second temperature values ​​is less than a first temperature threshold, a first temperature difference judgment is performed on the central processing unit and the graphics processing unit based on the plurality of first temperature values ​​and the plurality of second temperature values, including: determining a first temperature value of each central processing unit in the central processing unit from the plurality of first temperature values, and determining a second temperature value of each graphics processing unit in the graphics processing unit from the plurality of second temperature values, wherein the first temperature value of each central processing unit is used to determine an average temperature value of each central processing unit, and the second temperature value of each graphics processing unit is used to determine an average temperature value of each graphics processing unit; and determining a first temperature difference parameter corresponding to the central processing unit based on a first average temperature value having the highest average temperature value and a second average temperature value having the lowest average temperature value among the average temperature values ​​of the central processing units; determining a second temperature difference parameter corresponding to the graphics processing unit based on a third average temperature value having the highest average temperature value and a fourth average temperature value having the lowest average temperature value among the average temperature values ​​of the graphics processing units; determining that a result of the first temperature difference determination indicates a pass if the first temperature difference parameter is less than a temperature difference threshold and the second temperature difference parameter is less than the temperature difference threshold, and the first temperature threshold is a safety temperature upper limit set for the central processing unit and the graphics processing unit; If the result of the first temperature difference determination indicates a pass, performing a first temperature standard deviation determination on the central processing unit and the graphics processing unit according to the plurality of first temperature values ​​and the plurality of second temperature values; If the result of the first temperature standard deviation judgment indicates a pass, a first weighted judgment is performed on the central processing unit and the graphics processing unit based on the result of the first temperature difference judgment and the result of the first temperature standard deviation judgment, including: obtaining the first temperature difference parameter of the central processing unit and the second temperature difference parameter of the graphics processing unit based on the result of the first temperature difference judgment, and obtaining the first temperature standard deviation parameter of the central processing unit and the second temperature standard deviation parameter of the graphics processing unit based on the result of the first temperature standard deviation judgment; performing a first weighted fusion operation on the first temperature difference parameter and the first temperature standard deviation parameter to obtain a first comprehensive parameter of the central processing unit; performing a second weighted fusion operation on the second temperature difference parameter and the second temperature standard deviation parameter to obtain a second comprehensive parameter of the graphics processing unit; and determining that the result of the first weighted judgment indicates a pass if the first comprehensive parameter is less than a comprehensive parameter threshold and the second comprehensive parameter is less than the comprehensive parameter threshold; When the result of the first weighted judgment indicates a pass, it is determined that the cold plate meets the first heat dissipation performance condition.

2. The method according to claim 1, characterized in that Determining a first temperature difference parameter corresponding to the central processor according to a first average temperature value having the highest average temperature value and a second average temperature value having the lowest average temperature value among the average temperature values ​​of the central processors includes: Accumulating average temperature values ​​of the respective central processing units to obtain a first accumulated temperature value; Divide the first difference between the first average temperature value and the second average temperature value by the first accumulated temperature value to obtain the first temperature difference parameter; Determining the second temperature difference parameter corresponding to the graphics processor according to a third average temperature value having the highest average temperature value and a fourth average temperature value having the lowest average temperature value among the average temperature values ​​of the graphics processors includes: Accumulating average temperature values ​​of the graphics processors to obtain a second accumulated temperature value; A second difference between the third average temperature value and the fourth average temperature value is divided by the second accumulated temperature value to obtain the second temperature difference parameter.

3. The method according to claim 1, characterized in that The determining of a first temperature standard deviation of the central processing unit and the graphics processing unit according to the plurality of first temperature values ​​and the plurality of second temperature values ​​includes: When the average temperature values ​​of the respective CPUs in the CPUs are obtained, performing a standard deviation operation on the average temperature values ​​of the respective CPUs to obtain the first temperature standard deviation parameter; When an average temperature value of each graphics processor in the graphics processor is obtained, performing a standard deviation operation on the average temperature value of each graphics processor to obtain the second temperature standard deviation parameter; In a case where the first temperature standard deviation parameter is smaller than a temperature standard deviation threshold, and the second temperature standard deviation parameter is smaller than the temperature standard deviation threshold, it is determined that the result of the first temperature standard deviation judgment indicates a pass.

4. The method according to claim 1, wherein The performing a first weighted fusion operation on the first temperature difference parameter and the first temperature standard deviation parameter to obtain a first comprehensive parameter of the central processing unit includes: Obtaining a first weighting parameter corresponding to the central processing unit and a second weighting parameter corresponding to the graphics processing unit, and obtaining a weighting parameter and value of the first weighting parameter and the second weighting parameter; Obtaining a first cumulative parameter obtained by multiplying the first weighted parameter by the first temperature difference parameter, and obtaining a second cumulative parameter obtained by multiplying the second weighted parameter by the first temperature standard deviation parameter; Determine the result of dividing the accumulated value of the first accumulated parameter and the second accumulated parameter by the sum of the weighted parameters as the first comprehensive parameter; The performing a second weighted fusion operation on the second temperature difference parameter and the second temperature standard deviation parameter to obtain a second comprehensive parameter of the graphics processor includes: Obtaining a third cumulative parameter obtained by multiplying the first weighted parameter by the second temperature difference parameter, and obtaining a fourth cumulative parameter obtained by multiplying the second weighted parameter by the second temperature standard deviation parameter; The second comprehensive parameter is determined by dividing the accumulated value of the third accumulated parameter and the fourth accumulated parameter by the weighted parameter sum.

5. The method according to claim 1, characterized in that Before determining a first temperature difference between the central processing unit and the graphics processing unit based on the multiple first temperature values ​​and the multiple second temperature values, the method further includes: Obtaining a target material type of the cold plate; A temperature difference threshold corresponding to the target material type, a temperature standard deviation threshold corresponding to the target material type, and a comprehensive parameter threshold corresponding to the target material type are determined from a configuration file, wherein the temperature difference threshold is used for the first temperature difference judgment, the temperature standard deviation threshold is used for the first temperature standard deviation judgment, and the comprehensive parameter threshold is used for the first weighted judgment. The configuration file stores a plurality of sets of correspondences between material types and thresholds.

6. The method according to claim 1, characterized in that The method further comprises: When there is a temperature value greater than or equal to the first temperature threshold value among the multiple first temperature values ​​and the multiple second temperature values, or when the result of the first temperature difference judgment indicates failure, or when the result of the first temperature standard deviation judgment indicates failure, or when the result of the first weighted judgment indicates failure, it is determined that the cold plate does not meet the first heat dissipation performance condition.

7. The method according to claim 1, characterized in that The step of obtaining a plurality of first temperature values ​​of the central processing unit and a plurality of second temperature values ​​of the graphics processing unit when the cold plate is running includes: acquiring, at a first sampling frequency, the plurality of first temperature values ​​of the central processing unit and the plurality of second temperature values ​​of the graphics processing unit when the cold plate operates under a first load condition; After determining that the cold plate meets the first heat dissipation performance condition, the method further includes: Acquire, according to a second sampling frequency, a plurality of third temperature values ​​of the central processing unit and a plurality of fourth temperature values ​​of the graphics processing unit when the cold plate operates under a second load condition, wherein the second sampling frequency is higher than the first sampling frequency, and the plurality of third temperature values ​​and the plurality of fourth temperature values ​​are used to determine whether the cold plate meets a second heat dissipation performance condition.

8. The method according to claim 7, characterized in that After acquiring, according to the second sampling frequency, a plurality of third temperature values ​​of the central processing unit and a plurality of fourth temperature values ​​of the graphics processing unit when the cold plate operates under the second load condition, the method further includes: When each of the plurality of third temperature values ​​and the plurality of fourth temperature values ​​is less than a second temperature threshold, performing a second temperature difference determination on the central processing unit and the graphics processing unit according to the plurality of third temperature values ​​and the plurality of fourth temperature values; If the result of the second temperature difference determination indicates a pass, performing a second temperature standard deviation determination on the central processing unit and the graphics processing unit according to the plurality of third temperature values ​​and the plurality of fourth temperature values; When the result of the second temperature standard deviation judgment indicates a pass, performing a second weighted judgment on the central processing unit and the graphics processing unit according to the result of the second temperature difference judgment and the result of the second temperature standard deviation judgment; If the result of the second weighted judgment indicates a pass, it is determined that the cold plate meets the second heat dissipation performance condition.

9. The method according to claim 8, characterized in that The method further comprises: When the cold plate does not meet the first heat dissipation performance condition or does not meet the second heat dissipation performance condition, determining that the cold plate does not meet the expected heat dissipation performance condition; performing a correction operation on the cold plate; When the cold plate meets the first heat dissipation performance condition and the second heat dissipation performance condition, it is determined that the cold plate meets the expected heat dissipation performance condition.

10. A device for evaluating the heat dissipation performance of a cold plate, characterized in that: include: an acquisition unit, configured to acquire a plurality of first temperature values ​​of the central processing unit and a plurality of second temperature values ​​of the graphics processing unit when a cold plate is in operation, wherein the cold plate is configured to dissipate heat for the central processing unit and the graphics processing unit; a first judgment unit, configured to, when each of the plurality of first temperature values ​​and the plurality of second temperature values ​​is less than a first temperature threshold, perform a first temperature difference judgment on the central processing unit and the graphics processing unit based on the plurality of first temperature values ​​and the plurality of second temperature values, including: determining a first temperature value of each central processing unit in the central processing unit from the plurality of first temperature values, and determining a second temperature value of each graphics processing unit in the graphics processing unit from the plurality of second temperature values, the first temperature value of each central processing unit being used to determine an average temperature value of each central processing unit, and the second temperature value of each graphics processing unit being used to determine an average temperature value of each graphics processing unit value; determining a first temperature difference parameter corresponding to the central processing unit according to a first average temperature value with the highest average temperature value and a second average temperature value with the lowest average temperature value among the average temperature values ​​of the central processing units; determining a second temperature difference parameter corresponding to the graphics processing unit according to a third average temperature value with the highest average temperature value and a fourth average temperature value with the lowest average temperature value among the average temperature values ​​of the graphics processing units; if the first temperature difference parameter is less than a temperature difference threshold and the second temperature difference parameter is less than the temperature difference threshold, determining that the result of the first temperature difference judgment indicates a pass, and the first temperature threshold is a safety temperature upper limit set for the central processing unit and the graphics processing unit; a second judgment unit, configured to, when a result of the first temperature difference judgment indicates a pass, perform a first temperature standard deviation judgment on the central processing unit and the graphics processing unit according to the plurality of first temperature values ​​and the plurality of second temperature values; a third judgment unit, configured to, if the result of the first temperature standard deviation judgment indicates a pass, perform a first weighted judgment on the central processing unit and the graphics processing unit based on the result of the first temperature difference judgment and the result of the first temperature standard deviation judgment, including: obtaining the first temperature difference parameter of the central processing unit and the second temperature difference parameter of the graphics processing unit based on the result of the first temperature difference judgment, and obtaining the first temperature standard deviation parameter of the central processing unit and the second temperature standard deviation parameter of the graphics processing unit based on the result of the first temperature standard deviation judgment; performing a first weighted fusion operation on the first temperature difference parameter and the first temperature standard deviation parameter to obtain a first comprehensive parameter of the central processing unit; performing a second weighted fusion operation on the second temperature difference parameter and the second temperature standard deviation parameter to obtain a second comprehensive parameter of the graphics processing unit; and determining that the result of the first weighted judgment indicates a pass if the first comprehensive parameter is less than a comprehensive parameter threshold and the second comprehensive parameter is less than the comprehensive parameter threshold; A determining unit is configured to determine that the cold plate meets a first heat dissipation performance condition when a result of the first weighted judgment indicates a pass.

11. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein the computer program implements the steps of the method according to any one of claims 1 to 9 when executed by a processor.

12. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 9 are implemented.

13. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

Images