GPU fan speed correction method, correction device and server

Abstract

The present invention relates to a GPU fan speed correction method, correction device, server, equipment and storage medium. The correction method includes: obtaining a GPU fan theoretical speed value according to a first speed correlation relationship and a single GPU current temperature value; the first speed correlation relationship is a GPU fan speed value and a GPU temperature value correlation relationship; obtaining a GPU fan actual speed value according to a second speed correlation relationship and a GPU actual fan air volume value; the second speed correlation relationship is a GPU fan speed value and a GPU fan air volume value correlation relationship; when the GPU fan theoretical speed value is less than the GPU fan actual speed value, obtaining a difference between the GPU fan actual speed value and the GPU fan theoretical speed value as a speed correction value, and correcting the GPU fan theoretical speed value. Through the above technical solution, the problem that the GPU fan in the current server is easily affected by the system fan and has abnormal speed can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of server fans, and in particular to a GPU fan speed correction method, correction device and server. Background Art

[0002] At present, due to the improvement of performance requirements, PCIe slots are usually set in server design to install GPUs, accelerator cards, network cards and other components (hereinafter collectively referred to as GPU cards). Due to the high power consumption of GPUs when in use, some GPU cards have their own fans.

[0003] The following phenomena may occur during the use of the server:

[0004] When the CPU is operating at full pressure and high power consumption, it generates a lot of heat and the temperature is high. To ensure the CPU heat dissipation, the server's system fan will run at high speed. However, the current GPU may still be in a low-business working state, that is, the GPU fan speed is at a low speed state. The remaining air volume acts on the GPU fan blades under the action of wind pressure, thus forcing the GPU fan speed to increase.

[0005] According to the fan failure standard, when the fan speed deviates from the rated fan speed by ±15%, the fan is considered to have failed. Therefore, when the CPU is under high pressure and the GPU is under low pressure, the GPU fan speed abnormal alarm is very likely to occur. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides a GPU fan speed correction method, correction device and server. The GPU fan speed correction method is used to solve the problem that the GPU fan in the current server is easily affected by the system fan and has abnormal speed.

[0007] To achieve the above object, the present invention provides a GPU fan speed correction method, comprising:

[0008] Obtaining a theoretical speed value of a GPU fan according to a first speed association relationship and a current temperature value of a single GPU; wherein the first speed association relationship is an association relationship between a GPU fan speed value and a GPU temperature value;

[0009] Obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to a single GPU; wherein the second speed association relationship is the association relationship between the GPU fan speed value and the GPU fan air volume value;

[0010] When the theoretical speed value of the GPU fan is less than the actual speed value of the GPU fan, a difference between the actual speed value of the GPU fan and the theoretical speed value of the GPU fan is obtained as a speed correction value, and the theoretical speed value of the GPU fan is corrected.

[0011] Further, obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to a single GPU specifically includes:

[0012] Obtaining a total fan air volume value of all GPUs according to a system total air volume value of the server and a non-air outlet air volume value at a non-GPU air outlet position;

[0013] The actual fan air volume value of the GPU corresponding to a single GPU is obtained according to the total fan air volume value and the GPU quantity value.

[0014] Further, obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to a single GPU also includes:

[0015] Acquire a third rotation speed association relationship, where the third rotation speed association relationship is an association relationship between a system fan rotation speed value and a system fan air volume value of a single system fan in the server;

[0016] The total air volume value of the system is obtained according to the number of system fans and the third rotation speed association relationship.

[0017] Further, obtaining the third rotation speed association relationship specifically includes:

[0018] Obtain the third speed calculation formula corresponding to the third speed association relationship: RPM=(159.2Q) / (S*2r / η); wherein, RPM is the system fan speed value of a single system fan, Q is the system fan air volume value of a single system fan, S is the effective force area of ​​the impeller of a single system fan, r is the average force radius of a single system fan, and η is the efficiency of a single system fan.

[0019] Further, obtaining the total air volume value of the system according to the number of system fans and the third rotation speed association relationship specifically includes:

[0020] Obtain the number of system fans, and obtain the calculation formula for the total system air volume value based on the number of system fans and the third speed calculation formula: Qx=n(RPM*(S*2r / η)) / 159.2; wherein n is the number of system fans, and Qx is the total system air volume value.

[0021] Further, obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to a single GPU specifically includes:

[0022] Obtain the second speed calculation formula corresponding to the second speed association relationship: RPMture=[159.2(Qx-Qp) / m] / (Sgpu *2r gpu / η gpu ), where RPMture is the actual speed of the GPU fan, m is the number of GPU fans, S gpu is the effective force area of ​​the GPU fan impeller, r gpu is the average force radius of the GPU fan, η gpu is the efficiency of the GPU fan, Qp is the non-outlet air volume value;

[0023] The actual speed value of the GPU fan is calculated according to the second speed calculation formula and the actual GPU fan air volume value.

[0024] Further, obtaining the theoretical speed value of the GPU fan according to the first speed association relationship and the current temperature value of a single GPU specifically includes:

[0025] Obtain a first speed calculation formula corresponding to the first speed association relationship: RPMg=TgK+Bg; wherein RPMg is the theoretical speed value of the GPU fan, Tg is the current temperature value of a single GPU, and Bg and K are constants.

[0026] Further, obtaining the difference between the actual speed value of the GPU fan and the theoretical speed value of the GPU fan as the speed correction value specifically includes:

[0027] The speed correction calculation formula is: RPMture-RPMg=[159.2(Qx-Qp) / m] / (S gpu *2r gpu / η gpu )-TgK-Bg;

[0028] The speed correction value is calculated according to the speed correction calculation formula.

[0029] Furthermore, the correction method also includes:

[0030] When the theoretical speed value of the GPU fan is greater than or equal to the actual speed value of the GPU fan, the GPU fan speed information is sent to the BMC without being corrected.

[0031] Further, obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to a single GPU also includes:

[0032] The non-air outlet air volume value is obtained according to the test wind speed and air outlet area at the non-GPU air outlet position.

[0033] Further, obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to a single GPU also includes:

[0034] The non-air outlet air volume value is obtained according to the air volume thermal simulation information at the non-GPU air outlet position.

[0035] Furthermore, the system fan is an 8056 fan, and S, r, and η are fixed parameters.

[0036] Furthermore, before obtaining the theoretical speed value of the GPU fan according to the first speed association relationship and the current temperature value of the single GPU, the correction method further includes:

[0037] When the server is in a powered-on state, the BMC obtains rotation speed control information and controls the rotation speed values ​​of all system fans; wherein the rotation speed control information includes GPU temperature information.

[0038] Furthermore, the BMC obtains speed control information and controls the speed values ​​of all system fans, including:

[0039] The BMC controls the speed values ​​of all system fans according to a preset fan speed regulation mode; wherein the preset fan speed regulation mode includes one of the following: PID speed regulation, linear speed regulation, and segmented speed regulation.

[0040] The present invention also provides a GPU fan speed correction device, comprising:

[0041] A GPU fan theoretical speed value acquisition unit, used to acquire a GPU fan theoretical speed value according to a first speed association relationship and a current temperature value of a single GPU; wherein the first speed association relationship is an association relationship between a GPU fan speed value and a GPU temperature value;

[0042] A GPU fan actual speed value acquisition unit, used to acquire the GPU fan actual speed value according to a second speed association relationship and a GPU fan actual air volume value corresponding to a single GPU; wherein the second speed association relationship is an association relationship between the GPU fan speed value and the GPU fan air volume value;

[0043] The speed correction value acquisition unit is used to obtain the difference between the actual speed value of the GPU fan and the theoretical speed value of the GPU fan as the speed correction value when the theoretical speed value of the GPU fan is less than the actual speed value of the GPU fan, and to correct the theoretical speed value of the GPU fan.

[0044] Furthermore, the GPU fan actual speed value acquisition unit specifically includes:

[0045] a total fan air volume value acquisition unit, configured to acquire the total fan air volume value of all GPUs according to the system total air volume value of the server and the non-air outlet air volume value at the non-GPU air outlet position;

[0046] The GPU fan actual speed value acquisition unit is further used to acquire the GPU actual fan air volume value corresponding to a single GPU according to the total fan air volume value and the GPU quantity value.

[0047] Furthermore, the GPU fan actual speed value acquisition unit also includes:

[0048] A third rotation speed association relationship acquisition unit, configured to acquire a third rotation speed association relationship, wherein the third rotation speed association relationship is an association relationship between a system fan rotation speed value and a system fan air volume value of a single system fan in the server;

[0049] The system total air volume value acquisition unit is used to acquire the system total air volume value according to the system fan quantity value and the third rotation speed association relationship.

[0050] The present invention further provides a computer device, comprising a memory, a processor and a computer program, wherein the computer program is stored in the memory and can be run on the processor, and when the processor executes the computer program, the steps of the aforementioned PCIe device link training management method are implemented.

[0051] The present invention further provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the aforementioned PCIe device link training management method.

[0052] The present invention further provides a server, comprising a GPU and a GPU fan, wherein the rotation speed of the GPU fan is corrected by the aforementioned GPU fan rotation speed correction method.

[0053] The above technical solution of the present invention has the following technical effects compared with the prior art:

[0054] In the present invention, in order to correct the GPU fan speed in the server, the theoretical GPU fan speed value is first obtained according to the first speed correlation relationship and the current temperature value of a single GPU; wherein the first speed correlation relationship is the correlation relationship between the GPU fan speed value and the GPU temperature value;

[0055] Next, the actual speed value of the GPU fan is obtained according to the second speed association relationship and the actual GPU fan air volume value corresponding to the single GPU; wherein the second speed association relationship is the association relationship between the GPU fan speed value and the GPU fan air volume value;

[0056] When it is detected that the theoretical speed value of the GPU fan is less than the actual speed value of the GPU fan, the difference between the actual speed value of the GPU fan and the theoretical speed value of the GPU fan is obtained as a speed correction value, and the theoretical speed value of the GPU fan is corrected;

[0057] Therefore, the theoretical speed value of the GPU fan can be corrected to avoid abnormal speed of the GPU fan due to the influence of the system fan, which leads to speed error alarm information, improve the accuracy of GPU fan feedback parameters, improve GPU reliability, and reduce maintenance costs such as false alarms. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0059] Figure 1 It is a linear relationship diagram between GPU fan speed and temperature in the prior art;

[0060] Figure 2 is a flow chart of a method for correcting the speed of a GPU fan in Embodiment 1 of the present invention;

[0061] Figure 3 It is a schematic diagram of the overall process of the GPU fan speed correction method in the actual embodiment of the present invention;

[0062] Figure 4 is a structural block diagram of a GPU fan speed correction device in Embodiment 2 of the present invention;

[0063] Figure 5 This is a diagram of the internal structure of a computer device in Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0064] In the prior art, according to the GPU fan control strategy, the speed of the GPU fan is linearly related to the core temperature. That is, when the GPU core temperature is lower, the GPU fan speed is lower, and can even be as low as 0 speed; when the GPU temperature is higher, the GPU fan speed is higher. For the specific relationship, please refer to the table below, and the corresponding linear relationship diagram can be found in detail. Figure 1 :

[0065] temperature 0 10 20 30 40 50 60 70 80 90 100 Speed 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

[0066] Taking a 2U general-purpose server as an example, the following problems are likely to occur during the heat dissipation process of the server:

[0067] According to the thermal simulation results, when the CPU temperature is high and the system fan is at full speed, a single 8056 fan can provide 25cfm of air volume, and the four fans of the whole machine can provide 100cfm of air volume; when the chassis is fully equipped with 8 GPU cards, the GPU temperature is in a relatively low state, and the GPU fan speed is 2500 rpm, it can only output 5.2cfm of air volume to the outside. Even if the chassis is fully equipped with 8 GPU cards, it can only output 41.6cfm of air volume to the outside.

[0068] The remaining 58.4cfm of air volume acts on the GPU fan blades under the action of wind pressure (in the extreme case, when the GPU is powered on and the whole machine is operating normally, the GPU fan will automatically rotate under the action of wind pressure), thereby forcing the GPU fan speed to increase.

[0069] According to the fan failure standard, when the fan speed deviates from the rated fan speed by ±15%, the fan is considered to have failed. Therefore, when the CPU is under high pressure and the GPU is under low pressure, the GPU fan speed abnormal alarm is very likely to occur.

[0070] That is, in the prior art, when the CPU is at a high temperature, the fan provides a larger air volume to the CPU; however, the fan speed of the GPU card is controlled by the junction temperature of the GPU. When the system fan wants to provide a larger air volume to the GPU, the speed of the GPU fan is low, which will cause the air volume at the air outlet to be limited.

[0071] It can be seen that the problem of GPU fan speed deviation needs to be solved based on the system fan and GPU fan linkage technology.

[0072] To this end, the present invention provides a GPU fan speed correction method, correction device and server to solve the above problems.

[0073] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0074] Embodiment 1:

[0075] like Figure 2 As shown, an embodiment of the present invention provides a method for correcting the speed of a GPU fan, comprising:

[0076] S1 obtains a theoretical speed value of a GPU fan according to a first speed association relationship and a current temperature value of a single GPU; wherein the first speed association relationship is an association relationship between a GPU fan speed value and a GPU temperature value;

[0077] S2 obtains the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to the single GPU; wherein the second speed association relationship is the association relationship between the GPU fan speed value and the GPU fan air volume value;

[0078] S3: when the theoretical speed value of the GPU fan is less than the actual speed value of the GPU fan, obtain the difference between the actual speed value of the GPU fan and the theoretical speed value of the GPU fan as a speed correction value, and correct the theoretical speed value of the GPU fan.

[0079] In a specific embodiment, in order to correct the fan speed of a GPU in a server, a theoretical fan speed value of the GPU is first obtained according to a first speed correlation relationship and a current temperature value of a single GPU; wherein the first speed correlation relationship is a correlation relationship between a GPU fan speed value and a GPU temperature value;

[0080] Next, the actual speed value of the GPU fan is obtained according to the second speed association relationship and the actual GPU fan air volume value corresponding to the single GPU; wherein the second speed association relationship is the association relationship between the GPU fan speed value and the GPU fan air volume value;

[0081] When it is detected that the theoretical speed value of the GPU fan is less than the actual speed value of the GPU fan, the difference between the actual speed value of the GPU fan and the theoretical speed value of the GPU fan is obtained as a speed correction value, and the theoretical speed value of the GPU fan is corrected;

[0082] Therefore, the theoretical speed value of the GPU fan can be corrected to avoid abnormal speed of the GPU fan due to the influence of the system fan, which leads to speed error alarm information, improve the accuracy of GPU fan feedback parameters, improve GPU reliability, and reduce maintenance costs such as false alarms.

[0083] In an actual implementation, first, according to the GPU heat dissipation design guide, the relationship between the GPU fan speed and the GPU temperature is obtained: RPMg=TgK+Bg; wherein RPMg represents the theoretical speed of the GPU fan, Tg represents the temperature of the GPU, and Bg represents a constant.

[0084] Then, based on the relationship between the fan air volume and speed of the GPU card, the actual speed of the GPU fan, namely RPMture, is calculated.

[0085] When performing GPU fan anomaly detection, the GPU speed can be corrected according to different GPU temperatures and GPU fan control strategies. The correction rules are as follows:

[0086] (1) When RPMg ≥ RPMture, it means that the system fan speed is low and the GPU fan speed is high, and the GPU fan speed information is fed back to the BMC without correction.

[0087] (2) When RPMg<RPMture, it means that the system fan speed is high and the GPU fan speed is low. The GPU fan speed is increased under the wind pressure of the system fan, and the following information is fed back to the BMC: the GPU fan speed value needs to be corrected based on the value of RPMg.

[0088] In practice, fan failure means that when the real-time speed of the fan exceeds the rated speed by ±15%, the fan can be judged to have failed.

[0089] GPU fans refer to the fans that come with high-end GPU graphics cards, which can control their own speed and accelerate the GPU's self-heating to maintain high-performance operation. Common types include ADDA-5010 fans, which have a speed of 5500rpm / min.

[0090] The system fan refers to the fan that comes with the server chassis, which is used to provide the total power source and total air volume source for the heat dissipation of the entire system; common types include 8056 fan, 8056 fan, etc.

[0091] In a preferred embodiment, S2 specifically includes:

[0092] S231 obtains the total fan air volume value of all GPUs according to the total system air volume value of the server and the non-air outlet air volume value at the non-GPU air outlet position;

[0093] S232 obtains the GPU actual fan air volume value corresponding to a single GPU according to the total fan air volume value and the GPU quantity value.

[0094] In an actual embodiment, the total air volume Qx of the system is subtracted from the air volume Qp at the non-GPU air outlet position under the system configuration (which can be obtained through thermal simulation, or by testing the wind speed and air outlet area) to obtain the actual air volume when 8 GPU cards are fully configured. The air volume of each GPU card is Qture = (Qx-Qp) / 8.

[0095] In a preferred embodiment, S2 further includes:

[0096] S21 obtains a third rotation speed association relationship, where the third rotation speed association relationship is an association relationship between a system fan rotation speed value and a system fan air volume value of a single system fan in the server;

[0097] S22 obtains the total air volume value of the system according to the number of system fans and the third rotation speed correlation relationship.

[0098] In an actual embodiment, according to the calculation formulas of material mechanics and fluid mechanics (such as the chassis wind resistance curve), the overall result is RPM=(159.2Q) / (S*2r / η).

[0099] Based on the number of system fans in use, the total system air volume can be calculated: Qx=n(RPM*(S*2r / η)) / 159.2.

[0100] In a preferred embodiment, S21 specifically includes:

[0101] Obtain the third speed calculation formula corresponding to the third speed association relationship: RPM=(159.2Q) / (S*2r / η); wherein RPM is the system fan speed value of a single system fan, Q is the system fan air volume value of a single system fan, S is the effective force area of ​​the impeller of the single system fan, r is the average force radius of the single system fan, and η is the efficiency of the single system fan.

[0102] In an actual embodiment, RPM=(159.2Q) / (S*2r / η) can be obtained by comprehensively analyzing the calculation formulas of material mechanics and fluid mechanics (such as the chassis wind resistance curve).

[0103] In a preferred embodiment, S22 specifically includes:

[0104] Obtain the number of system fans, and obtain the total system air volume value based on the number of system fans and the third speed calculation formula: Qx=n(RPM*(S*2r / η)) / 159.2; wherein n is the number of system fans, and Qx is the total system air volume value.

[0105] In an actual embodiment, the total air volume of the system can be obtained according to the number of system fans in use: Qx=n(RPM*(S*2r / η)) / 159.2.

[0106] In a preferred embodiment, S2 specifically includes:

[0107] Obtain the second speed calculation formula corresponding to the second speed association relationship: RPMture = [159.2 (Qx-Qp) / m] / (S gpu *2r gpu / η gpu ), where RPMture is the actual speed of the GPU fan, m is the number of GPU fans, S gpu is the effective force area of ​​the GPU fan impeller, r gpu is the average force radius of the GPU fan, η gpu is the efficiency of the GPU fan, Qp is the non-outlet air volume value;

[0108] The actual speed value of the GPU fan is calculated according to the second speed calculation formula and the actual GPU fan air volume value.

[0109] In a preferred embodiment, S1 specifically includes:

[0110] The first speed calculation formula corresponding to the first speed association relationship is obtained: RPMg=TgK+Bg; wherein RPMg is the theoretical speed value of the GPU fan, Tg is the current temperature value of a single GPU, and Bg and K are constants.

[0111] In an actual embodiment, the relationship between the GPU fan speed and the GPU temperature can be obtained according to the GPU heat dissipation design guide: RPMg=TgK+Bg.

[0112] In a preferred embodiment, S3 specifically includes:

[0113] The speed correction calculation formula is: RPMture-RPMg=[159.2(Qx-Qp) / m] / (S gpu *2r gpu / η gpu )-TgK-Bg;

[0114] The speed correction value is calculated according to the speed correction calculation formula.

[0115] In an actual embodiment, when RPMg<RPMture, it means that the system fan speed is high and the GPU fan speed is low. Under the wind pressure of the system fan, the GPU fan speed is increased, and the following information is fed back to the BMC:

[0116] The GPU fan speed value needs to be corrected based on the RPMg value. The correction formula is RPMture-RPMg=[159.2(Qx-Qp) / m] / (S gpu *2r gpu / η gpu )-TgK-Bg.

[0117] In a preferred embodiment, the correction method further comprises:

[0118] S4: When the theoretical speed value of the GPU fan is greater than or equal to the actual speed value of the GPU fan, the GPU fan speed information is sent to the BMC without correction.

[0119] In an actual embodiment, when RPMg≥RPMture, it means that the system fan speed is low and the GPU fan speed is high, and the GPU fan speed information is fed back to the BMC without correction.

[0120] In a preferred embodiment, S2 further includes:

[0121] The non-air outlet air volume value is obtained based on the test wind speed and air outlet area at the non-GPU air outlet position.

[0122] In a preferred embodiment, S2 further includes:

[0123] The non-air outlet air volume value is obtained according to the air volume thermal simulation information at the non-GPU air outlet position.

[0124] In an actual embodiment, the air volume Qp at the non-GPU air outlet position may be obtained through thermal simulation, or by testing the wind speed and the air outlet area.

[0125] In a preferred embodiment, the system fan is an 8056 fan, and S, r, and η are fixed parameters.

[0126] In an actual embodiment, for a given model of 8056 fan, S, r, and η are all fixed parameters, and the rotation speed and air volume of all system fans are in linear proportional relationship.

[0127] In a preferred embodiment, before S1, the correction method further includes:

[0128] When the server is powered on, the BMC obtains speed control information and controls the speed values ​​of all system fans; the speed control information includes GPU temperature information.

[0129] In a preferred implementation, the BMC obtains the speed control information and controls the speed values ​​of all system fans, specifically including:

[0130] The BMC controls the speed values ​​of all system fans according to a preset fan speed control mode; wherein the preset fan speed control mode includes one of the following: PID speed control, linear speed control, and segmented speed control.

[0131] In the actual implementation, after the server is turned on, the BMC obtains information such as the hard disk, memory, air inlet temperature, CPU temperature, GPU temperature, etc., controls the speed of the system fan, and ensures the heat dissipation of the entire machine. Common server fan speed control methods include but are not limited to PID speed control, linear, segmented, etc.; when the system heat dissipation is stable, the system fan speed can be stable.

[0132] In a practical embodiment, Figure 3 As shown, the GPU fan speed correction method provided by the embodiment of the present invention has the following specific implementation process:

[0133] First, according to the GPU heat dissipation design guide, the relationship between the GPU fan speed and the GPU temperature is obtained: RPMg=TgK+Bg; where RPMg represents the theoretical speed of the GPU fan, Tg represents the temperature of the GPU, and Bg represents a constant.

[0134] After the server is turned on, the BMC obtains information such as the hard disk, memory, air inlet temperature, CPU temperature, and GPU temperature to control the speed of the system fan and ensure the heat dissipation of the entire machine. Common server fan speed control methods include but are not limited to PID speed control, linear, and segmented speed control. When the system heat dissipation is stable, the system fan speed can be stable.

[0135] According to the calculation formulas of material mechanics and fluid mechanics (such as the chassis wind resistance curve), the overall result is RPM=(159.2Q) / (S*2r / η); where RPM is the rotation speed of the system fan, Q is the air volume of a single system fan, S is the effective force area of ​​the system fan impeller, r is the average force radius of the system fan, and η is the efficiency of the system fan.

[0136] For example, for a given model of 8056 fan, S, r, and η are all fixed parameters, and the speed and air volume of all system fans are in linear proportion.

[0137] Based on the number of system fans in use, the total system air volume can be calculated: Qx=n(RPM*(S*2r / η)) / 159.2; where n represents the number of system fans and Qx is the total system air volume.

[0138] In addition, combined with the wind resistance parameters of the server wind tunnel test, the correction value of the total air volume (i.e. the air volume in the opening area) can also be obtained; for the sake of simplicity of calculation, no correction is made here, and Qx is still used as the total air volume of the system.

[0139] The total air volume Qx of the system minus the air volume Qp at the non-GPU air outlet position under the system configuration (which can be obtained through thermal simulation or by testing wind speed and air outlet area) is obtained to obtain the actual through-air volume when 8 GPU cards are fully configured. The air volume of each GPU card is Qture = (Qx-Qp) / 8.

[0140] Then, according to the relationship between the fan air volume and the speed of the GPU card, the actual speed of the GPU fan is calculated; that is, RPMture = [159.2 (Qx-Qp) / m] / (S gpu *2r gpu / η gpu ).

[0141] When performing GPU fan anomaly detection, the GPU speed is corrected according to different GPU temperatures and GPU fan control strategies. The correction rules are as follows:

[0142] (1) When RPMg ≥ RPMture, it means that the system fan speed is low and the GPU fan speed is high, and the GPU fan speed information is fed back to the BMC without correction.

[0143] (2) When RPMg < RPMture, it means that the system fan speed is high and the GPU fan speed is low. The GPU fan speed is increased under the wind pressure of the system fan, and the following information is fed back to the BMC: The GPU fan speed value needs to be corrected based on the value of RPMg. The correction formula is RPMture-RPMg = [159.2 (Qx-Qp) / m] / (S gpu *2r gpu / η gpu )-TgK-Bg.

[0144] It should be noted that, although the various steps in the flowchart are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless there is a clear description in this article, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowchart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0145] Embodiment 2:

[0146] like Figure 4 As shown, an embodiment of the present invention further provides a GPU fan speed correction device, comprising:

[0147] A GPU fan theoretical speed value acquisition unit, used to acquire a GPU fan theoretical speed value according to a first speed association relationship and a current temperature value of a single GPU; wherein the first speed association relationship is an association relationship between a GPU fan speed value and a GPU temperature value;

[0148] A GPU fan actual speed value acquisition unit, used to acquire the GPU fan actual speed value according to the second speed association relationship and the GPU actual fan air volume value corresponding to a single GPU; wherein the second speed association relationship is the association relationship between the GPU fan speed value and the GPU fan air volume value;

[0149] The speed correction value acquisition unit is used to obtain the difference between the actual speed value of the GPU fan and the theoretical speed value of the GPU fan as the speed correction value when the theoretical speed value of the GPU fan is less than the actual speed value of the GPU fan, so as to correct the theoretical speed value of the GPU fan.

[0150] In a preferred embodiment, the GPU fan actual speed value acquisition unit specifically includes:

[0151] A total fan air volume value acquisition unit, used to acquire the total fan air volume value of all GPUs according to the system total air volume value of the server and the non-air outlet air volume value at the non-GPU air outlet position;

[0152] The GPU fan actual speed value acquisition unit is further used to acquire the GPU actual fan air volume value corresponding to a single GPU according to the total fan air volume value and the GPU quantity value.

[0153] In a preferred embodiment, the GPU fan actual speed value acquisition unit further includes:

[0154] A third rotation speed association relationship acquisition unit, used to acquire a third rotation speed association relationship, where the third rotation speed association relationship is an association relationship between a system fan rotation speed value and a system fan air volume value of a single system fan in the server;

[0155] The system total air volume value acquisition unit is used to acquire the system total air volume value according to the system fan quantity value and the third rotation speed association relationship.

[0156] In a preferred embodiment, the third speed association relationship acquisition unit specifically includes:

[0157] The third speed calculation formula acquisition unit is used to obtain the third speed calculation formula corresponding to the third speed association relationship: RPM=(159.2Q) / (S*2r / η); wherein RPM is the system fan speed value of a single system fan, Q is the system fan air volume value of a single system fan, S is the effective force area of ​​the impeller of the single system fan, r is the average force radius of the single system fan, and η is the efficiency of the single system fan.

[0158] In a preferred embodiment, the system total air volume value acquisition unit specifically includes:

[0159] The system total air volume value calculation formula acquisition unit is used to obtain the system fan quantity value, and obtain the system total air volume value calculation formula according to the system fan quantity value and the third speed calculation formula: Qx=n(RPM*(S*2r / η)) / 159.2; wherein n is the system fan quantity value, and Qx is the system total air volume value.

[0160] In a preferred embodiment, the GPU fan actual speed value acquisition unit specifically includes:

[0161] The second speed calculation formula acquisition unit is used to acquire the second speed calculation formula corresponding to the second speed association relationship: RPMture=[159.2(Qx-Qp) / m] / (S gpu *2r gpu / η gpu); where RPMture is the actual speed of the GPU fan, m is the number of GPU fans, Sgpu is the effective force area of ​​the GPU fan impeller, rgpu is the average force radius of the GPU fan, ηgpu is the efficiency of the GPU fan, and Qp is the non-outlet air volume value;

[0162] The GPU fan actual speed value calculation unit is used to calculate the GPU fan actual speed value according to the second speed calculation formula and the GPU actual fan air volume value.

[0163] In a preferred embodiment, the GPU fan theoretical speed value acquisition unit specifically includes:

[0164] The first speed calculation formula acquisition unit is used to obtain the first speed calculation formula corresponding to the first speed association relationship: RPMg=TgK+Bg; wherein RPMg is the theoretical speed value of the GPU fan, Tg is the current temperature value of a single GPU, and Bg and K are constants.

[0165] In a preferred embodiment, the speed correction value acquisition unit specifically includes:

[0166] The speed correction calculation formula acquisition unit is used to obtain the speed correction calculation formula: RPMture-RPMg=[159.2(Qx-Qp) / m] / (S gpu *2r gpu / η gpu )-TgK-Bg;

[0167] The speed correction value calculation unit is used to calculate the speed correction value according to the speed correction calculation formula.

[0168] In a preferred embodiment, the correction device further comprises:

[0169] The GPU fan speed non-correction information sending unit is used to send the GPU fan speed non-correction information to the BMC when the GPU fan theoretical speed value is greater than or equal to the GPU fan actual speed value.

[0170] In a preferred embodiment, the GPU fan actual speed value acquisition unit further includes:

[0171] The non-air outlet air volume value acquisition unit is used to acquire the non-air outlet air volume value according to the test wind speed and air outlet area at the non-GPU air outlet position.

[0172] In a preferred embodiment, the GPU fan actual speed value acquisition unit further includes:

[0173] The non-air outlet air volume value acquisition unit is used to acquire the non-air outlet air volume value according to the air volume thermal simulation information at the non-GPU air outlet position.

[0174] In a preferred embodiment, the correction device further comprises a BMC, which is used to:

[0175] When the server is powered on, the BMC obtains speed control information and controls the speed values ​​of all system fans; the speed control information includes GPU temperature information.

[0176] In a preferred embodiment, the BMC is also used for:

[0177] The BMC controls the speed values ​​of all system fans according to a preset fan speed control mode; wherein the preset fan speed control mode includes one of the following: PID speed control, linear speed control, and segmented speed control.

[0178] For the specific limitations of the above-mentioned device, please refer to the limitations of the method above, which will not be repeated here.

[0179] Each module in the above device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0180] Among them, Figure 5 As shown, the above-mentioned computer device can be a terminal, which includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0181] It can be understood that the structure shown in the above figure is only a block diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0182] The implementation of all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods.

[0183] Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0184] It should be noted that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A GPU fan speed correction method, characterized in that: include: Obtaining a theoretical speed value of a GPU fan according to a first speed association relationship and a current temperature value of a single GPU; wherein the first speed association relationship is an association relationship between a GPU fan speed value and a GPU temperature value; Obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to a single GPU; wherein the second speed association relationship is the association relationship between the GPU fan speed value and the GPU fan air volume value; When the theoretical speed value of the GPU fan is less than the actual speed value of the GPU fan, a difference between the actual speed value of the GPU fan and the theoretical speed value of the GPU fan is obtained as a speed correction value, and the theoretical speed value of the GPU fan is corrected.

2. The GPU fan speed correction method according to claim 1, characterized in that: Obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to a single GPU specifically includes: Obtain the total fan air volume value of all GPUs based on the total system air volume value of the server and the non-air outlet air volume value at the non-GPU air outlet position; The actual fan air volume value of the GPU corresponding to a single GPU is obtained according to the total fan air volume value and the GPU quantity value.

3. The GPU fan speed correction method according to claim 2, characterized in that: The method further includes: obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to the single GPU; Acquire a third rotation speed association relationship, where the third rotation speed association relationship is an association relationship between a system fan rotation speed value and a system fan air volume value of a single system fan in the server; The total air volume value of the system is obtained according to the number of system fans and the third rotation speed association relationship.

4. The GPU fan speed correction method according to claim 3, characterized in that: The third speed association relationship is obtained, specifically including: Obtain the third speed calculation formula corresponding to the third speed association relationship: RPM=(159.2Q) / (S*2r / η); wherein, RPM is the system fan speed value of a single system fan, Q is the system fan air volume value of a single system fan, S is the effective force area of ​​the impeller of a single system fan, r is the average force radius of a single system fan, and η is the efficiency of a single system fan.

5. The GPU fan speed correction method according to claim 4, characterized in that: The total air volume value of the system is obtained according to the number of system fans and the third rotation speed association relationship, specifically including: Obtain the number of system fans, and obtain the calculation formula for the total system air volume value based on the number of system fans and the third speed calculation formula: Qx=n(RPM*(S*2r / η)) / 159.2; wherein n is the number of system fans, and Qx is the total system air volume value.

6. The GPU fan speed correction method according to claim 5, characterized in that: Obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to a single GPU specifically includes: Obtain the second speed calculation formula corresponding to the second speed association relationship: RPMture=[159.2(Qx-Qp) / m] / (S gpu *2r gpu / η gpu ), where RPMture is the actual speed of the GPU fan, m is the number of GPU fans, S gpu is the effective force area of ​​the GPU fan impeller, r gpu is the average force radius of the GPU fan, η gpu is the efficiency of the GPU fan, Qp is the non-outlet air volume value; The actual speed value of the GPU fan is calculated according to the second speed calculation formula and the actual GPU fan air volume value.

7. The GPU fan speed correction method according to claim 6, characterized in that: Obtaining a theoretical speed value of a GPU fan according to the first speed association relationship and a current temperature value of a single GPU includes: Obtain a first speed calculation formula corresponding to the first speed association relationship: RPMg=TgK+Bg; wherein RPMg is the theoretical speed value of the GPU fan, Tg is the current temperature value of a single GPU, and Bg and K are constants.

8. The GPU fan speed correction method according to claim 7, characterized in that: Obtaining the difference between the actual speed value of the GPU fan and the theoretical speed value of the GPU fan as a speed correction value, specifically includes: The speed correction calculation formula is: RPMture-RPMg=[159.2(Qx-Qp) / m] / (S gpu *2r gpu / η gpu )-TgK-Bg; The speed correction value is calculated according to the speed correction calculation formula.

9. The GPU fan speed correction method according to claim 1, characterized in that: The correction method also includes: When the theoretical speed value of the GPU fan is greater than or equal to the actual speed value of the GPU fan, the GPU fan speed information is sent to the BMC without being corrected.

10. The GPU fan speed correction method according to claim 2, characterized in that: The method further includes: obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to the single GPU; The non-air outlet air volume value is obtained according to the test wind speed and air outlet area at the non-GPU air outlet position.

11. The GPU fan speed correction method according to claim 2, characterized in that: The method further includes: obtaining the actual speed value of the GPU fan according to the second speed association relationship and the actual GPU fan air volume value corresponding to the single GPU; The non-air outlet air volume value is obtained according to the air volume thermal simulation information at the non-GPU air outlet position.

12. The GPU fan speed correction method according to claim 4, characterized in that: The system fan is an 8056 fan, and S, r, and η are fixed parameters.

13. The GPU fan speed correction method according to claim 1, characterized in that: Before obtaining the theoretical speed value of the GPU fan according to the first speed association relationship and the current temperature value of a single GPU, the correction method further includes: When the server is in the power-on state, the BMC obtains the rotation speed control information and controls the rotation speed values ​​of all system fans; wherein the rotation speed control information includes GPU temperature information.

14. The GPU fan speed correction method according to claim 13, characterized in that: The BMC obtains speed control information and controls the speed values ​​of all system fans, including: The BMC controls the speed values ​​of all system fans according to a preset fan speed regulation mode; wherein the preset fan speed regulation mode includes one of the following: PID speed regulation, linear speed regulation, and segmented speed regulation.

15. A GPU fan speed correction device, characterized in that: include: A GPU fan theoretical speed value acquisition unit, used to acquire a GPU fan theoretical speed value according to a first speed association relationship and a current temperature value of a single GPU; wherein the first speed association relationship is an association relationship between a GPU fan speed value and a GPU temperature value; A GPU fan actual speed value acquisition unit, used to acquire the GPU fan actual speed value according to a second speed association relationship and a GPU fan actual air volume value corresponding to a single GPU; wherein the second speed association relationship is an association relationship between the GPU fan speed value and the GPU fan air volume value; The speed correction value acquisition unit is used to obtain the difference between the actual speed value of the GPU fan and the theoretical speed value of the GPU fan as the speed correction value when the theoretical speed value of the GPU fan is less than the actual speed value of the GPU fan, and to correct the theoretical speed value of the GPU fan.

16. The GPU fan speed correction device according to claim 15, characterized in that: The GPU fan actual speed value acquisition unit specifically includes: A total fan air volume value acquisition unit, used to acquire the total fan air volume value of all GPUs according to the system total air volume value of the server and the non-air outlet air volume value at the non-GPU air outlet position; The GPU fan actual speed value acquisition unit is further used to acquire the GPU actual fan air volume value corresponding to a single GPU according to the total fan air volume value and the GPU quantity value.

17. The GPU fan speed correction device according to claim 16, characterized in that: The GPU fan actual speed value acquisition unit also includes: A third rotation speed association relationship acquisition unit, configured to acquire a third rotation speed association relationship, wherein the third rotation speed association relationship is an association relationship between a system fan rotation speed value and a system fan air volume value of a single system fan in the server; The system total air volume value acquisition unit is used to acquire the system total air volume value according to the system fan quantity value and the third rotation speed association relationship.

18. A computer device comprising a memory, a processor and a computer program, wherein the computer program is stored in the memory and can be run on the processor, wherein: When the processor executes the computer program, the steps of the GPU fan speed correction method according to any one of claims 1 to 14 are implemented.

19. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the GPU fan speed correction method according to any one of claims 1 to 14 are implemented.

20. A server, characterized in that: The invention comprises a GPU and a GPU fan, wherein the rotation speed of the GPU fan is corrected by the GPU fan rotation speed correction method described in any one of claims 1 to 14.

Images