Safety control method and system for AI server

1. A safety control method for an Artificial Intelligence (AI) server, which is applied in a Field Programmable Gate Array (FPGA), the safety control method comprising:

obtaining a current electrical current and a current power of a Graphics Processing Unit (GPU) in the AI server according to a preset frequency;

when the GPU satisfies a first control privilege transfer requirement, taking over control privilege of a heat dissipation system from a Baseboard Management Controller (BMC); and

controlling the heat dissipation system according to the current electrical current and the current power of the GPU;

wherein the first control privilege transfer requirement comprises: the current electrical current of the GPU exceeds a preset electrical current, or a rate of change of the current electrical current of the GPU exceeds a preset rate of change of electrical current, or the current power of the GPU exceeds a first preset power.

2. The safety control method according to claim 1 , further comprising:

according to the current electrical current and the current power of the GPU, controlling the GPU to lower a frequency to make the current electrical current of GPU lower than the preset electrical current.

3. The safety control method according to claim 1 , wherein controlling the heat dissipation system according to the current electrical current and the current power of the GPU comprises:

according to the current electrical current and the current power of the GPU, controlling the heat dissipation system to dissipate heat at full speed.

4. The safety control method according to claim 1 , wherein after obtaining the current electrical current and the current power of the GPU in the AI server according to the preset frequency, the safety control method further comprises:

obtaining a current power of a power supply module according to the preset frequency; and

accordingly, the first control privilege transfer requirement comprises: the current electrical current of the GPU exceeds the preset electrical current, or the rate of change of the current electrical current of the GPU exceeds the preset rate of change of electrical current, or the current power of the GPU exceeds the first preset power, or a rate of change of the current power of the power supply module exceeds a preset rate of change of power, or the current power of the power supply module exceeds a second preset power.

5. The safety control method according to claim 1 , further comprising:

when the GPU satisfies a second control privilege transfer requirement, returning the control privilege of the heat dissipation system to the BMC;

wherein the second control privilege transfer requirement comprises: the current electrical current of the GPU does not exceed the preset electrical current, the rate of change of the current electrical current of the GPU does not exceed the preset rate of change of electrical current, and the current power of the GPU does not exceed the first preset power.

6. The safety control method according to claim 5 , wherein the preset frequency is a sampling frequency at a millisecond level.

7. The safety control method according to claim 5 , wherein the preset frequency is a sampling frequency at a microsecond level.

8. The safety control method according to claim 1 , wherein the preset frequency is correspondingly and dynamically set among a set of values according to a fluctuation speed of different sampling targets.

9. A safety control system for an Artificial Intelligence (AI) server, the safety control system comprising a Field Programmable Gate Array (FPGA) and a Baseboard Management Controller (BMC), the FPGA being configured to:

obtain a current electrical current and a current power of a Graphics Processing Unit (GPU) in the AI server according to a preset frequency;

take over, when the GPU satisfies a first control privilege transfer requirement, control privilege of a heat dissipation system from the BMC; and

control the heat dissipation system according to the current electrical current and the current power of the GPU;

wherein the first control privilege transfer requirement comprises: the current electrical current of the GPU exceeds a preset electrical current, or a rate of change of the current electrical current of the GPU exceeds a preset rate of change of electrical current, or the current power of the GPU exceeds a first preset power.

10. The safety control system according to claim 9 , wherein the FPGA is further configured to:

control, according to the current electrical current and the current power of the GPU, the GPU to lower a frequency to make the current electrical current of GPU lower than the preset electrical current.

11. The safety control system according to claim 9 , wherein the FPGA is further configured to obtain a current power of a power supply module according to the preset frequency; and

accordingly, the first control privilege transfer requirement comprises: the current electrical current of the GPU exceeds the preset electrical current, or the rate of change of the current electrical current of the GPU exceeds the preset rate of change of electrical current, or the current power of the GPU exceeds the first preset power, or a rate of change of the current power of the power supply module exceeds a preset rate of change of power, or the current power of the power supply module exceeds a second preset power.

12. The safety control system according to claim 9 , wherein the FPGA is further configured to return, when the GPU satisfies a second control privilege transfer requirement, the control privilege of the heat dissipation system to the BMC;

wherein the second control privilege transfer requirement comprises:

the current electrical current of the GPU does not exceed the preset electrical current, the rate of change of the current electrical current of the GPU does not exceed the preset rate of change of electrical current, and the current power of the GPU does not exceed the first preset power.

13. The safety control system according to claim 9 , wherein the FPGA is further configured to: control, according to the current electrical current and the current power of the GPU, the heat dissipation system to dissipate heat at full speed.

14. The safety control system according to claim 9 , wherein the preset frequency is correspondingly and dynamically set among a set of values according to a fluctuation speed of different sampling targets.

15. The safety control system according to claim 12 , wherein the preset frequency is a sampling frequency at a millisecond level.

16. The safety control system according to claim 12 , wherein the preset frequency is a sampling frequency at a microsecond level.

17. One or more non-transitory computer-readable storage media storing computer-readable instructions, the computer-readable instructions, when executed by one or more processors, enabling the one or more processors to perform following operations applied in a Field Programmable Gate Array (FPGA):

obtaining a current electrical current and a current power of a Graphics Processing Unit (GPU) in the AI server according to a preset frequency;

when the GPU satisfies a first control privilege transfer requirement, taking over control privilege of a heat dissipation system from a Baseboard Management Controller (BMC); and

controlling the heat dissipation system according to the current electrical current and the current power of the GPU;

wherein the first control privilege transfer requirement comprises: the current electrical current of the GPU exceeds a preset electrical current, or a rate of change of the current electrical current of the GPU exceeds a preset rate of change of electrical current, or the current power of the GPU exceeds a first preset power.

18. The non-transitory computer-readable storage media according to claim 17 , wherein the computer-readable instructions, when executed by one or more processors, enable the one or more processors to further perform following operations applied in the FPGA: according to the current electrical current and the current power of the GPU, controlling the GPU to lower a frequency to make the current electrical current of GPU lower than the preset electrical current.

19. The non-transitory computer-readable storage media according to claim 17 , wherein the computer-readable instructions, when executed by one or more processors, enable the one or more processors to perform following operations applied in the FPGA: according to the current electrical current and the current power of the GPU, controlling the heat dissipation system to dissipate heat at full speed.

20. The non-transitory computer-readable storage media according to claim 17 , wherein the computer-readable instructions, when executed by one or more processors, enable the one or more processors to further perform following operations applied in the FPGA after obtaining the current electrical current and the current power of the GPU in the AI server according to the preset frequency: obtaining a current power of a power supply module according to the preset frequency; and accordingly, the first control privilege transfer requirement comprises: the current electrical current of the GPU exceeds the preset electrical current, or the rate of change of the current electrical current of the GPU exceeds the preset rate of change of electrical current, or the current power of the GPU exceeds the first preset power, or a rate of change of the current power of the power supply module exceeds a preset rate of change of power, or the current power of the power supply module exceeds a second preset power.