A supercomputer computing power management system based on artificial intelligence

Abstract

The present invention relates to the field of computer technology, and in particular to an artificial intelligence-based supercomputer computing power management system, comprising: a management module, which is used to run synchronously with a computer and monitor the running parameters of an installation program on the computer in real time when the computer is running; a capture module, which is used to receive the running parameters of the installation program on the computer monitored by the management module, and capture active running programs based on running parameter analysis; a configuration module, which is used to obtain the capture results of active running programs in the capture module. The present invention analyzes the activity of real-time running programs on the computer by monitoring the running parameters of real-time running programs on the computer, and thus configures independent computing power for each running program based on the activity analysis results, thereby realizing the distribution of computer computing power, ensuring the stable operation of each running program on the computer, and avoiding the problems of subsequent new running programs being slow and stuck due to unbalanced computer computing power configuration.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a supercomputer computing power management system based on artificial intelligence. Background Art

[0002] A supercomputer is an extremely powerful computing device that consists of numerous processors and massive memory. It has super computing capabilities and can process large and complex data and perform difficult computing tasks in a very short time. It is widely used in many fields such as weather forecasting, scientific research, engineering design, artificial intelligence, etc., providing key support for promoting scientific and technological progress and solving complex problems.

[0003] The invention patent with application number 202010288797.5 discloses a method for improving the computing power and reducing the power consumption computing power ratio of a computer with multiple computing cores, characterized in that the method includes: in each chip of the computer with multiple computing cores, a main time-frequency unit and at least one auxiliary time-frequency unit are set; the clock frequency of the main time-frequency unit, the computing core in the chip whose clock frequency is provided by the main time-frequency unit, and the computing core in the chip whose clock frequency is provided by the auxiliary time-frequency unit are determined through testing to meet the requirement that the normal operation pass rate of the computing core in the chip is greater than or equal to the set pass rate difference: the test is performed when the computer with multiple computing cores is initialized.

[0004] The application aims to solve the problem that "each computing core in a computer with multiple computing cores, under ideal conditions, is expected to achieve the highest computing power and the lowest power consumption ratio. However, considering the stability of the operation of the entire computer and the area and power consumption of the time-frequency unit such as the phase-locked loop module (PLL) or the frequency-locked loop module (FLL) that provides the clock frequency for each computing core, all computing cores in a chip are provided with clock frequencies by a time-frequency unit arranged in the chip at the same time. That is to say, each chip in the computer has only one time-frequency unit for providing a unified clock frequency for all computing cores in the chip. In this case, in order to ensure that all computing cores in the chip in the computer can operate normally, the clock frequency provided by the time-frequency unit of the chip is determined by the clock frequency required by the computing core with the worst computing power. For other computing cores in the chip, if they operate based on the determined clock frequency, their highest computing power cannot be achieved. In order to increase the computing power, the operating voltage of the chip will be increased, and the power consumption will increase accordingly, which brings about the loss of computing power and the increase of the power consumption-computing power ratio of the computer."

[0005] However, there is still a lack of reliable management services for computer computing power scheduling in the supercomputer field, which results in uneven distribution of computing power when supercomputers run multiple tasks at the same time, leading to a waste of supercomputer computing power, and causing subsequent computer programs to experience problems such as delays and freezes.

[0006] To this end, we proposed an artificial intelligence-based supercomputer computing power management system. Summary of the invention

[0007] In view of the above-mentioned shortcomings of the prior art, the present invention provides a supercomputer computing power management system based on artificial intelligence, which solves the technical problems raised in the above-mentioned background technology.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0009] A supercomputer computing power management system based on artificial intelligence, comprising:

[0010] The management module is used to run synchronously with the computer and monitor the running parameters of the installed program on the computer in real time when the computer is running; the capture module is used to receive the running parameters of the installed program on the computer monitored by the management module, and capture the actively running program based on the running parameter analysis; the configuration module is used to obtain the capture results of the actively running program in the capture module, and configure the computer computing power based on the capture results of the actively running program; the monitoring module is used to monitor the remaining computing power of the computer in real time; the refresh module is used to refresh the system operation and jump to the capture module operation stage; the feedback module is used to feedback the computing power configuration results of the programs running on the computer in real time.

[0011] Furthermore, the management module is internally provided with submodules, including:

[0012] A recording unit, used for recording the running parameters of the installation program on the computer monitored in real time by the management module;

[0013] Among them, the recording unit runs synchronously with the management module, and records the running parameters of the installation program on the computer monitored by the management module. When the recording unit records the running parameters of the installation program on the computer, based on the running parameter source program, the running parameters of the installation program on the computer are distinguished and recorded. The running parameters of the installation program on the computer include: running start time, running end time, running real-time memory occupied, and running task queue process.

[0014] Furthermore, the capture module is provided with submodules at the lower level, including:

[0015] A calling unit, used to call the running parameters of the installation program on the computer recorded in the recording unit;

[0016] An analyzing unit, used for acquiring the running parameters of the installation program on the computer retrieved by the retrieving unit, and analyzing the activity of each program based on the running parameters of the installation program on the computer;

[0017] The calling unit runs with the recording unit as the calling target, and calls the running parameters of the installation program on the computer in the recording unit. The recording unit synchronously resets the internally recorded parameters every time the computer is shut down.

[0018] Furthermore, the program activity analysis logic in the analysis unit is expressed as:

[0019]

[0020] Where: is the program activity performance value; t is the cumulative running time of the installation program on the computer; γ is the compensation; n is the total number of real-time memory usage records in the installation program running parameters on the computer; g i The real-time memory occupied by the installation program running on the computer recorded for the i-th time; in The processing rate of the task queue process currently running for the installation program on the computer; p over Added a rate for the currently running task queue process of the installed program on the computer;

[0021] Among them, the program activity performance value The larger the value, the higher the program activity. Conversely, the lower the program activity. The compensation γ value is 0 or 1. If the installation program on the computer is not finished running, the compensation γ value is 1. Conversely, the compensation γ value is 0. Table pair Seeking the average.

[0022] Furthermore, the processing rate of the task queue process currently running on the computer is p in That is: the average processing time of the tasks that have been completed in the task queue of the current installation program on the computer; the rate of new processes in the task queue of the current installation program on the computer p over That is: the average additional time of running tasks newly added to the running task queue by the installation program on the computer based on user operations;

[0023] Based on the program activity analysis logic, all programs currently running on the computer are analyzed. The capture module calculates the value based on the maximum The program corresponding to the value is used as the capture target.

[0024] Furthermore, the computer computing power configuration logic in the configuration module is expressed as:

[0025]

[0026] In the formula: m is the unit memory of the computer; M is the total amount of computer memory; u is the set of running programs that are not yet finished running on the computer; is the activity performance value of the vth group of running programs; m TP (α) is the memory configured for the running program α that is not yet finished running on the computer; M left The remaining allocatable memory; is the activity performance value of the running program α that is not yet finished running on the computer;

[0027] Wherein, based on formula (2), the running program other than the running program captured by the capture module is configured with running memory, and the running program captured by the capture module requires running memory less than hour, The memory required for running the program captured by the capture module is not less than When M′, M left =MM′, then the computing power of the active running program configuration captured by the capture module is MM left .

[0028] Furthermore, the refresh module uses the remaining monitoring result of the computer computing power in the monitoring module as a trigger condition, executes the trigger, refreshes the system operation, and jumps to the capture module operation stage;

[0029] An operation cycle is set in the monitoring module. The monitoring module continuously monitors the remaining computer computing power based on the operation cycle. When the monitoring module runs for three consecutive operation cycles and monitors that the remaining computer computing power is not less than 50% or is continuously less than 10%, the refresh module is triggered to run.

[0030] Furthermore, after the configuration module configures the running program that is not yet finished running on the computer, when a new running program is generated on the computer, the original configuration of the running program closed on the current computer is configured for the new running program, or the activity performance value of the current program closed on the computer is displayed. The lowest running program is to close the running program configuration and configure it in the new running program.

[0031] Furthermore, the feedback module is provided with submodules at the lower level, including:

[0032] An application unit is used to traverse the computing power configuration results of the running programs on the computer fed back in the feedback module, store the computing power configuration results of the running programs, and configure computing power for the corresponding running programs based on the stored computing power configuration results of the running programs when the computer is running next time;

[0033] Among them, when the system runs for the second time and subsequently, the system skips the capture module and configuration stage, and the monitoring module runs synchronously with the management module. The feedback module is integrated by the control panel, and the feedback module feeds back the computing power configuration results of the program running on the computer, that is, displays the computing power configuration results of the program running on the computer.

[0034] Furthermore, the management module is interactively connected to a recording unit via a local area network, the management module is interactively connected to a capture module via a local area network, the capture module is interactively connected to a retrieval unit and an analysis unit via a local area network, the retrieval unit is interactively connected to the recording unit via a local area network, the capture module is interactively connected to a configuration module via a local area network, the configuration module is interactively connected to a monitoring module, a refresh module and a feedback module via a local area network, the refresh module is interactively connected to the capture module and the configuration module via a local area network, and the feedback module is interactively connected to an application unit via a local area network.

[0035] Compared with the known public technology, the technical solution provided by the present invention has the following beneficial effects:

[0036] The present invention provides a supercomputer computing power management system based on artificial intelligence. During operation, the system analyzes the activity of real-time running programs on the computer by monitoring the running parameters of the real-time running programs on the computer, and then configures independent computing power for each running program based on the activity analysis result, thereby realizing the distribution of computer computing power, ensuring the stable operation of each running program on the computer, and avoiding the problem of subsequent new running programs being slow or stuck due to unbalanced computer computing power configuration. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0038] Figure 1 The figure is a schematic diagram of the structure of a supercomputer computing power management system based on artificial intelligence. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments 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.

[0040] The present invention will be further described below in conjunction with the embodiments.

[0041] Embodiment 1:

[0042] A supercomputer computing power management system based on artificial intelligence in this embodiment, such as Figure 1 As shown, including:

[0043] The management module is used to run synchronously with the computer and monitor the running parameters of the installation program on the computer in real time when the computer is running;

[0044] The management module is internally provided with submodules, including:

[0045] A recording unit, used to record the running parameters of the installation program on the computer monitored in real time by the management module;

[0046] The recording unit runs synchronously with the management module and records the running parameters of the installation program on the computer monitored by the management module. When the recording unit records the running parameters of the installation program on the computer, the running parameters of the installation program on the computer are distinguished and recorded based on the running parameter source program. The running parameters of the installation program on the computer include: running start time, running end time, running real-time memory usage, and running task queue process;

[0047] A capture module, used to receive the running parameters of the installation program on the computer monitored by the management module, and capture the active running program based on the running parameter analysis;

[0048] The capture module is equipped with submodules, including:

[0049] A calling unit, used to call the running parameters of the installation program on the computer recorded in the recording unit;

[0050] An analyzing unit, used for acquiring the running parameters of the installation program on the computer retrieved by the retrieving unit, and analyzing the activity of each program based on the running parameters of the installation program on the computer;

[0051] The calling unit runs with the recording unit as the calling target, and calls the running parameters of the installation program on the computer in the recording unit. The recording unit synchronously resets the internally recorded parameters every time the computer is shut down.

[0052] The program activity analysis logic in the analysis unit is expressed as:

[0053]

[0054] Where: is the program activity performance value; t is the cumulative running time of the installation program on the computer; γ is the compensation; n is the total number of real-time memory usage records in the installation program running parameters on the computer; g i The real-time memory occupied by the installation program running on the computer recorded for the i-th time; in The processing rate of the task queue process currently running for the installation program on the computer; p over Added a rate for the currently running task queue process of the installed program on the computer;

[0055] Among them, the program activity performance value The larger the value, the higher the program activity. Conversely, the lower the program activity. The compensation γ value is 0 or 1. If the installation program on the computer is not finished running, the compensation γ value is 1. Conversely, the compensation γ value is 0. Table pair ;

[0056] The processing rate of the task queue process currently running on the computer is pi n That is: the average processing time of the tasks that have been completed in the task queue of the current installation program on the computer; the rate of new processes in the task queue of the current installation program on the computer p over That is: the average additional time of running tasks newly added to the running task queue by the installation program on the computer based on user operations;

[0057] Based on the program activity analysis logic, all programs currently running on the computer are analyzed. The capture module calculates the value based on the maximum The program corresponding to the value is used as the capture target;

[0058] A configuration module, used to obtain the capture results of the active running programs in the capture module, and configure the computing power of the computer based on the capture results of the active running programs;

[0059] Monitoring module, used to monitor the remaining computing power of the computer in real time;

[0060] The refresh module is used to refresh the system operation and jump to the capture module operation stage;

[0061] Feedback module, used to provide real-time feedback on the computing power configuration results of programs running on the computer;

[0062] The feedback module is provided with submodules at the lower level, including:

[0063] An application unit is used to traverse the computing power configuration results of the running programs on the computer fed back in the feedback module, store the computing power configuration results of the running programs, and configure computing power for the corresponding running programs based on the stored computing power configuration results of the running programs when the computer is running next time;

[0064] When the system is run for the second time and subsequently, the system skips the capture module and the configuration stage, the monitoring module runs synchronously with the management module, the feedback module is integrated by the control panel, and the feedback module feeds back the computing power configuration result of the program running on the computer, that is, displays the computing power configuration result of the program running on the computer;

[0065] The management module is interactively connected with a recording unit through a local area network. The management module is interactively connected with a capture module through a local area network. The capture module is interactively connected with a retrieval unit and an analysis unit through the local area network. The retrieval unit is interactively connected with the recording unit through the local area network. The capture module is interactively connected with a configuration module through the local area network. The configuration module is interactively connected with a monitoring module, a refresh module and a feedback module through the local area network. The refresh module is interactively connected with the capture module and the configuration module through the local area network. The feedback module is interactively connected with an application unit through the local area network.

[0066] In this embodiment, the management module runs synchronously with the computer, and monitors the running parameters of the installation program on the computer in real time when the computer is running. The recording unit synchronously records the running parameters of the installation program on the computer monitored by the management module in real time. The capture module post-operates to receive the running parameters of the installation program on the computer monitored by the management module, and captures the actively running programs based on the running parameter analysis. The calling unit synchronously calls the running parameters of the installation program on the computer recorded in the recording unit. The analysis unit obtains the running parameters of the installation program on the computer called in the calling unit in real time, and analyzes the activity of each program based on the running parameters of the installation program on the computer. The configuration module further obtains the capture result of the actively running program in the capture module, and configures the computer computing power based on the capture result of the actively running program. The monitoring module then monitors the remaining computing power of the computer in real time. The refresh module refreshes the system operation in real time and jumps to the capture module operation stage. Finally, the computing power configuration result of the running program on the computer is fed back in real time through the feedback module, and based on the computing power configuration result of the running program on the computer fed back in the feedback module traversed by the application unit, the running program computing power configuration result is stored. When the computer runs next time, the computing power is configured for the corresponding running program based on the stored running program computing power configuration result.

[0067] Through the above-mentioned embodiments, the system brings effective operation computing power management to the supercomputer, avoiding the problem of unbalanced operation computing power distribution caused by the computer having no computing power configuration logic.

[0068] Embodiment 2:

[0069] In terms of specific implementation, based on Example 1, this example refers to Figure 1 A supercomputer computing power management system based on artificial intelligence in Example 1 is further described in detail:

[0070] The computer computing power configuration logic in the configuration module is expressed as:

[0071]

[0072] In the formula: m is the unit memory of the computer; M is the total amount of computer memory; u is the set of running programs that are not yet finished running on the computer; is the activity performance value of the vth group of running programs; m TP (α) is the memory configured for the running program α that is not yet finished running on the computer; M left The remaining allocatable memory; is the activity performance value of the running program α that is not yet finished running on the computer;

[0073] Wherein, based on formula (2), the running program other than the running program captured by the capture module is configured with running memory, and the running program captured by the capture module requires running memory less than hour, The memory required for running the program captured by the capture module is not less than When M′, M left =MM′, then the computing power of the active running program configuration captured by the capture module is MM left .

[0074] Through the above logic formula, the computer computing power configuration logic is further limited, providing further operation data support for the operation of subsequent operation modules of the system in Example 1.

[0075] like Figure 1 As shown, the refresh module uses the remaining monitoring result of the computer computing power in the monitoring module as a trigger condition, executes the trigger, refreshes the system operation, and jumps to the capture module operation stage;

[0076] An operation cycle is set in the monitoring module. The monitoring module continuously monitors the remaining computing power of the computer based on the operation cycle. When the monitoring module runs for three consecutive operation cycles and detects that the remaining computing power of the computer is not less than 50% or is continuously less than 10%, the refresh module is triggered to run;

[0077] After the configuration module is configured to run the running program that is not finished on the computer, when a new running program is generated on the computer, the original configuration of the closed running program on the current computer is configured to run the new running program, or the activity performance value of the closed computer current program is displayed. The lowest running program is to close the running program configuration and configure it in the new running program.

[0078] Through the above settings, the operating logic of the refresh module and the system in Example 1 is further limited, ensuring that the system can continue to serve the supercomputer and bring a stable management effect to the computer computing power.

[0079] In summary, during the operation of the system in the above embodiment, the activity of the real-time running programs on the computer is analyzed by monitoring the running parameters of the real-time running programs on the computer, and then based on the activity analysis results, independent computing power is configured for each running program, thereby realizing the distribution of computer computing power, ensuring the stable operation of each running program on the computer, and avoiding the problem of subsequent new running programs being slow or stuck due to unbalanced configuration of computer computing power.

[0080] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A supercomputer computing power management system based on artificial intelligence, characterized in that: include: The management module is used to run synchronously with the computer and monitor the running parameters of the installation program on the computer in real time when the computer is running; A capture module, used to receive the running parameters of the installation program on the computer monitored by the management module, and capture the active running program based on the running parameter analysis; A configuration module, used to obtain the capture results of the active running programs in the capture module, and configure the computing power of the computer based on the capture results of the active running programs; Monitoring module, used to monitor the remaining computing power of the computer in real time; The refresh module is used to refresh the system operation and jump to the capture module operation stage; Feedback module, used to provide real-time feedback on the computing power configuration results of programs running on the computer; The capture module is provided with submodules at the lower level, including: A calling unit, used to call the running parameters of the installation program on the computer recorded in the recording unit; An analyzing unit, used for acquiring the running parameters of the installation program on the computer retrieved by the retrieving unit, and analyzing the activity of each program based on the running parameters of the installation program on the computer; The calling unit runs with the recording unit as the calling target, and calls the running parameters of the installation program on the computer in the recording unit. The recording unit synchronously resets the internally recorded parameters every time the computer is shut down. The program activity analysis logic in the analysis unit is expressed as: Where: is the program activity performance value; t is the cumulative running time of the installation program on the computer; γ is the compensation; n is the total number of real-time memory usage records in the installation program running parameters on the computer; g i The real-time memory occupied by the installation program running on the computer recorded for the i-th time; in The processing rate of the task queue process currently running for the installation program on the computer; p over Added a rate for the currently running task queue process of the installed program on the computer; Among them, the program activity performance value The larger the value, the higher the program activity. Conversely, the lower the program activity. The compensation γ value is 0 or 1. If the installation program on the computer is not finished running, the compensation γ value is 1. Conversely, the compensation γ value is 0. Table pair ; The processing rate of the task queue process currently running on the computer is p in That is: the average processing time of the tasks that have been completed in the task queue of the current installation program on the computer; the rate of new processes in the task queue of the current installation program on the computer p over That is: the average additional time of running tasks newly added to the running task queue by the installation program on the computer based on user operations; Based on the program activity analysis logic, all programs currently running on the computer are analyzed. The capture module calculates the value based on the maximum The program corresponding to the value is used as the capture target; The computer computing power configuration logic in the configuration module is expressed as: In the formula: m is the unit memory of the computer; M is the total amount of computer memory; u is the set of running programs that are not yet finished running on the computer; is the activity performance value of the vth group of running programs; m TP (α) is the memory configured for the running program α that is not yet finished running on the computer; M left The remaining allocatable memory; is the activity performance value of the running program α that is not yet finished running on the computer; Wherein, based on formula (2), the running program other than the running program captured by the capture module is configured with running memory, and the running program captured by the capture module requires running memory less than hour, The memory required for running the program captured by the capture module is not less than When M′, M left =MM′, then the computing power of the active running program configuration captured by the capture module is MM left .

2. The supercomputer computing power management system based on artificial intelligence according to claim 1 is characterized in that: The management module is internally provided with submodules, including: A recording unit, used to record the running parameters of the installation program on the computer monitored in real time by the management module; Among them, the recording unit runs synchronously with the management module, and records the running parameters of the installation program on the computer monitored by the management module. When the recording unit records the running parameters of the installation program on the computer, based on the running parameter source program, the running parameters of the installation program on the computer are distinguished and recorded. The running parameters of the installation program on the computer include: running start time, running end time, running real-time memory occupied, and running task queue process.

3. The supercomputer computing power management system based on artificial intelligence according to claim 1 is characterized in that: The refresh module uses the remaining monitoring result of the computer computing power in the monitoring module as a trigger condition, executes the trigger, refreshes the system operation, and jumps to the capture module operation stage; An operation cycle is set in the monitoring module. The monitoring module continuously monitors the remaining computer computing power based on the operation cycle. When the monitoring module runs for three consecutive operation cycles and monitors that the remaining computer computing power is not less than 50% or is continuously less than 10%, the refresh module is triggered to run.

4. The supercomputer computing power management system based on artificial intelligence according to claim 3 is characterized in that: After the configuration module is configured to store the running program that is not yet finished on the computer, when a new running program is generated on the computer, the original configuration of the closed running program on the current computer is configured to store the new running program, or the activity performance value of the closed current program on the computer is displayed. The lowest running program is to close the running program configuration and configure it in the new running program.

5. The supercomputer computing power management system based on artificial intelligence according to claim 1 is characterized in that: The feedback module is provided with submodules at the lower level, including: An application unit is used to traverse the computing power configuration results of the running programs on the computer fed back in the feedback module, store the computing power configuration results of the running programs, and configure computing power for the corresponding running programs based on the stored computing power configuration results of the running programs when the computer is running next time; Among them, when the system runs for the second time and subsequently, the system skips the capture module and configuration stage, and the monitoring module runs synchronously with the management module. The feedback module is integrated by the control panel, and the feedback module feeds back the computing power configuration results of the program running on the computer, that is, displays the computing power configuration results of the program running on the computer.

6. The supercomputer computing power management system based on artificial intelligence according to claim 1 is characterized in that: The management module is interactively connected to a recording unit via a local area network, the management module is interactively connected to a capture module via a local area network, the capture module is interactively connected to a retrieval unit and an analysis unit at a lower level via a local area network, the retrieval unit is interactively connected to the recording unit via a local area network, the capture module is interactively connected to a configuration module via a local area network, the configuration module is interactively connected to a monitoring module, a refresh module and a feedback module via a local area network, the refresh module is interactively connected to the capture module and the configuration module via a local area network, and the feedback module is interactively connected to an application unit at a lower level via a local area network.

Images