Calculation power peak shifting scheduling method and device for intelligent calculation center of common calculation power

Abstract

The invention provides a computing power off-peak scheduling method and device for a common computing power intelligent computing center, and relates to the technical field of intelligent computing centers, intelligent computing centers and computing power infrastructures, and the method comprises the steps: S1, receiving a computing power task request; s2, acquiring computing power resource information corresponding to the task cycle; s3, determining a target time period according to the computing power resource information; and S4, running the task code in the target time period so as to process the target computing power task. According to the method, after the computing power task request of the user is received, the target time period with the lowest computing power resource cost is determined according to the computing power resource information to process the target computing power task, so that the user is automatically helped to allocate the time period with the lowest computing power resource cost; the problem of serious waste of computing power resources is greatly reduced, the economic cost of a user is greatly reduced, and wide application of the general computing power is facilitated.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent computing centers, smart computing centers and computing power infrastructure, and specifically to a computing power peak-shifting scheduling method and device for a universal computing power intelligent computing center. Background Art

[0002] With the rapid development of artificial intelligence technology, "intelligent computing centers" and "intelligent computing centers" have come into being.

[0003] "Intelligent computing center" refers to a facility that provides the required computing power, data and algorithms for artificial intelligence applications (such as artificial intelligence deep learning model development, model training and model reasoning) by using large-scale heterogeneous computing resources, including general computing power and intelligent computing power. The intelligent computing center covers facilities, hardware, and software, and can provide full-stack capabilities from bottom-level computing power to top-level application enablement.

[0004] “Intelligent Computing Center” includes but is not limited to “Smart Computing Center”.

[0005] "Intelligent Computing Center" refers to an artificial intelligence computing center. It is a type of computing infrastructure that is based on artificial intelligence theory, adopts artificial intelligence computing architecture, and provides computing power services, data services, and algorithm services required for artificial intelligence applications.

[0006] "Computing power" is the core of "intelligent computing center" and "intelligent computing center". It is the ability of computer equipment or computing / data center to process information. It is the ability of computer hardware and software to work together to execute certain computing needs. It is the computing power to process information data and achieve target result output. It is a new type of productivity that integrates information computing power, network carrying capacity, and data storage capacity. It mainly provides services to society through computing power infrastructure.

[0007] At present, intelligent computing centers can directly provide computing resources that meet user needs, and users can use computing resources to develop models and other tasks. However, intelligent computing centers have the problem of large fluctuations in computing resource utilization, such as peak computing demand during the day leading to resource premiums and serious waste of idle computing resources at night.

[0008] It can be seen that the existing technology has difficulties in adjusting the utilization rate of computing resources in intelligent computing centers, resulting in serious waste of computing resources and a substantial increase in economic costs, which is not conducive to the widespread application of inclusive computing power. Summary of the invention

[0009] The present invention provides a computing power peak-shifting scheduling method and device for a universal computing power intelligent computing center, which solves the problem in the prior art that the intelligent computing center has difficulty in adjusting the computing power resource utilization rate, resulting in serious waste of computing power resources and a substantial increase in economic costs.

[0010] To solve the above problems, the present invention is achieved as follows: In a first aspect, the present invention provides a computing power peak-shifting scheduling method for a universal computing power intelligent computing center, the method comprising: Step S1, receiving a computing task request, the computing task request is used to request processing of a target computing task, the computing task request includes a task period and a task code, the task code is the running code of the target computing task, and the task period is the time period for processing the target computing task; Step S2: Obtain computing power resource information corresponding to the task cycle, wherein the computing power resource information includes the computing power resource cost corresponding to each time period in a plurality of time periods included in the task cycle; Step S3: determining a target time period according to the computing power resource information, wherein the duration corresponding to the target time period is less than the duration corresponding to the task cycle, and the target time period is the time period in the task cycle with the lowest corresponding computing power resource cost; Step S4: Run the task code in the target time period to process the target computing power task.

[0011] Optionally, the step S2, obtaining computing resource information corresponding to the task cycle, includes: Step S21, determining N time units corresponding to the task cycle, where N is a positive integer; Step S22: determine N tidal price curves based on the N time units, and the N tidal price curves correspond one-to-one to the N time units, wherein the computing power resource information includes the N tidal price curves, and the time unit includes three time periods. The tidal price curve is used to represent the correspondence between each time period in the corresponding time unit and the computing power resource cost, and the three time periods include a first time period, a second time period and a third time period. The computing power resource cost corresponding to the first time period is higher than the computing power resource cost corresponding to the second time period, and the computing power resource cost corresponding to the second time period is higher than the computing power resource cost corresponding to the third time period.

[0012] Optionally, the step S3, determining the target time period according to the computing power resource information, includes: Step S31: Calculate the processing time of the target computing task according to the computing task request to obtain the calculation time; Step S32: Based on the calculation duration and the N tidal price curves, determine the target time period, where the target time period is equal to the calculation duration.

[0013] Optionally, the step S32, determining the target time period based on the calculation duration and the N tidal price curves, includes: Step S321: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculated duration, determine the target time period within the N third time periods; Step S322: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is less than the calculation duration, and the sum of the durations of the N third time periods and the N second time periods is greater than or equal to the calculation duration, the N third time periods are determined as first sub-time periods, and a second sub-time period is determined within the N second time periods, and the target time period includes the first sub-time period and the second sub-time period; Step S323: When the sum of the durations of the N third time periods and the N second time periods corresponding to the N tidal price curves is less than the calculation duration, and the sum of the durations of the N third time periods, the N second time periods, and the N first time periods is greater than or equal to the calculation duration, the N third time periods are determined as third sub-time periods, the N second time periods are determined as fourth sub-time periods, and a fifth sub-time period is determined within the N first time periods, and the target time period includes the third sub-time period, the fourth sub-time period, and the fifth sub-time period.

[0014] Optionally, the step S321, when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculated duration, determining the target time period within the N third time periods, includes: Step S3211, determining the target computing resources required for the target computing task; Step S3212: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculation duration, determine a fourth time period within the N third time periods, the fourth time period being a time period in which the computing power resources are greater than the target computing power resources; Step S3213: determine the target time period within the fourth time period.

[0015] Optionally, the step S4, running the task code in the target time period to process the target computing power task, includes: Step S41, sorting multiple target sub-time periods in the target time period to obtain sorting information; Step S42: Divide the target computing power task into a plurality of sub-target computing power tasks, wherein the plurality of target sub-time periods correspond one to one with the plurality of sub-target computing power tasks; Step S43: processing the multiple sub-target computing power tasks based on the sorting information and the task code, wherein the first sub-target computing power task among the multiple sub-target computing power tasks is processed, including: Step S431: Process the corresponding sub-target computing task based on the task code to obtain the first target checkpoint; Processing the i-th sub-target computing power task among the multiple sub-target computing power tasks includes: Step S432: Process the i-th sub-target computing power task based on the i-1-th target checkpoint and the task code to generate the i-th target checkpoint, where i is an integer greater than 1, and the target checkpoint is used to save the processing result of the corresponding sub-target computing power task.

[0016] In a second aspect, the present invention further provides a computing power peak-shifting scheduling device for a universal computing power intelligent computing center, the device comprising: A receiving module, used to receive a computing task request, wherein the computing task request is used to request processing of a target computing task, and the computing task request includes a task period and a task code, wherein the task code is an operating code of the target computing task, and the task period is a time period for processing the target computing task; An acquisition module, used to acquire computing power resource information corresponding to the task cycle, wherein the computing power resource information includes a computing power resource cost corresponding to each time period in a plurality of time periods included in the task cycle; A determination module, configured to determine a target time period according to the computing power resource information, wherein the duration corresponding to the target time period is less than the duration corresponding to the task cycle, and the target time period is the time period in the task cycle with the lowest corresponding computing power resource cost; A processing module is used to run the task code in the target time period to process the target computing power task.

[0017] In a third aspect, the present invention further provides an electronic device comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method described in the first aspect above.

[0018] In a fourth aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps in the method described in the first aspect are implemented.

[0019] In a fifth aspect, the present invention further provides a computer program product, comprising computer instructions, which, when executed by a processor, implement the steps in the method described in the first aspect above.

[0020] The present invention provides a computing power peak-shifting scheduling method and device for an inclusive computing power intelligent computing center, the method comprising: step S1, receiving a computing power task request; step S2, obtaining computing power resource information corresponding to the task cycle; step S3, determining a target time period according to the computing power resource information; step S4, running the task code in the target time period to process the target computing power task. After receiving a computing power task request from a user, the present invention determines the computing power resource information within the cycle according to the task cycle corresponding to the computing power task, thereby determining the target time period with the lowest computing power resource cost according to the computing power resource information to process the target computing power task, thereby automatically helping users to allocate the time period with the lowest computing power resource cost, greatly reducing the problem of serious waste of computing power resources and greatly reducing the economic cost of users, which is conducive to the wide application of inclusive computing power. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solution of the present invention, the accompanying drawings required for use in the description of the present invention will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0022] Figure 1 It is a flow chart of a computing power peak-shifting scheduling method for a universal computing power intelligent computing center provided by the present invention; Figure 2 It is a schematic diagram of the computing resource cost corresponding to the time period provided by the present invention; Figure 3 It is a structural diagram of a computing power peak-shifting scheduling device for a universal computing power intelligent computing center provided by the present invention; Figure 4 It is a structural diagram of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0023] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings 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.

[0024] The "computing power" mentioned in the present invention refers to: the ability of computer equipment or computing / data center to process information, the ability of computer hardware and software to work together to execute certain computing requirements, and the computing power to achieve target result output by processing information data. It is a new type of productivity that integrates information computing power, network carrying capacity, and data storage capacity, and it mainly provides services to the society through computing power infrastructure.

[0025] The "computing power" (Computational Power, CP) mentioned in the present invention refers to: it is the ability of a data center server to process data and output results. It is a comprehensive indicator to measure the computing power of a data center, including general computing power, super computing power and intelligent computing power. The commonly used unit of measurement is the number of floating-point operations performed per second (FLOPS, 1EFLOPS=10^18FLOPS). The larger the value, the stronger the comprehensive computing power. According to calculations, 1 EFLOPS is approximately the computing power output of 5 Tianhe-2A or 500,000 mainstream server CPUs or 2 million mainstream notebooks. The calculation formula is: CP=CP 通用 +CP 智能 +CP 超级 .

[0026] The "carrying capacity" (Network Power, NP) mentioned in the present invention refers to: it is the performance of the data transmission capacity of the computing power facilities, including the comprehensive capabilities of network architecture, network bandwidth, transmission delay, intelligent management and scheduling, etc. It involves network transmission within and between data centers, and is a comprehensive indicator for measuring network transmission scheduling capabilities.

[0027] The "Storage Power" (SP) mentioned in the present invention refers to: the comprehensive capabilities of a data center in terms of data storage capacity, performance, safety and reliability, and green and low-carbon. It is a comprehensive indicator for measuring the data storage capacity of a data center, including external storage devices such as storage arrays and server built-in storage devices. The commonly used unit of measurement for storage capacity is exabyte (EB, 1EB=2^60bytes), and the commonly used unit of measurement for performance is the number of reads and writes per second per unit capacity (IOPS / TB, Input / Output Operations Per Second / TB). The disaster recovery ratio is an important manifestation of safety and reliability.

[0028] The "computing power infrastructure" mentioned in the present invention refers to: a new type of information infrastructure that integrates information computing power, network carrying capacity, and data storage capacity, which can realize centralized computing, storage, transmission and application of information.

[0029] The "new information infrastructure" mentioned in the present invention refers to: mainly including network infrastructure such as 5G networks, fiber-optic broadband networks, backbone networks, international communication networks, satellite Internet, computing power infrastructure such as data centers, general computing power centers, intelligent computing centers, supercomputing centers, and new technology facilities such as artificial intelligence, blockchain, and quantum computing.

[0030] The “computing power” mentioned in the present invention includes: general computing power, intelligent computing power and super computing power.

[0031] The “general computing power” mentioned in the present invention refers to: the computing power provided by servers based on CPU (Central Processing Unit) chips, which is used to support basic general computing such as cloud computing and edge computing.

[0032] The "intelligent computing power" mentioned in the present invention refers to: a computing platform based on GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit) and other dedicated chips for various innovative artificial intelligence applications, such as natural language processing, machine vision, etc.

[0033] The "super computing power" mentioned in the present invention refers to: the computing power mainly provided by high-performance computing clusters such as supercomputers. It uses the centralized computing resources of multiple computer systems working in parallel and uses a dedicated operating system to handle extremely complex or data-intensive problems. It is mainly used for calculations in cutting-edge scientific fields, such as planetary simulation, drug molecule design, gene analysis, etc.

[0034] The "intelligent computing center" mentioned in the present invention refers to a facility that provides the required computing power, data and algorithms for artificial intelligence applications (such as artificial intelligence deep learning model development, model training and model reasoning) by using large-scale heterogeneous computing resources, including general computing power (CPU) and intelligent computing power (GPU, FPGA, ASIC, etc.). The intelligent computing center covers facilities, hardware, and software, and can provide full-stack capabilities from bottom-level computing power to top-level application enablement.

[0035] The “intelligent computing center” mentioned in the present invention includes but is not limited to the “intelligent computing center”.

[0036] The "intelligent computing center" mentioned in the present invention is an artificial intelligence computing center, which is a type of computing power infrastructure that is based on artificial intelligence theory, adopts artificial intelligence computing architecture, and provides computing power services, data services, and algorithm services required for artificial intelligence applications.

[0037] The "computing power center" mentioned in the present invention refers to: a facility that is mainly composed of infrastructure such as wind, fire, water, electricity, and IT hardware and software equipment, and has computing power, transportation capacity, and storage capacity, including general data centers, intelligent computing centers, supercomputing centers, etc.

[0038] The "supercomputing center" mentioned in the present invention refers to: a supercomputing data center, which is a data center based on a supercomputer or a large-scale computing cluster, which can provide functions such as large-scale computing, storage and network services, and is widely used in application scenarios such as aerospace, national defense, oil exploration, climate modeling and genome sequencing.

[0039] The "computing resources" mentioned in the present invention refer to: technologies and facilities with information computing, transmission, storage and application capabilities required for the development of a digital society, including but not limited to computing resources such as CPUs and GPUs, network resources such as switches and routers, storage resources such as storage arrays and distributed storage, security resources such as firewalls and intrusion detection systems, and supporting and guarantee resources such as wind, fire, water and electricity.

[0040] The "inclusive computing power" mentioned in the present invention refers to providing appropriate and effective computing power services at an affordable cost to all social classes and groups that have computing power service needs based on the requirements of equal opportunity and the principle of commercial sustainability.

[0041] The “model” mentioned in the present invention includes but is not limited to a “large language model” and a “multimodal large model”.

[0042] The “large language model” mentioned in the present invention refers to a large language model (LLM), which is a language model with a large parameter scale, designed to understand and generate human language, trained with a large amount of text data, and can perform a wide range of tasks including text summarization, translation, sentiment analysis, etc.

[0043] The "Multimodal Large Models" mentioned in the present invention refer to models that combine multimodal information such as text, images, videos, audio, etc. for training, including but not limited to multimodal large language models.

[0044] See also Figure 1 , Figure 1 : is a flowchart of a computing power peak-shifting scheduling method for a universal computing power intelligent computing center provided by the present invention, such as Figure 1 As shown, the following steps are included: Step S1, receiving a computing power task request, wherein the computing power task request is used to request processing of a target computing power task, and the computing power task request includes a task cycle and a task code, wherein the task code is the running code of the target computing power task, and the task cycle is the time period for processing the target computing power task.

[0045] In the present invention, the intelligent computing center can receive the computing power task request of the user, and process the target computing power task requested by the user according to the computing power task request. Specifically, the computing power task request includes a task cycle and a task code, wherein the task cycle refers to the computing power cycle that the user needs to process the target computing power task, for example, 7 days, 10 days, etc., which can be adjusted according to the actual needs of the user. The task code is the task code for processing the target computing power task, which varies according to the specific application scenario and needs. For example, it can be a code written in Python, which is not specifically limited in the present invention.

[0046] Step S2: Obtain computing power resource information corresponding to the task cycle, wherein the computing power resource information includes the computing power resource cost corresponding to each time period in multiple time periods included in the task cycle.

[0047] In the present invention, the computing power resource information may be the computing power resource cost, such as the consumption cost of each time unit of the user in the intelligent computing center. Specifically, the computing power resource information includes the computing power resource cost corresponding to each time period in the multiple time periods included in the task cycle, wherein the task cycle is divided into multiple time periods, and illustratively, for example, if the task cycle is 2 days, then 2 days can be divided into 8 time periods, wherein the length of each time period is 6 hours, and the computing power resource cost corresponding to each time period is the same or different, and can be adjusted according to actual conditions.

[0048] like Figure 2 As shown, Figure 2 The figure is a schematic diagram of the computing resource cost corresponding to the time period in the embodiment of the present invention, wherein the computing resource price is low during the time period between 6:00-12:00, medium during the time period between 12:00-18:00 and 0:00-6:00, and high during the time period between 18:00-0:000. Figure 2 It can be seen that the time period with the lowest computing resource cost in a day is between 6:00 and 12:00. Since this period is rest time, there are fewer users in the intelligent computing center and less computing resources are used. Therefore, the cost consumption of users for processing computing resource tasks during this period is also lower.

[0049] Step S3: determine a target time period according to the computing power resource information, wherein the duration corresponding to the target time period is less than the duration corresponding to the task cycle, and the target time period is the time period in the task cycle with the lowest corresponding computing power resource cost.

[0050] In the present invention, a target time period is determined within the task cycle based on the determined computing power resource information. Specifically, the target time period is a time period composed of one or more time periods in the task cycle. For example, the target time period can be divided into multiple sub-target time periods, and the target computing power tasks are processed separately in different time periods. The computing power resource cost corresponding to each time period is the lowest, thereby minimizing the computing power resource cost of the overall processing of the target computing power tasks.

[0051] In some feasible implementations, the target time period can be determined by a large model or a deep neural network, that is, the target time period must be ensured to be the time period with the lowest corresponding computing resource cost in the task cycle.

[0052] It should be noted that, in the case where other users have already used computing resources in multiple time periods with the lowest computing resource costs, the target time period is determined by screening the time periods corresponding to higher computing resource costs according to the lowest principle, that is, the time period with the lowest computing resource cost considered as a whole within the task cycle. Through the above steps, users are allocated to process target computing tasks in relatively idle time periods with lower computing resource costs, which improves the computing resource utilization rate of the intelligent computing center and reduces the consumption of computing resource costs by users.

[0053] Step S4: Run the task code in the target time period to process the target computing power task.

[0054] In the present invention, the task code is run in a determined target time period to process the target computing power task, and finally after completing the processing of the target computing power task, the processing result is returned to the user before or at the end of the task cycle.

[0055] After receiving the computing power task request from the user, the present invention determines the computing power resource information within the cycle according to the task cycle corresponding to the computing power task, and thus determines the target time period with the lowest computing power resource cost to process the target computing power task according to the computing power resource information, thereby automatically helping the user to allocate the time period with the lowest computing power resource cost, greatly reducing the problem of serious waste of computing power resources and greatly reducing the economic cost of users, which is conducive to the widespread application of inclusive computing power.

[0056] In some feasible implementations, optionally, the step S2 of obtaining computing resource information corresponding to the task cycle includes: Step S21, determining N time units corresponding to the task cycle, where N is a positive integer; Step S22: determine N tidal price curves based on the N time units, and the N tidal price curves correspond one-to-one to the N time units, wherein the computing power resource information includes the N tidal price curves, and the time unit includes three time periods. The tidal price curve is used to represent the correspondence between each time period in the corresponding time unit and the computing power resource cost, and the three time periods include a first time period, a second time period and a third time period. The computing power resource cost corresponding to the first time period is higher than the computing power resource cost corresponding to the second time period, and the computing power resource cost corresponding to the second time period is higher than the computing power resource cost corresponding to the third time period.

[0057] In the present invention, the task cycle is divided into N time units, where N is a positive integer. For example, the task cycle can be 7 days, and the 7-day task cycle is divided into 7 time units, that is, each time unit corresponds to one day, wherein each time unit corresponds to a tidal price curve. It should be noted that the tidal price curve refers to a curve in which prices change over time in certain markets (such as electricity, finance or commodity markets), especially when considering short-term supply and demand changes and market dynamics. This curve is usually used to describe the price fluctuation pattern within a day or a period of time, similar to the rise and fall of the tide.

[0058] The computing power resource information includes the N tidal price curves, and each time unit includes three time periods, that is, each day is divided into three time periods, specifically, including the first time period, the second time period and the third time period. Among them, the computing power resource cost corresponding to the first time period is higher than the computing power resource cost corresponding to the second time period, and the computing power resource cost corresponding to the second time period is higher than the computing power resource cost corresponding to the third time period, that is, the cost consumption of the user in processing the target computing power task in the first time period is greater than the cost consumption in the second time period, and the cost consumption of the user in processing the target computing power task in the second time period is greater than the cost consumption in the third time period.

[0059] In the present invention, by determining N time units, the N time units are divided, and each time unit is split into three time periods, so that the time period with the lowest computing power resource cost can be determined among the N time units, that is, the third time period. When processing the user's target computing power task, the target computing power task can be preferentially arranged in the third time period for processing. For example, if the computing power resource arrangement of the third time period of a certain day is full, the target computing power task will be arranged in the second time period, thereby optimizing the user's consumption cost to the greatest extent.

[0060] Optionally, the step S3, determining the target time period according to the computing power resource information, includes: Step S31: Calculate the processing time of the target computing task according to the computing task request to obtain the calculation time; Step S32: Based on the calculation duration and the N tidal price curves, determine the target time period, where the target time period is equal to the calculation duration.

[0061] In the present invention, when the intelligent computing center receives a request for a computing task, the processing time for processing the target computing task can be calculated according to the task cycle and task code provided by the user to obtain the computing time. For example, when the task cycle is 7 days, the computing time can be 2 days, that is, in the 7 days within the task cycle, the time period with lower computing resource cost is selected to form a 2-day time period to process the computing task.

[0062] In some feasible implementations, the calculation time can be provided by the user when inputting a computing power task request. Specifically, the user makes a preliminary estimate of the processing time of the target computing power task, obtains a calculated value, and sends the calculated value to the intelligent computing center. The intelligent computing center can use the calculated value as a reference to further calculate the processing time of the target computing power task, thereby obtaining the calculation time. This calculation method is more accurate.

[0063] After determining the calculation duration, the length of the target time period is determined according to the calculation duration. For example, when the calculation duration is determined to be 2 days, the target time period is determined to be two days. Therefore, when the task cycle is 7 days, 7 tidal price curves can be determined, and a target time period of two days can be determined among the 7 tidal price curves. Specifically, multiple time periods with the lowest computing power resource cost can be selected from the 7 tidal price curves to form the target time period. The calculation duration can more accurately determine the duration range of the target time period, thereby accurately screening out the target time period from the N tidal price curves.

[0064] Optionally, the step S32, determining the target time period based on the calculation duration and the N tidal price curves, includes: Step S321: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculated duration, determine the target time period within the N third time periods; Step S322: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is less than the calculation duration, and the sum of the durations of the N third time periods and the N second time periods is greater than or equal to the calculation duration, the N third time periods are determined as first sub-time periods, and a second sub-time period is determined within the N second time periods, and the target time period includes the first sub-time period and the second sub-time period; Step S323: When the sum of the durations of the N third time periods and the N second time periods corresponding to the N tidal price curves is less than the calculation duration, and the sum of the durations of the N third time periods, the N second time periods, and the N first time periods is greater than or equal to the calculation duration, the N third time periods are determined as third sub-time periods, the N second time periods are determined as fourth sub-time periods, and a fifth sub-time period is determined within the N first time periods, and the target time period includes the third sub-time period, the fourth sub-time period, and the fifth sub-time period.

[0065] In the present invention, the computing power resource information includes the N tidal price curves, each time unit includes three time periods, and exemplarily, the time unit is the day, that is, the time of each day is divided into three time periods, specifically, including the first time period, the second time period and the third time period. Among them, the computing power resource cost corresponding to the first time period is higher than the computing power resource cost corresponding to the second time period, and the computing power resource cost corresponding to the second time period is higher than the computing power resource cost corresponding to the third time period, that is, the cost consumption of the user in processing the target computing power task in the first time period is greater than the cost consumption in the second time period, and the cost consumption of the user in processing the target computing power task in the second time period is greater than the cost consumption in the third time period.

[0066] In the present invention, N is 7 for illustration, which includes 7 tidal price curves, wherein, when the sum of the durations of the 7 third time periods is greater than or equal to the calculation duration, that is, the target time period can be completely allocated within the third time period with the lowest computing resource cost, the target time period can be determined within the 7 third time periods. It should be noted that the target time period can be split into multiple sub-target time periods within the 7 third time periods to form a target time period, and the specific division of the target time period can be adjusted according to the load situation and the actual queuing situation, which is not specifically limited in this embodiment.

[0067] When the sum of the durations of the 7 third time periods corresponding to the 7 tidal price curves is less than the calculation duration, and the sum of the durations of the 7 third time periods and the 7 second time periods is greater than or equal to the calculation duration, that is, the target time period cannot be completely allocated in the third time period, and another part of the target time period needs to be allocated in the 7 second time periods. Specifically, the 7 third time periods can be determined as the first sub-time periods, and the second sub-time periods can be determined in the 7 second time periods, so that the first sub-time period and the second sub-time period constitute the target time period.

[0068] When the sum of the durations of the 7 third time periods and the 7 second time periods corresponding to the 7 tidal price curves is less than the calculation duration, and the sum of the 7 third time periods, the 7 second time periods, and N first time periods is greater than or equal to the calculation duration, that is, the target time period cannot be completely allocated in the third time period and the second time period, and another part of the target time period needs to be allocated in the 7 first time periods. Specifically, the 7 third time periods can be determined as the third sub-time period, the 7 second time periods can be determined as the fourth sub-time period, and the fifth sub-time period can be determined in the 7 first time periods, so that the third sub-time period, the fourth sub-time period and the fifth sub-time period are determined as the target time period.

[0069] Through the method of determining the target time period in this embodiment, the target time period can be better allocated to the third time period with lower computing power resource cost. When the third time period cannot meet the allocated time, the second time period and the first time period are considered in turn, thereby reducing the user's computing power resource cost consumption as much as possible.

[0070] Optionally, the step S321, when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculated duration, determining the target time period within the N third time periods, includes: Step S3211, determining the target computing resources required for the target computing task; Step S3212: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculation duration, determine a fourth time period within the N third time periods, the fourth time period being a time period in which the computing power resources are greater than the target computing power resources; Step S3213: determine the target time period within the fourth time period.

[0071] In the present invention, when the sum of the durations of the N third time periods is greater than or equal to the computing duration, the method for determining the target time period may further include determining the target computing power resources corresponding to the target computing power task, wherein the target computing power resources are the computing power resources required to process the target computing power task, such as computing resources, storage resources, and network bandwidth, etc. The target computing power resources may be filled in by the user when the user sends a computing power task request to the intelligent computing center, or the intelligent computing center may analyze the target computing power task after receiving the computing power task request to determine the target computing power resources required for the target computing power task.

[0072] After the target computing power resources are determined, the fourth time period is determined by matching the target computing power resources in the third time period. It should be noted that the computing power resources may fluctuate in the third time period, that is, the computing power resources in some time periods are less and the computing power resources in some time periods are more.

[0073] Specifically, the computing power resources corresponding to the fourth time period are greater than the target computing power resources, which ensures the normal processing of the target computing power task. When the computing power resources are less than the target computing power resources, the processing of the target computing power task is suspended. Thus, a target time period is determined within the fourth time period, so that the task code is run in the target time period to process the target computing power task.

[0074] Optionally, the step S4, running the task code in the target time period to process the target computing power task, includes: Step S41, sorting multiple target sub-time periods in the target time period to obtain sorting information; Step S42: Divide the target computing power task into a plurality of sub-target computing power tasks, wherein the plurality of target sub-time periods correspond one to one with the plurality of sub-target computing power tasks; Step S43: processing the multiple sub-target computing power tasks based on the sorting information and the task code, wherein the first sub-target computing power task among the multiple sub-target computing power tasks is processed, including: Step S431: Process the corresponding sub-target computing task based on the task code to obtain the first target checkpoint; Processing the i-th sub-target computing power task among the multiple sub-target computing power tasks includes: Step S432: Process the i-th sub-target computing power task based on the i-1-th target checkpoint and the task code to generate the i-th target checkpoint, where i is an integer greater than 1, and the target checkpoint is used to save the processing result of the corresponding sub-target computing power task.

[0075] In the present invention, the target time period includes multiple target sub-time periods, and the multiple target sub-time periods are sorted to determine the sorted sorting information. Correspondingly, the target computing task is divided into multiple sub-target computing tasks, wherein each target sub-time period corresponds to each sub-target computing task one by one.

[0076] In the process of processing multiple sub-target computing tasks according to the sorting information and task code, the first sub-target computing task is first processed based on the task code to generate the first target checkpoint, which is the save point (checkpoint) of the processing result corresponding to the first sub-target computing task. It should be noted that during the training of the model, the task model parameters can be saved as a checkpoint so that the next time the training is resumed, when the next lower period is reached, the checkpoint of the model is loaded into the GPU computing memory, and training continues until the next training time. When the i-th sub-target computing task is subsequently processed, the saved information and task code of the i-1 target checkpoints need to be used to continue processing the i-th sub-target computing task, where i is an integer greater than 1. After all sub-target computing tasks are processed, the processing of the target computing task is completed.

[0077] By using the target checkpoint, the training process of the model is saved during the processing of the target computing power task, thereby splitting the target computing power task into multiple sub-target computing power tasks. This allows multiple sub-target computing power tasks to be processed separately in multiple target sub-time periods, reducing the waste of computing power resources while reducing the economic cost of users processing target computing power tasks.

[0078] After receiving the computing power task request from the user, the present invention determines the computing power resource information within the cycle according to the task cycle corresponding to the computing power task, and thus determines the target time period with the lowest computing power resource cost to process the target computing power task according to the computing power resource information, thereby automatically helping the user to allocate the time period with the lowest computing power resource cost, greatly reducing the problem of serious waste of computing power resources and greatly reducing the economic cost of users, which is conducive to the widespread application of inclusive computing power.

[0079] See also Figure 3 , Figure 3 : is a structural diagram of a computing power peak-shifting scheduling device for a universal computing power intelligent computing center provided by the present invention, such as Figure 3 As shown, the computing power peak-shifting scheduling device 300 for the inclusive computing power intelligent computing center includes: The receiving module 310 is used to receive a computing task request, wherein the computing task request is used to request processing of a target computing task, and the computing task request includes a task period and a task code, wherein the task code is the running code of the target computing task, and the task period is the time period for processing the target computing task; An acquisition module 320 is used to acquire computing power resource information corresponding to the task cycle, wherein the computing power resource information includes a computing power resource cost corresponding to each time period in a plurality of time periods included in the task cycle; A determination module 330 is used to determine a target time period according to the computing power resource information, wherein the duration corresponding to the target time period is less than the duration corresponding to the task cycle, and the target time period is the time period in the task cycle with the lowest corresponding computing power resource cost; The processing module 340 is used to run the task code in the target time period to process the target computing power task.

[0080] Optionally, the acquisition module 320 includes: A first determination submodule is used to determine N time units corresponding to the task cycle, where N is a positive integer; The second determination submodule is used to determine N tidal price curves based on the N time units, and the N tidal price curves correspond to the N time units one by one, wherein the computing power resource information includes the N tidal price curves, and the time unit includes three time periods. The tidal price curve is used to represent the correspondence between each time period in the corresponding time unit and the computing power resource cost, and the three time periods include a first time period, a second time period and a third time period. The computing power resource cost corresponding to the first time period is higher than the computing power resource cost corresponding to the second time period, and the computing power resource cost corresponding to the second time period is higher than the computing power resource cost corresponding to the third time period.

[0081] Optionally, the determination module 330 includes: A calculation submodule, used to calculate the processing time of processing the target computing task according to the computing task request to obtain the calculation time; The third determination submodule is used to determine the target time period based on the calculation duration and the N tidal price curves, and the target time period is equal to the calculation duration.

[0082] Optionally, the third determining submodule includes: A first determining unit, configured to determine the target time period within the N third time periods when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculated duration; a second determining unit, configured to determine the N third time periods as first sub-time periods, and determine a second sub-time period within the N second time periods, when the sum of the durations of the N third time periods corresponding to the N tidal price curves is less than the calculation duration, and the sum of the durations of the N third time periods and the N second time periods is greater than or equal to the calculation duration, and the target time period includes the first sub-time period and the second sub-time period; A third determination unit is used to determine N third time periods as third sub-time periods, N second time periods as fourth sub-time periods, and a fifth sub-time period within the N first time periods, when the sum of the durations of the N third time periods and the N second time periods corresponding to the N tidal price curves is less than the calculation duration, and the sum of the durations of the N third time periods, the N second time periods, and the N first time periods is greater than or equal to the calculation duration, and the target time period includes the third sub-time period, the fourth sub-time period, and the fifth sub-time period.

[0083] Optionally, the first determining unit includes: A first determination subunit is used to determine the target computing power resources required for the target computing power task; A second determining subunit is configured to determine a fourth time period within the N third time periods when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculation duration, the fourth time period being a time period in which the computing power resources are greater than the target computing power resources; The third determining subunit is configured to determine the target time period within the fourth time period.

[0084] Optionally, the processing module 340 includes: A sorting submodule, used to sort multiple target sub-time periods in the target time period to obtain sorting information; A division submodule, used to divide the target computing power task into a plurality of sub-target computing power tasks, wherein the plurality of target sub-time periods correspond one to one with the plurality of sub-target computing power tasks; A processing submodule, configured to process the plurality of sub-target computing power tasks based on the sorting information and the task code, wherein the first sub-target computing power task among the plurality of sub-target computing power tasks is processed, comprising: A first processing unit, configured to process the corresponding sub-target computing task based on the task code to obtain a first target checkpoint; Processing the i-th sub-target computing power task among the multiple sub-target computing power tasks includes: The second processing unit is used to process the i-th sub-target computing power task based on the i-1-th target checkpoint and the task code to generate the i-th target checkpoint, wherein i is an integer greater than 1, and the target checkpoint is used to save the processing result of the corresponding sub-target computing power task.

[0085] After receiving the computing power task request from the user, the present invention determines the computing power resource information within the cycle according to the task cycle corresponding to the computing power task, and thus determines the target time period with the lowest computing power resource cost to process the target computing power task according to the computing power resource information, thereby automatically helping the user to allocate the time period with the lowest computing power resource cost, greatly reducing the problem of serious waste of computing power resources and greatly reducing the economic cost of users, which is conducive to the widespread application of inclusive computing power.

[0086] The present application also provides an electronic device. Figure 4 , the electronic device may include a processor 401, a memory 402, and a program 4021 stored in the memory 402 and executable on the processor 401.

[0087] When the program 4021 is executed by the processor 401, it can achieve Figure 1 Any step in the corresponding method embodiment: Step S1, receiving a computing task request, the computing task request is used to request processing of a target computing task, the computing task request includes a task period and a task code, the task code is the running code of the target computing task, and the task period is the time period for processing the target computing task; Step S2: Obtain computing power resource information corresponding to the task cycle, wherein the computing power resource information includes the computing power resource cost corresponding to each time period in a plurality of time periods included in the task cycle; Step S3: determining a target time period according to the computing power resource information, wherein the duration corresponding to the target time period is less than the duration corresponding to the task cycle, and the target time period is the time period in the task cycle with the lowest corresponding computing power resource cost; Step S4: Run the task code in the target time period to process the target computing power task.

[0088] Optionally, the step S2, obtaining computing resource information corresponding to the task cycle, includes: Step S21, determining N time units corresponding to the task cycle, where N is a positive integer; Step S22: determine N tidal price curves based on the N time units, and the N tidal price curves correspond one-to-one to the N time units, wherein the computing power resource information includes the N tidal price curves, and the time unit includes three time periods. The tidal price curve is used to represent the correspondence between each time period in the corresponding time unit and the computing power resource cost, and the three time periods include a first time period, a second time period and a third time period. The computing power resource cost corresponding to the first time period is higher than the computing power resource cost corresponding to the second time period, and the computing power resource cost corresponding to the second time period is higher than the computing power resource cost corresponding to the third time period.

[0089] Optionally, the step S3, determining the target time period according to the computing power resource information, includes: Step S31: Calculate the processing time of the target computing task according to the computing task request to obtain the calculation time; Step S32: Based on the calculation duration and the N tidal price curves, determine the target time period, where the target time period is equal to the calculation duration.

[0090] Optionally, the step S32, determining the target time period based on the calculation duration and the N tidal price curves, includes: Step S321: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculated duration, determine the target time period within the N third time periods; Step S322: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is less than the calculation duration, and the sum of the durations of the N third time periods and the N second time periods is greater than or equal to the calculation duration, the N third time periods are determined as first sub-time periods, and a second sub-time period is determined within the N second time periods, and the target time period includes the first sub-time period and the second sub-time period; Step S323: When the sum of the durations of the N third time periods and the N second time periods corresponding to the N tidal price curves is less than the calculation duration, and the sum of the durations of the N third time periods, the N second time periods, and the N first time periods is greater than or equal to the calculation duration, the N third time periods are determined as third sub-time periods, the N second time periods are determined as fourth sub-time periods, and a fifth sub-time period is determined within the N first time periods, and the target time period includes the third sub-time period, the fourth sub-time period, and the fifth sub-time period.

[0091] Optionally, the step S321, when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculated duration, determining the target time period within the N third time periods, includes: Step S3211, determining the target computing resources required for the target computing task; Step S3212: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculation duration, determine a fourth time period within the N third time periods, the fourth time period being a time period in which the computing power resources are greater than the target computing power resources; Step S3213: determine the target time period within the fourth time period.

[0092] Optionally, the step S4, running the task code in the target time period to process the target computing power task, includes: Step S41, sorting multiple target sub-time periods in the target time period to obtain sorting information; Step S42: Divide the target computing power task into a plurality of sub-target computing power tasks, wherein the plurality of target sub-time periods correspond one to one with the plurality of sub-target computing power tasks; Step S43: processing the multiple sub-target computing power tasks based on the sorting information and the task code, wherein the first sub-target computing power task among the multiple sub-target computing power tasks is processed, including: Step S431: Process the corresponding sub-target computing task based on the task code to obtain the first target checkpoint; Processing the i-th sub-target computing power task among the multiple sub-target computing power tasks includes: Step S432: Process the i-th sub-target computing power task based on the i-1-th target checkpoint and the task code to generate the i-th target checkpoint, where i is an integer greater than 1, and the target checkpoint is used to save the processing result of the corresponding sub-target computing power task.

[0093] After receiving the computing power task request from the user, the present invention determines the computing power resource information within the cycle according to the task cycle corresponding to the computing power task, and thus determines the target time period with the lowest computing power resource cost to process the target computing power task according to the computing power resource information, thereby automatically helping the user to allocate the time period with the lowest computing power resource cost, greatly reducing the problem of serious waste of computing power resources and greatly reducing the economic cost of users, which is conducive to the widespread application of inclusive computing power.

[0094] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, each process of the above-mentioned embodiment of the computing power peak-shifting scheduling method for the inclusive computing power intelligent computing center is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. Among them, the computer-readable storage medium is such as a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc.

[0095] The embodiment of the present application further provides a computer program product, which is stored in a storage medium. The computer program product is executed by at least one processor to implement the various processes of the above-mentioned embodiment of the computing power peak-shifting scheduling method for the inclusive computing power intelligent computing center, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0096] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0097] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0098] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

Claims

1. A computing power peak-shifting scheduling method for a universal computing power intelligent computing center, characterized in that: The method comprises: Step S1, receiving a computing task request, the computing task request is used to request processing of a target computing task, the computing task request includes a task period and a task code, the task code is the running code of the target computing task, and the task period is the time period for processing the target computing task; Step S2: Obtain computing power resource information corresponding to the task cycle, wherein the computing power resource information includes the computing power resource cost corresponding to each time period in a plurality of time periods included in the task cycle; Step S3: determining a target time period according to the computing power resource information, wherein the duration corresponding to the target time period is less than the duration corresponding to the task cycle, and the target time period is the time period in the task cycle with the lowest corresponding computing power resource cost; Step S4: Run the task code in the target time period to process the target computing power task.

2. The method according to claim 1, characterized in that The step S2, obtaining computing resource information corresponding to the task cycle, includes: Step S21, determining N time units corresponding to the task cycle, where N is a positive integer; Step S22: determine N tidal price curves based on the N time units, and the N tidal price curves correspond one-to-one to the N time units, wherein the computing power resource information includes the N tidal price curves, and the time unit includes three time periods. The tidal price curve is used to represent the correspondence between each time period in the corresponding time unit and the computing power resource cost, and the three time periods include a first time period, a second time period and a third time period. The computing power resource cost corresponding to the first time period is higher than the computing power resource cost corresponding to the second time period, and the computing power resource cost corresponding to the second time period is higher than the computing power resource cost corresponding to the third time period.

3. The method according to claim 2, characterized in that The step S3, determining the target time period according to the computing power resource information, includes: Step S31: Calculate the processing time of the target computing task according to the computing task request to obtain the calculation time; Step S32: Based on the calculation duration and the N tidal price curves, determine the target time period, where the target time period is equal to the calculation duration.

4. The method according to claim 3, characterized in that The step S32, determining the target time period based on the calculation duration and the N tidal price curves, comprises: Step S321: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculated duration, determine the target time period within the N third time periods; Step S322: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is less than the calculation duration, and the sum of the durations of the N third time periods and the N second time periods is greater than or equal to the calculation duration, the N third time periods are determined as first sub-time periods, and a second sub-time period is determined within the N second time periods, and the target time period includes the first sub-time period and the second sub-time period; Step S323: When the sum of the durations of the N third time periods and the N second time periods corresponding to the N tidal price curves is less than the calculation duration, and the sum of the durations of the N third time periods, the N second time periods, and the N first time periods is greater than or equal to the calculation duration, the N third time periods are determined as third sub-time periods, the N second time periods are determined as fourth sub-time periods, and a fifth sub-time period is determined within the N first time periods, and the target time period includes the third sub-time period, the fourth sub-time period, and the fifth sub-time period.

5. The method according to claim 4, characterized in that The step S321, when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculated duration, determining the target time period within the N third time periods, includes: Step S3211, determining the target computing resources required for the target computing task; Step S3212: when the sum of the durations of the N third time periods corresponding to the N tidal price curves is greater than or equal to the calculation duration, determine a fourth time period within the N third time periods, the fourth time period being a time period in which the computing power resources are greater than the target computing power resources; Step S3213: determine the target time period within the fourth time period.

6. The method according to any one of claims 1 to 5, characterized in that: The step S4, running the task code in the target time period to process the target computing power task, includes: Step S41, sorting multiple target sub-time periods in the target time period to obtain sorting information; Step S42: Divide the target computing power task into a plurality of sub-target computing power tasks, wherein the plurality of target sub-time periods correspond one to one with the plurality of sub-target computing power tasks; Step S43: processing the multiple sub-target computing power tasks based on the sorting information and the task code, wherein the first sub-target computing power task among the multiple sub-target computing power tasks is processed, including: Step S431: Process the corresponding sub-target computing task based on the task code to obtain the first target checkpoint; Processing the i-th sub-target computing power task among the multiple sub-target computing power tasks includes: Step S432: Process the i-th sub-target computing power task based on the i-1-th target checkpoint and the task code to generate the i-th target checkpoint, where i is an integer greater than 1, and the target checkpoint is used to save the processing result of the corresponding sub-target computing power task.

7. A computing power peak-shifting scheduling device for a universal computing power intelligent computing center, characterized in that: The device comprises: A receiving module, used to receive a computing task request, wherein the computing task request is used to request processing of a target computing task, and the computing task request includes a task period and a task code, wherein the task code is an operating code of the target computing task, and the task period is a time period for processing the target computing task; An acquisition module, used to acquire computing power resource information corresponding to the task cycle, wherein the computing power resource information includes a computing power resource cost corresponding to each time period in a plurality of time periods included in the task cycle; A determination module, configured to determine a target time period according to the computing power resource information, wherein the duration corresponding to the target time period is less than the duration corresponding to the task cycle, and the target time period is the time period in the task cycle with the lowest corresponding computing power resource cost; A processing module is used to run the task code in the target time period to process the target computing power task.

8. An electronic device, characterized in that: include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The method comprises computer instructions, which, when executed by a processor, implement the steps of the method according to any one of claims 1 to 6.

Images