Calculation network scheduling method, storage medium, equipment and computer program product based on Xinzhou intelligent calculation platform

Abstract

The invention discloses a network scheduling method, storage medium, equipment and computer program product based on an intelligent loam computing platform, and the method comprises the steps: transmitting collected operation data of each terminal device to a cloud through the intelligent loam computing platform, and obtaining network nodes passing from each terminal device to the cloud through a tracking route, determining cross nodes of the terminal devices in the operation data transmission process according to the IP addresses of the network nodes; idle computing power on the cross nodes is calculated, and if the computing power required by the operation data transmitted to the cross nodes for computing power calculation is smaller than the idle computing power of the cross nodes, the operation data is continuously transmitted to the cloud after computing power calculation; and otherwise, performing node contraction on the cross node and the neighbor node thereof, updating the cross node, if the idle computing power for updating the cross node does not meet the computing power requirement, not performing node contraction, introducing the cross node into the computing power space through a star large model, performing computing power calculation on the operation data transmitted to the cross node, and continuously transmitting the operation data to the cloud.

Description

Technical Field

[0001] The present invention relates to the field of computing network scheduling technology, and specifically, to a computing network scheduling method, storage medium, device and computer program product based on the Xiran intelligent computing platform. Background Art

[0002] With the acceleration of global digital transformation and the rapid development of artificial intelligence technology, the demand for computing power has shown explosive growth, and the scale of computing power is also continuing to expand. In the context of cloud computing, there is a tendency to build a centralized supercomputing pool to solve computing power problems. However, due to network latency and bandwidth limitations, the response speed of centralized processing will be slowed down. In addition, due to the huge amount of data involved in the task, a lot of energy will be consumed in the process of transmitting data to the supercomputing pool. Especially in long-distance transmission, due to problems such as signal attenuation and transmission delay, more energy is required to ensure data stability and reliability.

[0003] In order to solve the above problems, some studies have proposed computing network scheduling, which can achieve unified management and scheduling of computing resources and network resources, solve the problem of uneven distribution of computing resources, and achieve efficient resource utilization and task processing. However, in the process of computing network scheduling, due to the lack of management strategies for computing power nodes, computing power nodes will be abused during data transmission, resulting in unnecessary waste of resources and increased energy consumption. Summary of the invention

[0004] In response to the problems existing in the prior art, the present invention provides a computing network scheduling method, storage medium, device and computer program product based on the Xiran intelligent computing platform. By uniformly performing computing power calculations on cross-nodes of data transmission, it is possible to achieve coordinated scheduling and flexible use of diversified computing power resources and reduce energy consumption.

[0005] To achieve the above technical objectives, the present invention adopts the following technical solution: a computing network scheduling method based on the Xirang intelligent computing platform, specifically comprising the following steps:

[0006] Step S1: The collected operation data of each terminal device is transmitted to the cloud through the Xirang intelligent computing platform, the network nodes passed from each terminal device to the cloud are obtained by tracing the route, and the cross-nodes of each terminal device during the operation data transmission process are determined according to the IP address of the network node;

[0007] Step S2, calculating the idle computing power on the cross node, if the computing power required for computing the computing power of the running data transmitted to the cross node is less than the idle computing power of the cross node, the running data transmitted to the cross node is calculated for computing power and then continued to be transmitted to the cloud;

[0008] Step S3, otherwise, the cross node and its neighbor nodes are contracted, the cross node is updated, and the idle computing power of the updated cross node is calculated to see whether it meets the computing power required for transmitting the running data to the updated cross node for computing power calculation. If so, the running data transmitted to the updated cross node is calculated for computing power and then continued to be transmitted to the cloud;

[0009] Step S4: If not, the node is not shrunk, and the cross node is introduced into the computing power space through the Star Big Model. After the computing power of the operating data transmitted to the cross node is calculated, it continues to be transmitted to the cloud.

[0010] Furthermore, the calculation process of the idle computing power is:

[0011]

[0012] Among them, C br is the idle computing power, n is the number of logic operation chips, i is the index of n, f(a i ) is the mapping function of logical operation, α i is the mapping ratio coefficient of the i-th logic operation chip, q 1 (TOPS) is the redundant computing power of logical operations; m is the number of parallel computing chips, j is the index of m, f(b j ) is the mapping function for parallel computing, β j is the mapping ratio coefficient of the jth parallel computing chip, q 2 (FLOPS) is the redundant computing power of parallel computing; p is the number of neural network acceleration chips, k is the index of p, f(c k ) is the mapping function of neural network acceleration, γ k is the mapping ratio coefficient of the kth neural network acceleration chip, q 3 (FLOPS) is the redundant computing power for neural network acceleration.

[0013] Furthermore, the specific process of performing node contraction on the cross node and its neighboring nodes in step S3 is: calculating the node similarity between the cross node and each neighboring node, if the node similarity exceeds 60%, merging the cross node with the corresponding neighboring node, updating the cross node, and broadcasting the updated cross node.

[0014] Furthermore, the calculation process of the node similarity is:

[0015]

[0016] Among them, S ij represents the similarity between the intersection node i and the neighbor node j, T i represents all neighbor nodes of intersection node i, T jRepresents all neighbor nodes of neighbor node j.

[0017] Furthermore, the specific process of introducing the computing power space of the cross node through the Star Big Model in step S4 is: the IP address, idle computing power and running data of the cross node are transmitted to the cross node for computing power calculation, and the computing power required is input into the Star Big Model. The Star Big Model finds the manufacturer of the corresponding cross node according to the input IP address, and according to the computing power difference between the required computing power and the idle computing power, searches for computing power space greater than the computing power difference in the same manufacturer, and accesses the cross node through the API interface.

[0018] Furthermore, the present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the computing network scheduling method based on the Xirang intelligent computing platform.

[0019] Furthermore, the present invention also provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the computing network scheduling method based on the Xirang intelligent computing platform is implemented.

[0020] Furthermore, the present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the computing network scheduling method based on the Xirang intelligent computing platform.

[0021] Compared with the prior art, the present invention has the following beneficial effects: the present invention performs computing power calculation at the cross-node based on the computing network scheduling method of the Xirang intelligent computing platform, which can realize the coordinated scheduling and flexible use of diversified computing power resources, improve the overall computing power utilization efficiency and business flexibility, and at the same time, reduce the time for data to travel back and forth to the cloud, thereby reducing network latency, alleviating the computing pressure on the cloud, and enabling the cloud to process other tasks more efficiently. In addition, in order to make the idle computing power of the cross-node meet the computing power calculation of the running data, the present invention performs node contraction by calculating the node similarity, thereby merging similar nodes to reduce noise and errors, thereby improving the calculation accuracy and expanding the idle computing power of the cross-node, which can more effectively utilize the computing resources in the network node and reduce energy consumption; further, the present invention can also expand the computing power space of the cross-node through the Xingchen big model, which can reduce the dependence on hardware resources, thereby reducing the procurement and maintenance costs of hardware resources, and the Xingchen big model can better protect the security of the running data in the cross-node. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flow chart of the computing network scheduling method based on the Xirang intelligent computing platform of the present invention;

[0023] Figure 2This is a schematic diagram of introducing computing power space through the Star Big Model at the cross node. DETAILED DESCRIPTION

[0024] The technical solution of the present invention is further explained below in conjunction with the accompanying drawings.

[0025] like Figure 1 The flowchart of the network scheduling method based on the Xirang intelligent computing platform of the present invention is as follows:

[0026] Step S1: The collected operating data of each terminal device is transmitted to the cloud through the Xirang intelligent computing platform, the network nodes passed from each terminal device to the cloud are obtained by tracing the route, and the cross-nodes of each terminal device during the data transmission process are determined according to the IP address of the network node.

[0027] Step S2, calculate the idle computing power on the cross node. If the computing power required for transmitting the operating data to the cross node for computing power calculation is less than the idle computing power of the cross node, the operating data transmitted to the cross node will be calculated for computing power and then continued to be transmitted to the cloud. Computing power calculation is performed at the cross node, which can realize the coordinated scheduling and flexible use of diversified computing power resources, improve the overall computing power utilization efficiency and business flexibility, and at the same time, reduce the time for operating data to travel to and from the cloud, thereby reducing network latency, alleviating the computing pressure on the cloud, and enabling the cloud to process other tasks more efficiently.

[0028] The calculation process of the idle computing power in the present invention is:

[0029]

[0030] Among them, C br is the idle computing power, n is the number of logic operation chips, i is the index of n, f(a i ) is the mapping function of logical operation, α i is the mapping ratio coefficient of the i-th logic operation chip, q 1 (TOPS) is the redundant computing power of logical operations; m is the number of parallel computing chips, j is the index of m, f(b j ) is the mapping function for parallel computing, β j is the mapping ratio coefficient of the jth parallel computing chip, q 2 (FLOPS) is the redundant computing power of parallel computing; p is the number of neural network acceleration chips, k is the index of p, f(c k ) is the mapping function of neural network acceleration, γ k is the mapping ratio coefficient of the kth neural network acceleration chip, q 3 (FLOPS) is the redundant computing power for neural network acceleration.

[0031] Step S3, otherwise, the cross node and its neighbor nodes are contracted and the cross node is updated. Specifically, the cross node and each neighbor node are calculated for node similarity. If the node similarity exceeds 60%, the cross node is merged with the corresponding neighbor node, the cross node is updated, and the updated cross node is broadcasted. It is calculated whether the idle computing power of the updated cross node meets the computing power required for the operation data to be transmitted to the updated cross node for computing power calculation. If so, the operation data transmitted to the updated cross node is calculated for computing power and then continued to be transmitted to the cloud. By calculating node similarity and contracting nodes, similar nodes are merged to reduce noise and errors, thereby improving calculation accuracy and expanding the idle computing power of cross nodes, which can more effectively utilize computing resources in network nodes and reduce energy consumption.

[0032] In the present invention, the node similarity is determined by calculating the degree of overlap between the neighbor nodes of the intersection node and the neighbor nodes of its neighbor nodes. The specific calculation process is:

[0033]

[0034] Among them, S ij represents the similarity between the intersection node i and the neighbor node j, T i represents all neighbor nodes of intersection node i, T j Represents all neighbor nodes of neighbor node j.

[0035] Step S4: If no, no node contraction is performed, and the cross node is introduced into the computing power space through the star big model, such as Figure 2 , the IP address, idle computing power and running data of the cross node are transmitted to the cross node for computing power calculation, and the computing power required is input into the Star Big Model. The Star Big Model finds the manufacturer of the corresponding cross node according to the input IP address, and according to the computing power difference between the required computing power and the idle computing power, searches for computing power space greater than the computing power difference in the same manufacturer, accesses the cross node through the API interface, calculates the computing power of the running data transmitted to the cross node, and then continues to transmit it to the cloud. By expanding the computing power space of the cross node through the Star Big Model, intelligent identification of the computing power space of the same manufacturer can be realized, resource waste can be avoided, and overall resource utilization can be improved. It can also reduce dependence on hardware resources, thereby reducing the procurement and maintenance costs of hardware resources. At the same time, the Star Big Model can better protect the security of running data in the cross node.

[0036] Through the computing network scheduling method based on the Xirang intelligent computing platform of the present invention, the operating data collected and transmitted to the cloud will be calculated at the cross-nodes of the transmission network, which can realize the coordinated scheduling and flexible use of diversified computing resources, improve the overall computing power utilization efficiency and business flexibility, and at the same time, reduce the time for data to travel to and from the cloud, thereby reducing network latency, alleviating the computing pressure on the cloud, and enabling the cloud to process other tasks more efficiently.

[0037] In one technical solution of the present invention, a computer-readable storage medium is also provided, storing a computer program, wherein the computer program enables a computer to execute the computing network scheduling method based on the Xirang intelligent computing platform.

[0038] In one technical solution of the present invention, an electronic device is also provided, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the computing network scheduling method based on the Xirang intelligent computing platform is implemented.

[0039] In one technical solution of the present invention, a computer program product is also provided, including a computer program, which, when executed by a processor, implements the computing network scheduling method based on the Xirang intelligent computing platform.

[0040] In the embodiments disclosed in the present application, the computer storage medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. The computer storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the above. More specific examples of computer storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0041] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0042] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A computing network scheduling method based on Xirang intelligent computing platform, characterized in that: The specific steps include: Step S1: The collected operation data of each terminal device is transmitted to the cloud through the Xirang intelligent computing platform, the network nodes passed from each terminal device to the cloud are obtained by tracing the route, and the cross-nodes of each terminal device during the operation data transmission process are determined according to the IP address of the network node; Step S2, calculating the idle computing power on the cross node, if the computing power required for computing the computing power of the running data transmitted to the cross node is less than the idle computing power of the cross node, the running data transmitted to the cross node is calculated for computing power and then continued to be transmitted to the cloud; Step S3, otherwise, the cross node and its neighbor nodes are contracted, the cross node is updated, and the idle computing power of the updated cross node is calculated to see whether it meets the computing power required for transmitting the running data to the updated cross node for computing power calculation. If so, the running data transmitted to the updated cross node is calculated for computing power and then continued to be transmitted to the cloud; Step S4: If not, the node is not shrunk, and the cross node is introduced into the computing power space through the Star Big Model. After the computing power of the operating data transmitted to the cross node is calculated, it continues to be transmitted to the cloud.

2. According to claim 1, a computing network scheduling method based on the Xirang intelligent computing platform is characterized in that: The calculation process of the idle computing power is as follows: Among them, C br is the idle computing power, n is the number of logic operation chips, i is the index of n, f(a i ) is the mapping function of logical operation, α i is the mapping ratio coefficient of the i-th logic operation chip, q1 (TOPS) is the redundant computing power of the logic operation; m is the number of parallel computing chips, j is the index of m, f(b j ) is the mapping function for parallel computing, β j is the mapping ratio coefficient of the jth parallel computing chip, q2 (FLOPS) is the redundant computing power of parallel computing; p is the number of neural network acceleration chips, k is the index of p, and f(c k ) is the mapping function of neural network acceleration, γ k is the mapping ratio coefficient of the kth neural network acceleration chip, and q3 (FLOPS) is the redundant computing power of neural network acceleration.

3. According to claim 1, a computing network scheduling method based on Xirang intelligent computing platform is characterized in that: The specific process of performing node contraction on the cross node and its neighbor nodes in step S3 is: calculating the node similarity between the cross node and each neighbor node. If the node similarity exceeds 60%, merging the cross node with the corresponding neighbor node, updating the cross node, and broadcasting the updated cross node.

4. According to claim 3, a computing network scheduling method based on the Xirang intelligent computing platform is characterized in that: The calculation process of the node similarity is: Among them, S ij represents the similarity between the intersection node i and the neighbor node j, T i represents all neighbor nodes of intersection node i, T j Represents all neighbor nodes of neighbor node j.

5. According to claim 1, a computing network scheduling method based on Xirang intelligent computing platform is characterized in that: The specific process of introducing the computing power space of the cross node through the Star Big Model in step S4 is: the IP address, idle computing power and running data of the cross node are transmitted to the cross node for computing power calculation, and the computing power required is input into the Star Big Model. The Star Big Model finds the manufacturer of the corresponding cross node according to the input IP address, and according to the computing power difference between the required computing power and the idle computing power, searches for computing power space greater than the computing power difference in the same manufacturer, and accesses the cross node through the API interface.

6. A computer-readable storage medium storing a computer program, characterized in that: The computer program enables the computer to execute the computing network scheduling method based on the Xirang intelligent computing platform as described in any one of claims 1-5.

7. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the computing network scheduling method based on the Xirang intelligent computing platform as described in any one of claims 1 to 5 is implemented.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by the processor, it implements the computing network scheduling method based on the Xirang intelligent computing platform as described in any one of claims 1-5.

Images