A multi-motherboard storage and computing system supporting automatic function configuration

Abstract

The present invention discloses a multi-motherboard storage and computing integrated system supporting automatic function configuration, including a system for configuring resources for the function requirements input by the user, and executing the business data of the corresponding function requirements after the configuration is completed. The system includes an automatic configuration unit, a storage unit, a complex computing unit and a preset number of general motherboards, and the storage unit includes a preset number of storage media. In this system, the user selects the corresponding demand category in the configuration interface according to the demand function, and inputs the corresponding demand parameters. The automatic configuration unit configures the resources of the corresponding general motherboard, storage medium, AI computing chip, edge computing chip and acquisition and recording chip according to the demand parameters, so as to realize the functions of different special systems such as data storage, AI computing, edge computing and acquisition and recording, so as to meet the user's needs for quickly realizing a variety of differentiated storage and computing functions when using the system.

Description

Technical Field

[0001] The present invention belongs to the field of storage computing technology, and in particular relates to a multi-mainboard storage and computing integrated system that supports automatic function configuration. Background Art

[0002] At present, in the field of modern storage computing technology, dedicated systems such as data storage, AI computing, edge computing, and collection and recording are playing an increasingly important role. Although there are differences in the functions of these dedicated types of systems, their hardware architecture itself has a certain degree of commonality, and differentiated functions can be achieved through different software and hardware configurations. However, in the prior art, the software and hardware configuration process of dedicated systems is complex and requires manual intervention, making it difficult to achieve efficient automated configuration, and each different functional requirement usually requires a different system hardware platform, which further increases the complexity of hardware procurement and management.

[0003] The prior art discloses the CN112948300A server, integrated storage and computing device, and server system. Although an integrated storage and computing architecture is realized, there are only two nodes, which can only realize two functions, computing and storage, and the functions are relatively simple. At the same time, the hardware and bus of the prior art are fixed, and the software and hardware configurations cannot be automatically modified to meet the ever-changing dynamic business needs. As a result, the prior art has the disadvantages of complex function expansion process, inflexible configuration, and high cost of expansion configuration, and is completely unable to fully adapt to diversified and differentiated business needs. Summary of the invention

[0004] The purpose of the present invention is to solve the problems raised in the background technology and to propose a multi-motherboard storage and computing integrated system that supports automatic function configuration.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] The present invention proposes a multi-mainboard storage-computing integrated system supporting automatic function configuration, including a system for configuring resources for function requirements input by a user, and executing business data corresponding to the function requirements after the configuration is completed. The system includes an automatic configuration unit, a storage unit, a complex computing unit, and a preset number of general mainboards. The storage unit includes a preset number of storage media, and the complex computing unit includes a preset number of AI computing chips, edge computing chips, and acquisition and recording chips, wherein:

[0007] During resource allocation:

[0008] The automated configuration unit is used to provide a configuration interface including demand categories such as data storage, AI computing, edge computing, and collection records. The user selects the corresponding demand category in the configuration interface according to the functional requirements, and enters the demand parameters corresponding to each demand category in the configuration interface;

[0009] It is also used to convert the demand parameters input by the user into structured data, and then match the demand parameters in the structured data according to the preset rule library, configure each general motherboard as the function of the corresponding demand category, and configure storage media, AI computing chips, edge computing chips or acquisition and recording chips for each demand category;

[0010] In the process of executing the business data of each functional requirement:

[0011] Each of the general mainboards is used to allocate business data of corresponding demand categories to corresponding storage media for storage or computing chips for computing according to configured resources;

[0012] Each of the storage media is used to receive the business data assigned by the general mainboard and store the data according to the configured resources;

[0013] Each computing chip is used to accept business data assigned by the general motherboard or business data directly input by the user and perform calculations or collection according to the configured resources;

[0014] The automatic configuration unit is also used to monitor the performance of each demand category in real time and make dynamic adjustments to resources.

[0015] Preferably, the system also includes a high-speed bus switching unit, which includes a high-speed bus connected to each universal mainboard, storage medium and each chip respectively, and the channel state of each high-speed bus is in an off state in an initial state until the automation configuration unit activates the channel of the corresponding high-speed bus according to the configuration.

[0016] Preferably, in the structured data, the data storage requirement parameters include: storage capacity, storage type, storage performance and data protection mode;

[0017] The AI ​​computing requirement parameters include: AI computing performance, memory size, AI required bandwidth and latency requirements;

[0018] The edge computing requirement parameters include: computing density, network latency, number of distributed computing nodes, and local storage cache performance;

[0019] The required parameters for the collection and recording include: collection data type, collection and recording interface and collection and recording rate.

[0020] Preferably, the AI ​​computing chip is an NPU chip, and the edge computing chip and the acquisition and recording chip are both FPGA chips;

[0021] Each of the storage media is a storage hard disk of a different type.

[0022] Preferably, the preset rule base defines different configurations of the channel states of the general mainboard, storage medium, AI computing chip, edge computing chip, acquisition and recording chip and high-speed bus corresponding to the demand parameters of each demand category in the structured data. Then, the automatic configuration unit matches the demand parameters in the structured data according to the preset rule base, configures each general mainboard as the function of the corresponding demand category, and configures the storage medium, AI computing chip, edge computing chip or acquisition and recording chip for each demand category, and operates in the following manner:

[0023] For the demand category of data storage, the automatic configuration unit configures the corresponding type of storage medium according to the storage type in the demand parameter, configures the corresponding number of general mainboards according to the storage performance in the demand parameter, selects and executes data storage software in the software library of the general mainboard, jointly configures the corresponding number of storage media according to the size of the storage capacity and the storage performance in the demand parameter, stores the data in the selected storage medium in the corresponding mode according to the data protection mode in the demand parameter, and sets the channel state of the high-speed bus connected to the configured general mainboard and the storage medium to the activated state;

[0024] For the demand category of AI computing, a general-purpose motherboard is selected as the function of the demand category of AI computing, and AI computing software is selected to be executed in the software library of the general-purpose motherboard, the automatic configuration unit configures a corresponding number of NPU chips according to the memory size and AI computing performance in the demand parameters, and performs computing power configuration on the NPU chip according to the AI ​​computing performance, adjusts the width of the high-speed bus connected to each general-purpose motherboard and chip of the configuration according to the AI ​​demand bandwidth in the demand parameters, and sets the channel state of the high-speed bus connected to each general-purpose motherboard and chip of the configuration to an activated state;

[0025] For the edge computing demand category, the automated configuration unit jointly configures a corresponding number of general-purpose motherboards and FPGA chips according to the computing density and the number of distributed computing nodes in the demand parameters, and the number of FPGA chips corresponds to the number of general-purpose motherboards, and selects edge computing software to be executed in the software library of the general-purpose motherboard, configures a corresponding number of storage media according to the local storage cache performance in the demand parameters, and sets the channel status of the high-speed bus connected to each configured general-purpose motherboard and storage medium to an activated state;

[0026] For the demand category of collection and recording, a universal mainboard is selected to serve as the function of the demand category of collection and recording, and the collection and recording software is selected to be executed in the software library of the universal mainboard. The automatic configuration unit configures a corresponding number of FPGA chips according to the number of types of the collection data type in the demand parameters, and calls the collection module of the corresponding collection data type on the FPGA chip according to the collection data type in the demand parameters, configures the interface of the FPGA chip according to the collection and recording interface in the demand parameters, configures a corresponding number of storage media according to the collection and recording rate in the demand parameters, adjusts the width of the high-speed bus connected to the configured universal mainboard, FPGA chip and storage medium according to the collection and recording rate in the demand parameters, and sets the channel status of the high-speed bus connected to the configured universal mainboard, FPGA chip and storage medium to the activated state.

[0027] Preferably, the system further comprises a network interface and a switching unit connected to each universal mainboard, the network interface and the switching unit are provided with an external interface for receiving data storage and business data of AI computing demand category input by a user, and the network interface and the switching unit are used to transmit the data storage and business data of AI computing demand category input by the user to the universal mainboard of the corresponding demand category, and are used for data interaction between the universal mainboards configured corresponding to the edge computing demand category, and for the automatic configuration unit to configure the universal mainboards;

[0028] Each of the universal mainboards is provided with a network port connected to the network interface and the switching unit, and the network bandwidth of each network port is initially set to a first preset value.

[0029] Preferably, the system further comprises a fiber optic acquisition interface arranged on the FPGA chip, and the fiber optic acquisition interface is used to receive edge computing input by a user and to collect and record business data of a required category.

[0030] Preferably, in the process of executing the business data of each functional requirement, the following operations are performed according to the configured resources:

[0031] For the data storage demand category, the universal mainboard used for data storage function allocates the received business data of the corresponding demand category to the corresponding configured storage medium for storage, and the storage medium stores the data according to the corresponding data protection mode;

[0032] For the demand categories of AI computing, the general motherboard used for AI computing functions distributes the received business data of the corresponding demand categories to the corresponding configured NPU chips for computing;

[0033] For edge computing demand categories, each configured FPGA chip will calculate the received business data of the corresponding demand category and forward it to the corresponding configured general motherboard. Each general motherboard will exchange data with the switching unit through the network interface, and then each general motherboard will forward the received business data to the corresponding configured storage medium for storage;

[0034] For the demand category of the collection record, the configured FPGA chip will parse the received business data of the corresponding demand category and forward it to the corresponding configured general motherboard. The general motherboard will then forward the received business data to the corresponding configured storage medium for storage.

[0035] Preferably, in the process of executing the business data of each functional requirement, the automatic configuration unit is also used to monitor the performance of each requirement category in real time and make dynamic adjustments to resources in the following manner:

[0036] For the data storage demand category, in the process of executing the business data corresponding to the functional requirements, the automated configuration unit monitors the storage performance in the demand parameters in real time and dynamically adjusts the number of storage media;

[0037] For the demand category of AI computing, during the execution of business data corresponding to the functional requirements, the automated configuration unit monitors the AI ​​computing performance and latency requirements in the demand parameters in real time, and dynamically adjusts the number of NPU chips and the width of the corresponding high-speed bus in turn;

[0038] For edge computing demand categories, during the execution of business data corresponding to functional requirements, the automated configuration unit monitors the network delay in the demand parameters in real time and dynamically adjusts the network bandwidth of the corresponding general motherboard Internet port;

[0039] For the demand categories of collection and recording, in the process of executing the business data of the corresponding functional requirements, the automated configuration unit monitors the collection and recording rate in real time and dynamically adjusts the bandwidth of the corresponding high-speed bus.

[0040] Preferably, the system further comprises a power management unit for supplying power to each general mainboard, the automatic configuration unit, the network interface and switching unit, each storage medium, each chip and the high-speed bus.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] In this system, the user selects the corresponding demand category in the configuration interface according to the demand function and enters the corresponding demand parameters. The automatic configuration unit configures the corresponding general motherboard, storage medium, AI computing chip, edge computing chip and acquisition and recording chip resources according to the demand parameters to realize the automatic configuration function and realize the functions of different special systems such as data storage, AI computing, edge computing and acquisition and recording, so as to meet the user's needs for quickly realizing a variety of differentiated storage and computing functions when using the system;

[0043] At the same time, in the process of executing the business data of various functional requirements, the system monitors the performance of each demand category in real time through the automated configuration unit and makes dynamic adjustments to resources, thereby making the configuration method more flexible and dynamically adapting to the ever-changing business data requirements, thereby improving the scalability and flexibility of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A module block diagram of a multi-motherboard storage-computing integrated system that supports automatic function configuration according to the present invention. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0046] like Figure 1 As shown, a multi-mainboard storage and computing integrated system supporting automatic function configuration is provided, which is used to configure resources according to the function requirements input by the user, and execute the business data corresponding to the function requirements after the configuration is completed. The system includes an automatic configuration unit, a storage unit, a complex computing unit and a preset number of general mainboards (the number of general mainboards is set according to actual needs, Figure 1 In the example, n represents the number of general motherboards), the storage unit includes a preset number of storage media, and the complex computing unit includes a preset number of AI computing chips, edge computing chips, and acquisition and recording chips, wherein: the AI ​​computing chip is an NPU chip, and the edge computing chip and the acquisition and recording chip are both FPGA chips (edge ​​computing and acquisition and recording are realized by different FPGA chips); each storage medium is a different type of storage hard disk (the type of storage hard disk can be SSD, HDD, Figure 1 SSD is used in both tables).

[0047] The system also includes a high-speed bus switching unit, which includes high-speed buses (such as Figure 1As shown, in this embodiment, the high-speed bus is a PCIE high-speed bus), and the channel state of each high-speed bus is in an off state in the initial state until the automatic configuration unit activates the channel of the corresponding high-speed bus according to the configuration.

[0048] During resource allocation:

[0049] The automated configuration unit is used to provide a configuration interface for demand categories including data storage, AI computing, edge computing, and collection records. Users select the corresponding demand category in the configuration interface according to functional requirements and enter the demand parameters corresponding to each demand category in the configuration interface;

[0050] It is also used to convert the demand parameters input by the user into structured data (text information can be converted into structured data of demand through natural language processing (NLP) technology), and then match the demand parameters in the structured data according to the preset rule library, configure each general motherboard as the function of the corresponding demand category, and configure storage media, AI computing chips, edge computing chips or acquisition and recording chips for each demand category; Figure 1 As shown, the automation configuration unit configures the storage medium, AI computing chip, edge computing chip, acquisition and recording chip and high-speed bus switching unit through I2C signals (configuring the high-speed bus switching unit includes configuring the channel state of the high-speed bus and the width of the high-speed bus).

[0051] Among them, in structured data, the data storage requirement parameters include: storage capacity, storage type, storage performance and data protection mode;

[0052] The required parameters for AI computing include: AI computing performance, memory size, AI required bandwidth, and latency requirements;

[0053] The required parameters for edge computing include: computing density, network latency, number of distributed computing nodes, and local storage cache performance;

[0054] The required parameters for collection and recording include: collection data type, collection and recording interface and collection and recording rate.

[0055] The preset rule base defines different configurations of the channel states of the general mainboard, storage medium, AI computing chip, edge computing chip, acquisition and recording chip, and high-speed bus corresponding to the demand parameters of each demand category in the structured data (that is, the channel states of the general mainboard, storage medium, AI computing chip, edge computing chip, acquisition and recording chip, and high-speed bus are configured according to the demand parameters of each demand category in the structured data, wherein the rule base is a resource configuration base formed based on existing experience, and the corresponding resources are configured according to the demand parameters). The automatic configuration unit matches the demand parameters in the structured data according to the preset rule base, configures each general mainboard as the function of the corresponding demand category, and configures the storage medium, AI computing chip, edge computing chip, or acquisition and recording chip for each demand category, and operates as follows:

[0056] For the data storage demand category, the automatic configuration unit configures the corresponding type of storage medium according to the storage type in the demand parameters, configures the corresponding number of general motherboards according to the storage performance in the demand parameters (for example, the rule base defines the number of general motherboards required for storage performance in different ranges), and selects the execution data storage software in the software library of the general motherboard (that is, the general motherboard is configured to have the data storage management function), and jointly configures the corresponding number of storage media according to the storage capacity and storage performance in the demand parameters (for example, the rule base defines the number of storage media required for storage performance in different ranges, and the number of storage media required for storage capacity in different ranges, and the final number of storage media depends on the storage capacity). The larger number of storage media is required according to the size of the storage capacity and the storage performance, such as two storage media are required according to the size of the storage capacity and three storage media are required according to the storage performance, and the number of storage media finally configured is three), the data is stored in the selected storage medium in the corresponding mode according to the data protection mode in the demand parameters, and the channel status of the high-speed bus connected to the configured general mainboard and the storage medium is set to the activated state; for example, for the demand category of data storage, the demand parameters in the structured data are storage capacity: 30TB, storage type: SSD storage, storage performance: bandwidth above 2GB / s, data protection mode: RAID5 (medium intensity), then 8 4TB NVME SSD hard disks are configured, the storage mode is stored according to the RAID5 mode, the number of general mainboards is one, and the channel status of each configured hard disk and the high-speed bus connected to the general mainboard is set to the activated state.

[0057] For the demand category of AI computing, a general-purpose motherboard is selected as the function of the demand category of AI computing (if the functional requirement for AI computing is defined in the rule base, a general-purpose motherboard is selected), and the execution of AI computing software is selected in the software library of the general-purpose motherboard (that is, the general-purpose motherboard is configured to have the management function of AI computing), and the automatic configuration unit configures the corresponding number of NPU chips according to the memory size and AI computing performance in the demand parameters (the rule base defines the number of NPU chips required for memory sizes within different ranges and the number of NPU chips required for AI computing performance within different ranges, and the final number of NPU chips depends on the memory size and AI computing performance. The larger number of NPU chips is required). And configure the computing power of the NPU chip according to the AI ​​computing performance, adjust the width of the high-speed bus connected to each general motherboard and chip of the configuration according to the AI ​​required bandwidth in the demand parameters (the rule base defines the width of the high-speed bus required for the AI ​​required bandwidth in different ranges), and set the channel state of the high-speed bus connected to each general motherboard and chip of the configuration to the activated state; for example, for the demand category of AI computing, the demand parameters in the structured data are AI computing performance: 50TFLOPS, memory size: 80GB, AI required bandwidth: 64Gb / s, then three NPU chips (such as NVIDIAA100) are configured, the width of the high-speed bus connected to each general motherboard and NPU chip is configured to PCIe4.0 X8, and the channel state of the high-speed bus is set to the activated state.

[0058] For the demand category of edge computing, the automatic configuration unit jointly configures the corresponding number of general-purpose motherboards and FPGA chips according to the computing density and the number of distributed computing nodes in the demand parameters (that is, the corresponding number of general-purpose motherboards are jointly configured according to the computing density and the number of distributed computing nodes in the demand parameters, and the corresponding number of FPGA chips are jointly configured according to the computing density and the number of distributed computing nodes in the demand parameters, where the computing density is divided into low, medium and high levels, and the number of general-purpose motherboards and FPGA chips required for different levels of computing density and the number of general-purpose motherboards and FPGA chips required for different numbers of distributed computing nodes are defined in the rule base. In general, the number of distributed computing nodes corresponds to the number of motherboards and FPGA chips), and the FPGA chip The number of chips corresponds to the number of general mainboards, and the edge computing software is selected to be executed in the software library of the general mainboard (that is, the general mainboard is configured to have the management function of edge computing), and the corresponding number of storage media is configured according to the local storage cache performance in the demand parameters (the number of storage media required for storage cache performance in different ranges is defined in the rule library), and the channel state of the high-speed bus connected to each general mainboard and storage medium of the configuration is set to the activated state; for example, for the demand category of edge computing, the demand parameters in the structured data are computing density: medium, number of distributed computing nodes: 3, local storage cache performance: 2GB / s performance, and 4TB capacity, then the number of general mainboards and FPGA chips are configured to be three, and one NVME SSD hard disk with a capacity of 4TB is configured, and the channel state of the high-speed bus connected to each general mainboard, FPGA chip and storage medium is set to the activated state.

[0059] For the demand category of collection and recording, a general mainboard is selected to be used as the function of the demand category of collection and recording (if the functional requirement for collection and recording is defined in the rule library, a general mainboard is selected), and the collection and recording software is selected to be executed in the software library of the general mainboard (that is, the general mainboard is configured to have the management function of collection and recording), the automatic configuration unit configures the corresponding number of FPGA chips according to the number of types of collection data types in the demand parameters (generally, one FPGA chip is configured for each type), and calls the collection module of the corresponding collection data type on the FPGA chip according to the collection data type in the demand parameters (that is, the collection function of the FPGA chip is configured according to the collection data type in the demand parameters), configures the interface of the FPGA chip according to the collection and recording interface in the demand parameters, configures the corresponding number of storage media according to the collection and recording rate in the demand parameters, adjusts the width of the high-speed bus connected to the configured general mainboard, FPGA chip and storage medium according to the collection and recording rate in the demand parameters, and sets the channel state of the high-speed bus connected to the configured general mainboard, FPGA chip and storage medium to the activated state; for example, for the demand category of collection and recording, the demand parameters in the structured data are collection data type: radar data, collection and recording interface: 6-way SRIO, collection and recording rate: 6 x10Gbps, configure a general motherboard, three 4TB NVME SSD hard drives and an FPGA chip, configure the interface of the FPGA chip as a 6-way 10Gb / s SRIO interface, configure the width of the high-speed bus connected on the general motherboard and the PFGA chip to PCIe4.0 X8, and set the channel state of the high-speed bus to activated state.

[0060] In the process of executing the business data of each functional requirement:

[0061] Each general-purpose mainboard is used to allocate the business data of the corresponding demand category to the corresponding storage medium for storage or computing chip for calculation according to the configured resources;

[0062] Each storage medium is used to receive the business data assigned by the general mainboard and store the data according to the configured resources;

[0063] Each computing chip is used to accept business data assigned by the general motherboard or business data directly input by the user and perform calculations or collection according to the configured resources;

[0064] The automated configuration unit is also used to monitor the performance of each demand category in real time and make dynamic adjustments to resources.

[0065] The system further includes a network interface and a switching unit connected to each general motherboard, and the network interface and the switching unit are provided with an external interface (the external interface is an Ethernet interface) for receiving business data of the data storage and AI computing demand categories input by the user, and the network interface and the switching unit are used to transmit the business data of the data storage and AI computing demand categories input by the user to the general motherboard of the corresponding demand category (it should be noted that in the process of executing business data, the user directly inputs the business data corresponding to the data storage and AI computing demand categories into the general motherboard of the corresponding demand category function through the external interface, the network interface and the switching unit), and is used for data interaction between the general motherboards corresponding to the edge computing demand category (because the edge computing configures the corresponding number of general motherboards according to the computing density and the number of distributed computing nodes, when the number of general motherboards is multiple, multiple general motherboards for edge computing demand categories exchange data through the network interface and the switching unit), and is used for the automatic configuration unit to configure each general motherboard (the automatic configuration unit configures each general motherboard through the network interface and the switching unit);

[0066] Each general mainboard is provided with a network port connected to the network interface and the switching unit (the network port is an Ethernet port, and each mainboard is connected to the network interface and the switching unit through the network port to receive business data), and the network bandwidth of each network port is initially set to a first preset value (the first preset value is set according to actual needs).

[0067] The system also includes a fiber optic acquisition interface arranged on the FPGA chip, which is used to receive edge computing input by the user and collect and record business data of the required category.

[0068] In the process of executing the business data of each functional requirement, perform the following operations according to the configured resources:

[0069] For the data storage requirement category, the universal mainboard used for data storage function allocates the received business data of the corresponding requirement category to the corresponding configured storage medium for storage, and the storage medium stores the data according to the corresponding data protection mode;

[0070] For the demand categories of AI computing, the general motherboard used for AI computing functions distributes the received business data of the corresponding demand categories to the corresponding configured NPU chips for computing;

[0071] For edge computing demand categories, each configured FPGA chip will calculate the received business data of the corresponding demand category and forward it to the corresponding configured general mainboard. Each general mainboard exchanges data with the switching unit through a network interface, and then each general mainboard forwards the received business data to the corresponding configured storage medium for storage (because edge computing configures a corresponding number of general mainboards according to the computing density and the number of distributed computing nodes, when there are multiple general mainboards, multiple general mainboards for edge computing demand categories exchange data with the switching unit through a network interface);

[0072] For the demand category of the collection and recording, the configured FPGA chip will perform frame parsing on the received business data of the corresponding demand category (for example, if the corresponding business data is radar video data, it will be parsed into frames of data), and then forwarded to the corresponding configured general mainboard. The general mainboard will then forward the received business data to the corresponding configured storage medium for storage.

[0073] Among them, in the process of executing the business data of each functional requirement, the automatic configuration unit is also used to monitor the performance of each requirement category in real time and make dynamic adjustments to resources (the dynamic adjustment in this embodiment is to add corresponding resources on the basis of current resources, one at a time, and the process of adding resources is also monitored in real time by the automatic configuration unit and dynamically adjusted), and operates as follows:

[0074] For the data storage demand category, in the process of executing the business data of the corresponding functional requirements, the automatic configuration unit monitors the storage performance in the demand parameters in real time and dynamically adjusts the number of storage media (when the storage performance is lower than the first threshold, the number of storage media is increased, and the specific value of the first threshold is set according to actual needs and is not limited);

[0075] For the demand category of AI computing, in the process of executing the business data of the corresponding functional requirements, the automatic configuration unit monitors the AI ​​computing performance and delay requirements in the demand parameters in real time (the delay requirement can be set according to actual needs, such as the delay requirement in this embodiment is not more than 100ms), and dynamically adjusts the number of NPU chips and the width of the corresponding high-speed bus in turn (when the AI ​​computing performance is lower than the second threshold, the number of NPU chips is increased, and when the delay requirement exceeds 100ms, the width of the high-speed bus is increased. The specific value of the second threshold is set according to actual needs and is not limited);

[0076] For edge computing demand categories, in the process of executing the business data of the corresponding functional requirements, the automated configuration unit monitors the network delay in the demand parameters in real time (the network delay can be set according to actual needs, such as the network delay in this embodiment is not more than 10ms), and dynamically adjusts the network bandwidth of the corresponding universal motherboard Internet port (when the network delay exceeds 10ms, the width of the network bandwidth of the universal motherboard Internet port is increased);

[0077] For the demand category of collection and recording, in the process of executing the business data of the corresponding functional requirements, the automated configuration unit monitors the collection and recording rate in real time and dynamically adjusts the bandwidth of the corresponding high-speed bus (when the monitored collection and recording rate is lower than the third threshold, the width of the high-speed bus is increased, and the specific value of the third threshold is set according to actual needs without restriction).

[0078] The system also includes a power management unit for supplying power to each general mainboard, an automatic configuration unit, a network interface and switching unit, each storage medium, each chip and a high-speed bus.

[0079] This system allows users to select the corresponding demand category in the configuration interface according to the demand function and enter the corresponding demand parameters. The automatic configuration unit configures the resources of the corresponding general motherboard, storage medium, AI computing chip, edge computing chip and acquisition and recording chip according to the demand parameters to realize the automatic configuration function and realize the functions of different special systems such as data storage, AI computing, edge computing and acquisition and recording to meet the user's rapid realization of various differentiated storage and computing function requirements when using the system; at the same time, in the process of executing the business data of various functional requirements, the system monitors the performance of each demand category in real time through the automatic configuration unit and makes dynamic adjustments to the resources, thereby making the configuration method more flexible and dynamically adapting to the ever-changing business data requirements, thereby improving the scalability and flexibility of the system.

[0080] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A multi-motherboard storage and computing integrated system supporting automatic function configuration, characterized in that: The system is used to configure resources for the functional requirements input by the user, and execute the business data corresponding to the functional requirements after the configuration is completed. The system includes an automated configuration unit, a storage unit, a complex computing unit and a preset number of general motherboards. The storage unit includes a preset number of storage media, and the complex computing unit includes a preset number of AI computing chips, edge computing chips and acquisition and recording chips, wherein: During resource allocation: The automated configuration unit is used to provide a configuration interface including demand categories such as data storage, AI computing, edge computing, and collection records. The user selects the corresponding demand category in the configuration interface according to the functional requirements, and enters the demand parameters corresponding to each demand category in the configuration interface; It is also used to convert the demand parameters input by the user into structured data, and then match the demand parameters in the structured data according to the preset rule library, configure each general motherboard as the function of the corresponding demand category, and configure storage media, AI computing chips, edge computing chips or acquisition and recording chips for each demand category; In the process of executing the business data of each functional requirement: Each of the general mainboards is used to allocate business data of corresponding demand categories to corresponding storage media for storage or computing chips for computing according to configured resources; Each of the storage media is used to receive the business data assigned by the general mainboard and store the data according to the configured resources; Each computing chip is used to accept business data assigned by the general motherboard or business data directly input by the user and perform calculations or collection according to the configured resources; The automatic configuration unit is also used to monitor the performance of each demand category in real time and make dynamic adjustments to resources; The system further comprises a high-speed bus switching unit, wherein the high-speed bus switching unit comprises high-speed buses respectively connected to each universal mainboard, storage medium and each chip, and the channel state of each high-speed bus is in an off state in an initial state until the automatic configuration unit activates the channel of the corresponding high-speed bus according to the configuration; The preset rule base defines different configurations of the channel states of the general motherboard, storage medium, AI computing chip, edge computing chip, acquisition and recording chip and high-speed bus corresponding to the demand parameters of each demand category in the structured data.

2. The multi-motherboard storage and computing integrated system supporting automatic function configuration as claimed in claim 1, characterized in that: In structured data, the data storage requirement parameters include: storage capacity, storage type, storage performance and data protection mode; The AI ​​computing requirement parameters include: AI computing performance, memory size, AI required bandwidth and latency requirements; The edge computing requirement parameters include: computing density, network latency, number of distributed computing nodes, and local storage cache performance; The required parameters for the collection and recording include: collection data type, collection and recording interface and collection and recording rate.

3. The multi-motherboard storage and computing integrated system supporting automatic function configuration as claimed in claim 2, characterized in that: The AI ​​computing chip is an NPU chip, and the edge computing chip and the acquisition and recording chip are both FPGA chips; Each of the storage media is a storage hard disk of a different type.

4. The multi-motherboard storage and computing integrated system supporting automatic function configuration as claimed in claim 3, characterized in that: The automatic configuration unit matches the requirement parameters in the structured data according to the preset rule base, configures each general mainboard as the function of the corresponding requirement category, and configures the storage medium, AI computing chip, edge computing chip or acquisition and recording chip for each requirement category, and operates in the following manner: For the demand category of data storage, the automatic configuration unit configures the corresponding type of storage medium according to the storage type in the demand parameter, configures the corresponding number of general mainboards according to the storage performance in the demand parameter, selects and executes data storage software in the software library of the general mainboard, jointly configures the corresponding number of storage media according to the size of the storage capacity and the storage performance in the demand parameter, stores the data in the selected storage medium in the corresponding mode according to the data protection mode in the demand parameter, and sets the channel state of the high-speed bus connected to the configured general mainboard and the storage medium to the activated state; For the demand category of AI computing, a general-purpose motherboard is selected as the function of the demand category of AI computing, and AI computing software is selected to be executed in the software library of the general-purpose motherboard, the automatic configuration unit configures a corresponding number of NPU chips according to the memory size and AI computing performance in the demand parameters, and performs computing power configuration on the NPU chip according to the AI ​​computing performance, adjusts the width of the high-speed bus connected to each general-purpose motherboard and chip of the configuration according to the AI ​​demand bandwidth in the demand parameters, and sets the channel state of the high-speed bus connected to each general-purpose motherboard and chip of the configuration to an activated state; For the edge computing demand category, the automated configuration unit jointly configures a corresponding number of general-purpose motherboards and FPGA chips according to the computing density and the number of distributed computing nodes in the demand parameters, and the number of FPGA chips corresponds to the number of general-purpose motherboards, and selects edge computing software to be executed in the software library of the general-purpose motherboard, configures a corresponding number of storage media according to the local storage cache performance in the demand parameters, and sets the channel status of the high-speed bus connected to each configured general-purpose motherboard and storage medium to an activated state; For the demand category of collection and recording, a universal mainboard is selected to serve as the function of the demand category of collection and recording, and the collection and recording software is selected to be executed in the software library of the universal mainboard. The automatic configuration unit configures a corresponding number of FPGA chips according to the number of types of the collection data type in the demand parameters, and calls the collection module of the corresponding collection data type on the FPGA chip according to the collection data type in the demand parameters, configures the interface of the FPGA chip according to the collection and recording interface in the demand parameters, configures a corresponding number of storage media according to the collection and recording rate in the demand parameters, adjusts the width of the high-speed bus connected to the configured universal mainboard, FPGA chip and storage medium according to the collection and recording rate in the demand parameters, and sets the channel status of the high-speed bus connected to the configured universal mainboard, FPGA chip and storage medium to the activated state.

5. The multi-motherboard storage and computing integrated system supporting automatic function configuration as claimed in claim 4, characterized in that: The system also includes a network interface and a switching unit connected to each universal mainboard, the network interface and the switching unit are provided with an external interface for receiving data storage and business data of AI computing demand categories input by a user, and the network interface and the switching unit are used to transmit the data storage and business data of AI computing demand categories input by the user to the universal mainboard of the corresponding demand category, and are used for data interaction between the universal mainboards configured corresponding to the edge computing demand category, and for the automatic configuration unit to configure the universal mainboards; Each of the universal mainboards is provided with a network port connected to the network interface and the switching unit, and the network bandwidth of each network port is initially set to a first preset value.

6. The multi-motherboard storage and computing integrated system supporting automatic function configuration as claimed in claim 5, characterized in that: The system also includes a fiber optic acquisition interface arranged on the FPGA chip, and the fiber optic acquisition interface is used to receive edge computing input by a user and collect and record business data of a required category.

7. The multi-motherboard storage and computing integrated system supporting automatic function configuration according to claim 6, characterized in that: In the process of executing the business data of each functional requirement, perform the following operations according to the configured resources: For the data storage demand category, the universal mainboard used for data storage function allocates the received business data of the corresponding demand category to the corresponding configured storage medium for storage, and the storage medium stores the data according to the corresponding data protection mode; For the demand categories of AI computing, the general motherboard used for AI computing functions distributes the received business data of the corresponding demand categories to the corresponding configured NPU chips for computing; For edge computing demand categories, each configured FPGA chip will calculate the received business data of the corresponding demand category and forward it to the corresponding configured general motherboard. Each general motherboard will exchange data with the switching unit through the network interface, and then each general motherboard will forward the received business data to the corresponding configured storage medium for storage; For the demand category of the collection record, the configured FPGA chip will parse the received business data of the corresponding demand category and forward it to the corresponding configured general motherboard. The general motherboard will then forward the received business data to the corresponding configured storage medium for storage.

8. The multi-motherboard storage-computing integrated system supporting automatic function configuration as claimed in claim 5, characterized in that: In the process of executing the business data of each functional requirement, the automatic configuration unit is also used to monitor the performance of each requirement category in real time and make dynamic adjustments to resources in the following manner: For the data storage demand category, in the process of executing the business data corresponding to the functional requirements, the automated configuration unit monitors the storage performance in the demand parameters in real time and dynamically adjusts the number of storage media; For the demand category of AI computing, during the execution of business data corresponding to the functional requirements, the automated configuration unit monitors the AI ​​computing performance and latency requirements in the demand parameters in real time, and dynamically adjusts the number of NPU chips and the width of the corresponding high-speed bus in turn; For edge computing demand categories, during the execution of business data corresponding to functional requirements, the automated configuration unit monitors the network delay in the demand parameters in real time and dynamically adjusts the network bandwidth of the corresponding general motherboard Internet port; For the demand categories of collection and recording, in the process of executing the business data of the corresponding functional requirements, the automated configuration unit monitors the collection and recording rate in real time and dynamically adjusts the bandwidth of the corresponding high-speed bus.

9. The multi-motherboard storage-computing integrated system supporting automatic function configuration according to claim 4, characterized in that: The system also includes a power management unit for supplying power to each universal mainboard, the automatic configuration unit, the network interface and switching unit, each storage medium, each chip and the high-speed bus.

Images