CN113608861B - Virtualized distribution method and device for software load computing resources - Google Patents

Abstract

The invention discloses a software-based load computing resource virtualization allocation method and device. The method includes: monitoring the status of pending tasks, the status of tasks being processed, and the usage status of heterogeneous computing resources of the space-borne computing platform; performing virtualization processing on the heterogeneous computing resources of the space-borne computing platform according to the status of pending tasks; According to the data and processing task requirements, the heterogeneous computing resources of the satellite computing platform are dynamically allocated or reallocated; the processing tasks are deployed according to the dynamic reallocation results of the hardware computing resources; the heterogeneous computing resource subscription and publishing clients are configured according to the processing task deployment method to achieve Interaction of pre- and post-processed data. The present invention proposes a virtualization method for controlling and dynamically reconfiguring multi-type heterogeneous processing such as field programmable gate arrays, digital signal processors and general-purpose processors, and solves the problem of efficient implementation of virtualization of spaceborne computing platforms.

Description

A software-based load computing resource virtualization allocation method and device

Technical field

The present invention relates to the technical field of satellite load computing resource virtualization, in particular to a computing resource virtualization allocation method and device.

Background technique

In complex mission environments, satellite payload systems are required to quickly adapt to complex dynamic environments and mission requirements, such as quickly switching between different functional waveforms to implement different processing processes. Traditionally, payloads designed with the hardware platform as the center and designed for special functions are becoming more and more popular. The more difficult it is to meet application requirements, how to build a software service-centered payload system so that the payload system has scalable hardware reconfiguration capabilities and virtualization-based software reconfiguration capabilities to meet the needs of current and future satellite computing mission expansion is issues that need resolving.

Virtualization is the configuration and management of hardware resources. Virtualization technology realizes the logical separation of service acquisition and the physical resources that actually provide services, making it the first and most important step to have the ability to define requirements. The satellite payload computing platform based on virtualization provides functional application services on the upper side to adapt to the deployment and execution of various functional tasks, and adopts a general, modular, standardized, and networked design on the lower side, making it scalable, upgradable, and evolvable. Ability. Compared with the common CPU computing resources of desktop systems, satellite payload systems are composed of a large number of heterogeneous processing resources, including FPGA resources that are good at logic processing, DSP resources that are good at floating point operations, and CPU resources that are good at real-time calculations. At present, the development of virtualization in desktop systems is relatively mature. How to realize resource virtualization in embedded systems with many heterogeneous processors is a key technology to support the next generation of satellite payload processing platforms. The development trend of space-borne computing platforms is hybrid heterogeneity. The present invention proposes a virtualization technology implementation method for space-borne heterogeneous loads to realize operating system-based multi-type heterogeneous processing of FPGAs, DSPs, and general-purpose processors. control and dynamic reconstruction to solve the problem of efficient implementation of virtualization of spaceborne computing platforms.

Contents of the invention

The technical problem solved by the present invention is to overcome the shortcomings of the existing technology, provide a computing resource virtualization allocation method and device, and solve the problem of flexible use of satellite-borne heterogeneous computing resources.

The technical solution of the present invention is:

The first aspect is a software-based load computing resource virtualization allocation method, including the following steps:

1) Monitor the task status of the spaceborne computing platform and the usage status of heterogeneous computing resources of the spaceborne computing platform, and obtain monitoring data; the task status includes: the status of tasks to be processed and the status of tasks being processed;

2) According to the task status, perform virtualization processing on the heterogeneous computing resources of the spaceborne computing platform and obtain the virtualization division results of the heterogeneous computing resources;

3) Based on the task status and heterogeneous computing resource usage status monitoring data obtained in step 1) and the heterogeneous computing resource virtualization division results obtained in step 2), dynamically allocate the heterogeneous computing resources of the satellite computing platform to obtain the satellite The allocation plan of heterogeneous computing resources on the load computing platform is used as the dynamic allocation result of hardware computing resources;

4) According to the dynamic allocation result of hardware computing resources described in step 3), deploy the tasks to be processed and the tasks being processed, and obtain the deployment method of processing tasks;

5) According to the deployment method of the processing task in step 4), configure the subscription publishing client on the heterogeneous processor to realize the interaction of pre-processing and post-processing data.

Optionally, the heterogeneous computing resources include: processing resources corresponding to field programmable gate arrays, digital signal processors, and general-purpose processors.

Optionally, the method for virtualizing the heterogeneous computing resources of the spaceborne computing platform described in step 2) includes: a field programmable gate array virtualization processing method, a digital signal processor virtualization processing method, and a general-purpose virtualization processing method. Processor virtualization processing methods;

The virtualization processing method of the field programmable gate array is: dividing the field programmable gate array logic resources into a high-speed communication client implementation area and multiple virtualization computing resource areas according to the task status. The virtualization computing The resource area communicates externally through the communication client;

The virtualization processing method of the digital signal processor is: according to the task status, the information is sent and received based on the special digital signal processor task, and the remaining virtualized computing resources are allocated as virtual resources.

The virtualization processing method of the general-purpose processor is to divide the virtualized computing resources based on the embedded operating system.

Optionally, the method described in step 1) for monitoring the task status of the spaceborne computing platform and the usage status of heterogeneous computing resources of the spaceborne computing platform includes:

Monitor and obtain the number of free area resources of each field programmable gate array FPGA and the data communication rate of the tasks being processed;

Monitor and obtain the total number of resources of each available digital signal processor and the data communication rate of the tasks being processed by the available digital signal processors;

Monitor and obtain the total number of resources of each available general-purpose processor and the data communication rate of the tasks being processed by the available general-purpose processor.

When there is a task to be processed, the delay requirements, computing requirements and data communication requirements of the task to be processed are obtained according to the task configuration file to be processed.

Optionally, the method for dynamically allocating heterogeneous computing resources on a spaceborne computing platform described in step 3) includes: a dynamic allocation method for general-purpose processors and digital signal processors, and a dynamic allocation method for field programmable gate arrays. method.

Optionally, the dynamic allocation method of the general processor and digital signal processor is specifically:

A31) Assume that the input rate of data to be processed by task i to be processed is Di bit/s, Ci execution cycles need to be processed, and the tolerated delay is Ti; the main frequency of processor j is Fj, the current usage rate is Pj, and the available communication bandwidth is Dj bit/s;

Find processors that satisfy the conditions of Dj>Di and Ci/(Fj*(1-Pj))<Ti among the general processors and digital signal processors of the spaceborne computing platform, and use (Dj-Di) to satisfy the conditions The processors are sorted, and the processor with the smallest (Dj-Di) value is used as the task execution processor; if there is no task execution processor that meets the conditions in the spaceborne computing platform processor, enter step A32), otherwise, Then go directly to step 4);

A32) Reallocate running tasks and pending tasks in general processors and digital signal processors. The reallocation method is to search for tasks that meet the task execution delay and communication bandwidth requirements based on the branch and bound method in order of task priority. Task combination, obtain the best match between processor and task combination, as the result of task reallocation; if all tasks are allocated, the allocation ends, and enters step 4), otherwise, enters step A33); the processor includes: General purpose processors and digital signal processors

A33) The unassigned tasks wait for other tasks to be executed and the system computing resources are released before executing step 4).

Optionally, the dynamic allocation method of the field programmable gate array is specifically:

B31) Assume that the input rate of data to be processed by task i to be processed is Di bit/s, the number of virtualized computing resources required is Ai, the number of resources in the virtualized computing resource area is Aj, and the available communication bandwidth is Dj bit/s;

Search the virtualized computing resource areas that meet the conditions of Aj>Ai and Dj>Di in the field programmable gate array of the computing platform, and sort the virtualized computing resource areas that meet the conditions by (Dj-Di), (Dj-Di ) The virtualized computing resource area with the smallest value is used as the task execution virtualized computing resource area; when there is no virtualized computing resource area that meets the conditions in the computing platform, enter step B32); otherwise, end the allocation and enter step 5);

B32) Reallocate the tasks being processed and the tasks to be processed in the virtualized computing resource area. The reallocation method is to set the size of the virtualized computing resource area according to the task requirements in order of task priority, by completely reconfiguring the field programmable gate array. To allocate, if all tasks have been allocated, end the allocation and go to step 5), otherwise, go to step B33);

B33) The unallocated task waits for other tasks to be completed and the system computing resources are released before executing step 5).

Optionally, the method for realizing pre-processing and post-processing data interaction described in step 5) is specifically:

51) Each processor on the satellite computing platform is interconnected through a high-speed switching network, and one of the general-purpose processors is selected as a communication management control node for data communication management between tasks on each processor;

52) Implement the subscription publishing system client on each satellite computing platform processor;

53) The subscription publishing client on each satellite computing platform processor creates subscribers and publishers based on the deployment method of processing tasks obtained in step 4) for data interaction between tasks on the processor.

Optionally, the high-speed switching network between processors includes: any one of a high-speed optical switching network, a high-speed RapidIO switching network, and/or an Ethernet switching network.

In a second aspect, a software-based load computing resource virtualization allocation device for implementing the method described in the first aspect includes:

System resource monitoring module: Responsible for monitoring system resources and task status, monitoring the status of pending tasks, status of tasks being processed, and usage status of heterogeneous computing resources on the space-borne computing platform, and obtaining monitoring data;

System task management module: Responsible for task configuration and runtime control, dynamically allocate heterogeneous computing resources of the space-borne computing platform based on the status of tasks to be processed and the status of tasks being processed, and obtain the allocation plan for the heterogeneous computing resources of the space-borne computing platform;

Task dynamic deployment module: Responsible for user task deployment, deploying pending tasks and ongoing tasks based on the dynamic allocation results of hardware computing resources;

Data subscription and publishing module: Responsible for communication between heterogeneous processor tasks. According to the deployment method of processing tasks, the subscription and publishing client is configured on the heterogeneous processor to realize the interaction of pre-processing and post-processing data.

The advantages of the present invention compared with the prior art are:

The computing resource virtualization allocation method and device provided by the embodiments of the present invention and the heterogeneous resource access method based on data subscription publishing provide task mobility and location transparency, so that calculations can be performed on the same processor and in a remote processor. There is no difference in processing, and it is suitable for multi-star collaborative processing. Based on interconnection, the hardware is used as a schedulable resource. Different from the traditional hardware acceleration method, the FPGA is completely decoupled from the general processor and directly connected to the data network as a computing resource. It is conducive to the decoupling of task processing, on-demand allocation according to task needs, and improves resource interconnection capabilities, processing throughput, computing resource utilization and system flexibility.

Description of the drawings

Figure 1 is a step flow chart of a software-based load computing resource virtualization allocation method provided by an embodiment of the present invention;

Figure 2 is a schematic structural diagram of a software-based load computing resource virtualization allocation device provided by an embodiment of the present invention.

Detailed ways

A software-based load computing resource virtualization allocation method of the present invention includes the following steps:

1) Monitor the task status of the spaceborne computing platform and the usage status of heterogeneous computing resources of the spaceborne computing platform, and obtain monitoring data; the task status includes: the status of tasks to be processed and the status of tasks being processed;

2) According to the task status, perform virtualization processing on the heterogeneous computing resources of the spaceborne computing platform and obtain the virtualization division results of the heterogeneous computing resources;

3) Based on the task status and heterogeneous computing resource usage status monitoring data obtained in step 1) and the heterogeneous computing resource virtualization division results obtained in step 2), dynamically allocate the heterogeneous computing resources of the satellite computing platform to obtain the satellite The allocation plan of heterogeneous computing resources on the load computing platform is used as the dynamic allocation result of hardware computing resources;

4) According to the dynamic allocation result of hardware computing resources described in step 3), deploy the tasks to be processed and the tasks being processed, and obtain the deployment method of processing tasks;

5) According to the deployment method of the processing task in step 4), configure the subscription publishing client on the heterogeneous processor to realize the interaction of pre-processing and post-processing data. specific:

51) Each processor on the satellite computing platform is interconnected through a high-speed switching network, and one of the general-purpose processors is selected as a communication management control node for data communication management between tasks on each processor;

52) Implement the subscription publishing system client on each satellite computing platform processor;

53) The subscription publishing client on each satellite computing platform processor creates subscribers and publishers based on the deployment method of processing tasks obtained in step 4) for data interaction between tasks on the processor.

Heterogeneous computing resources include: field programmable gate arrays, digital signal processors, and general-purpose processor processing resources.

Step 2) The method for virtualizing heterogeneous computing resources of a spaceborne computing platform includes: a virtualization processing method for a field programmable gate array, a virtualization processing method for a digital signal processor, and a virtualization processing method for a general-purpose processor. Chemical treatment methods;

21) The virtualization processing method of the field programmable gate array is to divide the field programmable gate array logical resources into a high-speed communication client implementation area and multiple virtualized computing resource areas according to the task status, and the virtualization The computing resource area communicates externally through the communication client;

22) The virtualization processing method of the digital signal processor is to send and receive information based on a dedicated digital signal processor task according to the task status, and the remaining computing resources are allocated as virtual resources.

23) The virtualization processing method of the general-purpose processor is to divide computing resources based on the embedded operating system.

The method described in step 1) for monitoring the mission status of the spaceborne computing platform and the usage status of heterogeneous computing resources of the spaceborne computing platform includes:

Monitor and obtain the number of free area resources of each field programmable gate array FPGA and the data communication rate of the tasks being processed;

Monitor and obtain the total number of resources of each available digital signal processor and the data communication rate of the tasks being processed by the available digital signal processors;

Monitor and obtain the total number of resources of each available general-purpose processor and the data communication rate of the tasks being processed by the available general-purpose processor.

When there is a task to be processed, the delay requirements, computing requirements and data communication requirements of the task to be processed are obtained according to the task configuration file to be processed.

Step 3) The method for dynamically allocating heterogeneous computing resources on a spaceborne computing platform includes: a dynamic allocation method for general-purpose processors and digital signal processors, and a dynamic allocation method for field programmable gate arrays;

The dynamic allocation method of the general processor and the digital signal processor is specifically:

A31) Assume that the input rate of data to be processed by task i to be processed is Di bit/s, Ci execution cycles need to be processed, and the tolerated delay is Ti; the main frequency of processor j is Fj, the current usage rate is Pj, and the available communication bandwidth is Dj bit/s;

Find processors that satisfy the conditions of Dj>Di and Ci/(Fj*(1-Pj))<Ti among the general processors and digital signal processors of the spaceborne computing platform, and use (Dj-Di) to satisfy the conditions The processors are sorted, and the processor with the smallest (Dj-Di) value is used as the task execution processor; if there is no task execution processor that meets the conditions in the spaceborne computing platform processor, enter step A32), otherwise, Then go directly to step 4);

A32) Reallocate running tasks and pending tasks in general processors and digital signal processors. The reallocation method is to search for tasks that meet the task execution delay and communication bandwidth requirements based on the branch and bound method in order of task priority. Task combination, obtain the best match between processor and task combination, as the result of task reallocation; if all tasks are allocated, the allocation ends, and enters step 4), otherwise, enters step A33); the processor includes: General purpose processors and digital signal processors

A33) The unassigned tasks wait for other tasks to be executed and the system computing resources are released before executing step 4).

The dynamic allocation method of the field programmable gate array is specifically:

B31) Assume that the input rate of data to be processed by task i to be processed is Di bit/s, the number of virtualized computing resources required is Ai, the number of resources in the virtualized computing resource area is Aj, and the available communication bandwidth is Dj bit/s;

Search the virtualized computing resource areas that meet the conditions of Aj>Ai and Dj>Di in the field programmable gate array of the computing platform, and sort the virtualized computing resource areas that meet the conditions by (Dj-Di), (Dj-Di ) The virtualized computing resource area with the smallest value is used as the task execution virtualized computing resource area; when there is no virtualized computing resource area that meets the conditions in the computing platform, enter step B32); otherwise, end the allocation and enter step 5);

B32) Reallocate the tasks being processed and the tasks to be processed in the virtualized computing resource area. The reallocation method is to set the size of the virtualized computing resource area according to the task requirements in order of task priority, by completely reconfiguring the field programmable gate array. To allocate, if all tasks have been allocated, end the allocation and go to step 5), otherwise, go to step B33);

B33) The unallocated task waits for other tasks to be completed and the system computing resources are released before executing step 5).

Methods for interacting with pre-processing and post-processing data in step 5) include:

Data is sent and received between processors based on a high-speed switching network based on a system architecture based on data subscription and publishing. A general-purpose processor is used as the main control node to implement node communication management. The remaining processors implement the subscription and publishing system client. The client is created on the main control node. Communication entity, the communication entity is stored by the master;

Communication entities include writing data entities and reading data; communication entities are used to implement data communication; each communication entity is interconnected through a high-speed switching network to realize heterogeneous interconnection data sharing; control data flow and business data flow use different themes to carry out data Send and receive.

The high-speed switching network between the processors includes:

Computing resources adopt a full switching method, which can be a high-speed optical switching network, a high-speed RapidIO switching network, or an Ethernet switching network.

A software-based load computing resource virtualization allocation device includes four functional modules:

System resource monitoring module: Responsible for monitoring system resources and task status, monitoring the status of pending tasks, status of tasks being processed, and usage status of heterogeneous computing resources on the space-borne computing platform, and obtaining monitoring data;

System task management module: Responsible for task configuration and runtime control, dynamically allocate heterogeneous computing resources of the space-borne computing platform based on the status of tasks to be processed and the status of tasks being processed, and obtain the allocation plan for the heterogeneous computing resources of the space-borne computing platform;

Task dynamic deployment module: Responsible for user task deployment, deploying pending tasks and ongoing tasks based on the dynamic allocation results of hardware computing resources;

Data subscription and publishing module: Responsible for communication between heterogeneous processor tasks. According to the deployment method of processing tasks, the subscription and publishing client is configured on the heterogeneous processor to realize the interaction of pre-processing and post-processing data.

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

Referring to Figure 1, there is shown a step flow chart of a software-based load computing resource virtualization allocation method provided by an embodiment of the present invention. As shown in Figure 1, the software-based load calculation resource virtualization method can be applied to satellites. In this embodiment, the satellite computing platform includes four boards. Board 1 includes 2 Power PC processors and 1 memory chip. Board 2 includes 4 FPGAs. Each FPGA is divided into 4 Reconfigurable area, board 3 includes 4 DSPs, each DSP includes 8 cores. The computing chips of each board realize full exchange and transmission of data through the Ethernet protocol.

As shown in Figure 1, the specific steps may include:

Step 101: Monitor the status of pending tasks, the status of tasks being processed, and the usage status of heterogeneous computing resources on the spaceborne computing platform;

When there is a task to be processed, the delay requirements, computing requirements and data communication requirements of the task to be processed are obtained according to the task configuration file to be processed;

In addition, system resource status information must be obtained, including:

The number of free area resources of each field programmable gate array and the data communication rate of the tasks being processed;

The total number of resources of each available digital signal processor and the data communication rate of the task being processed;

The total resources of each available general-purpose processor and the communication rate of the task being processed.

Step 102: Virtualize the heterogeneous computing resources of the spaceborne computing platform according to the status of the tasks to be processed;

The Power PC processor runs an embedded real-time operating system and serves as the system master control node to obtain task information and system status information, manage system computing tasks, and perform system computing resource allocation control.

The FPGA is divided into a high-speed communication client implementation area and multiple virtualized computing resource areas. The field programmable gate array is connected to other processors through high-speed interfaces. The virtualized computing resource area communicates externally through the high-speed communication client. According to the virtualization The communication interface is configured according to the communication requirements and task priority of execution tasks in the computing resource area to ensure the execution of high data requirements and high priority tasks. Each virtualized computing resource region has an address/data interface, an interrupt interface, and clock and reset signals. Achieve multiple tasks running at the same time.

In DSP, Core0 implements the subscription publishing client to realize the sending and receiving of data. The remaining resources are allocated as virtual resources. The communication interface is configured according to the communication requirements and task priority of the execution task to ensure high data requirements and high priority tasks. implement.

Step 103: Dynamically allocate or reallocate the heterogeneous computing resources of the spaceborne computing platform according to the monitoring data and processing task requirements;

Assume that the input rate of data to be processed by task 1 is 1M bit/s, 10M execution cycles need to be processed, and the tolerated delay is 50ms; the main frequency of the processor Power PC 1 is 1GHz, the current usage rate is 20%, and communication is available The bandwidth is 20M bit/s. The main frequency of the processor Power PC 2 is 1GHz, the current utilization rate is 30%, and the available communication bandwidth is 10Mbit/s.

Search the computing platform processors for processors that satisfy both Dj>Di and Ci/(Fj*(1-Pj))<Ti conditions, and sort the processors that meet the conditions by (Dj-Di), (Dj- The processor with the smallest Di) value is used as the task execution processor. The processor Power PC 2 is assigned to process task 1.

When there is no processor that meets the conditions in the computing platform processor, the processor's running tasks and to-be-processed tasks are reallocated. The reallocation method is to search for tasks that satisfy the task execution based on the branch and bound method in order of task priority. The task combination of delay and communication bandwidth requirements is used to obtain the best match between the processor and the task combination as a specific implementation method of task redistribution.

If all tasks are assigned to corresponding processing, the assignment ends. Otherwise, the unassigned tasks wait for other tasks to be completed, the system computing resources are released, and then step 103 is performed.

For tasks executed on field programmable gate arrays, assume that the input rate of data to be processed by task 2 is 1G bit/s, and the number of programmable logic blocks in the required virtualized computing resource area is 2300. Assume that the virtualized computing resource area 1. The number of programmable logic blocks is 5000, and the available communication bandwidth is 500M bit/s. Assume that the number of programmable logic blocks in the virtualized computing resource area 2 is 3000, and the available communication bandwidth is 1500M bit/s. In the computing platform field programmable gate array Find the virtualized computing resource areas that satisfy both Aj>Ai and Dj>Di conditions, and sort the virtualized computing resource areas that meet the conditions by (Dj-Di). The virtualized computing resource area with the smallest (Dj-Di) value Execute virtualized computing resource areas as tasks. Therefore, virtualized computing resource area 2 is used as the task execution area.

When there is no virtualized computing resource area that meets the conditions in the computing platform, the running tasks and pending tasks in the virtualized computing resource area are reallocated. The reallocation method is, in order of task priority, the virtualized computing resource area is based on The task requirement setting size is changed by completely reconfiguring the field programmable gate array. If all tasks are allocated to the corresponding processing, the allocation is completed. Otherwise, the unallocated tasks wait for other tasks to be completed, the system computing resources are released, and then step 103 is performed. process.

Step 104: Deploy processing tasks according to the dynamic reallocation results of hardware computing resources;

The compiled task execution bitstream file and the corresponding configuration file are stored in the storage unit of the computing platform. The computing platform loads the corresponding bitstream file to the processor assigned in step 103 according to the user's task requirements, and starts, configures, and executes the task. Stop and uninstall management.

Step 105: Configure the heterogeneous computing resource subscription and publishing client according to the processing task deployment method to realize the interaction of pre-processing and post-processing data.

The processors use a high-speed Ethernet switching network to send and receive data based on a system architecture based on data subscription and release. The general processor is used as the main control node to implement node communication management. The other processors implement the subscription and release system client. The client is in the main control unit. The node creates a communication entity, and the master is responsible for storing the communication entity.

Communication entities include writing data entities and reading data entities for realizing data communication. Each entity is interconnected based on high-speed Ethernet switching network to realize heterogeneous interconnection data sharing. Control data flow and business data flow use different topics to send and receive data.

Set corresponding operations according to the actual needs of the application module, including initialization, configuration, query, start, stop, application data transmission, etc., and code the operations. The identification and processing of the operations are completed by the subscription-publishing adaptation unit. All operations requiring data transmission are implemented by the subscription-publish mechanism.

Referring to Figure 2, a schematic structural diagram of a software-based load computing resource virtualization allocation device provided by an embodiment of the present invention is shown. As shown in Figure 2, it may specifically include the following modules:

System resource monitoring module: responsible for monitoring system resources and task status;

System task management module: responsible for task configuration and runtime control;

Task dynamic deployment module: responsible for user task deployment and dynamic and flexible configuration of computing resources;

Data subscription and publishing module: responsible for communication between processor tasks.

The data subscription and publishing module is divided into 5 functional units, including:

Subscription and publication management unit: responsible for subscription and publication communication management;

Subscription publishing receiving unit: responsible for parsing communication message header information;

Subscription publishing sending unit: responsible for encapsulating communication message header information;

Receiving operation adaptation unit: responsible for parsing the operation names and parameters related to each application module, and then passing the data to a specific FPGA or DSP or CPU logic module for processing;

Send operation adaptation unit: Responsible for encapsulating the operation name and parameters related to each application module.

If the processing results need to be sent externally, the subscription publishing engine encapsulates the results into a request message or a response message based on the preconfigured information, and transmits it through the subscription publishing mechanism.

The processing function is implemented by the application module. Each application module has its own processing unit for processing operation names and parameters. The operation names and parameters are obtained from the configuration file. The application module completes corresponding calculation processing based on the control information and subscription and publication data, gives the processing results, and sends them to the further processing module through the subscription and publication mechanism.

Although the present invention has been disclosed above in terms of preferred embodiments, they are not intended to limit the present invention. Any person skilled in the art can utilize the methods and technical contents disclosed above to improve the present invention without departing from the spirit and scope of the present invention. Possible changes and modifications are made to the technical solution. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention, all belong to the technical solution of the present invention. protected range.

Contents not described in detail in the specification of the present invention belong to the well-known techniques of those skilled in the art.

Claims (3)

1. A software-based load computing resource virtualization allocation method, characterized by including the following steps: 1) Monitor the task status of the spaceborne computing platform and the usage status of heterogeneous computing resources of the spaceborne computing platform, and obtain monitoring data; the task status includes: the status of tasks to be processed and the status of tasks being processed; 2) According to the task status, perform virtualization processing on the heterogeneous computing resources of the spaceborne computing platform and obtain the virtualization division results of the heterogeneous computing resources; 3) Based on the task status and heterogeneous computing resource usage status monitoring data obtained in step 1) and the heterogeneous computing resource virtualization division results obtained in step 2), dynamically allocate the heterogeneous computing resources of the satellite computing platform to obtain the satellite The allocation plan of heterogeneous computing resources on the load computing platform is used as the dynamic allocation result of hardware computing resources; 4) According to the dynamic allocation result of hardware computing resources described in step 3), deploy the tasks to be processed and the tasks being processed, and obtain the deployment method of processing tasks; 5) According to the deployment method of processing tasks in step 4), configure the subscription publishing client on the heterogeneous processor to realize the interaction of pre-processing and post-processing data; The heterogeneous computing resources include: processing resources corresponding to field programmable gate arrays, digital signal processors and general-purpose processors; Step 2) The method for virtualizing heterogeneous computing resources of a spaceborne computing platform includes: a virtualization processing method for a field programmable gate array, a virtualization processing method for a digital signal processor, and a virtualization processing method for a general-purpose processor. Chemical treatment methods; The virtualization processing method of the field programmable gate array is: dividing the field programmable gate array logic resources into a high-speed communication client implementation area and multiple virtualization computing resource areas according to the task status. The virtualization computing The resource area communicates externally through the communication client; The virtualization processing method of the digital signal processor is: according to the task status, based on the dedicated digital signal processor task to send and receive information, the remaining virtualized computing resources are allocated as virtual resources; The virtualization processing method of the general processor is: dividing virtualized computing resources based on the embedded operating system; Step 3) The method for dynamically allocating heterogeneous computing resources on a spaceborne computing platform includes: a dynamic allocation method for general-purpose processors and digital signal processors, and a dynamic allocation method for field programmable gate arrays; The dynamic allocation method of the general processor and the digital signal processor is specifically: A31) Assume that the input rate of data to be processed by task i to be processed is Di bit/s, Ci execution cycles need to be processed, and the tolerated delay is Ti; the main frequency of processor j is Fj, the current usage rate is Pj, and the available communication bandwidth is Dj bit/s; Find processors that satisfy the conditions of Dj>Di and Ci/(Fj*(1-Pj))<Ti among the general processors and digital signal processors of the spaceborne computing platform, and use (Dj-Di) to satisfy the conditions The processors are sorted, and the processor with the smallest (Dj-Di) value is used as the task execution processor; if there is no task execution processor that meets the conditions in the spaceborne computing platform processor, enter step A32), otherwise, Then go directly to step 4); A32) Reallocate running tasks and pending tasks in general processors and digital signal processors. The reallocation method is to search for tasks that meet the task execution delay and communication bandwidth requirements based on the branch and bound method in order of task priority. Task combination, obtain the best match between processor and task combination, as the result of task reallocation; if all tasks are allocated, the allocation ends, and enters step 4), otherwise, enters step A33); the processor includes: General purpose processors and digital signal processors A33) The unallocated tasks wait for other tasks to be executed and the system computing resources are released before executing step 4); The dynamic allocation method of the field programmable gate array is specifically: B31) Assume that the input rate of data to be processed by task i to be processed is Di bit/s, the number of virtualized computing resources required is Ai, the number of resources in the virtualized computing resource area is Aj, and the available communication bandwidth is Dj bit/s; Search the virtualized computing resource areas that meet the conditions of Aj>Ai and Dj>Di in the field programmable gate array of the computing platform, and sort the virtualized computing resource areas that meet the conditions by (Dj-Di), (Dj-Di ) The virtualized computing resource area with the smallest value is used as the task execution virtualized computing resource area; when there is no virtualized computing resource area that meets the conditions in the computing platform, enter step B32); otherwise, end the allocation and enter step 5); B32) Reallocate the tasks being processed and the tasks to be processed in the virtualized computing resource area. The reallocation method is to set the size of the virtualized computing resource area according to the task requirements in order of task priority, by completely reconfiguring the field programmable gate array. To allocate, if all tasks have been allocated, end the allocation and go to step 5), otherwise, go to step B33); B33) The unallocated task waits for other tasks to be executed and the system computing resources are released before executing step 5); The method for realizing pre-processing and post-processing data interaction described in step 5) is specifically: 51) Each processor on the satellite computing platform is interconnected through a high-speed switching network, and one of the general-purpose processors is selected as a communication management control node for data communication management between tasks on each processor; 52) Implement the subscription publishing system client on each satellite computing platform processor; 53) The subscription publishing client on each satellite computing platform processor creates subscribers and publishers based on the deployment method of processing tasks obtained in step 4) for data interaction between tasks on the processor; The high-speed switching network between processors includes: any one of a high-speed optical switching network, a high-speed RapidIO switching network, and/or an Ethernet switching network. 2. A software-based load computing resource virtualization allocation method according to claim 1, characterized in that step 1) describes the task status of the spaceborne computing platform and the heterogeneous computing resource usage status of the spaceborne computing platform. Monitoring methods include: Monitor and obtain the number of free area resources of each field programmable gate array FPGA and the data communication rate of the tasks being processed; Monitor and obtain the total number of resources of each available digital signal processor and the data communication rate of the tasks being processed by the available digital signal processors; Monitor and obtain the total number of resources of each available general-purpose processor and the data communication rate of the tasks being processed by the available general-purpose processor; When there is a task to be processed, the delay requirements, computing requirements and data communication requirements of the task to be processed are obtained according to the task configuration file to be processed. 3. A software-based load computing resource virtualization allocation device used to implement a software-based load computing resource virtualization allocation method as claimed in claim 2, characterized in that it includes: System resource monitoring module: Responsible for monitoring system resources and task status, monitoring the status of pending tasks, status of tasks being processed, and usage status of heterogeneous computing resources on the space-borne computing platform, and obtaining monitoring data; System task management module: Responsible for task configuration and runtime control, dynamically allocate heterogeneous computing resources of the space-borne computing platform based on the status of tasks to be processed and the status of tasks being processed, and obtain the allocation plan for the heterogeneous computing resources of the space-borne computing platform; Task dynamic deployment module: Responsible for user task deployment, deploying pending tasks and ongoing tasks based on the dynamic allocation results of hardware computing resources; Data subscription and publishing module: Responsible for communication between heterogeneous processor tasks. According to the deployment method of processing tasks, the subscription and publishing client is configured on the heterogeneous processor to realize the interaction of pre-processing and post-processing data.