Scheduling method and system for periodic dependency tasks of vehicle heterogeneous central computing platform

Abstract

The invention discloses a scheduling method and system for periodic dependency tasks of a vehicle heterogeneous central computing platform, and relates to the field of task scheduling, and the method comprises the steps: building a vehicle heterogeneous processing platform; according to the number of dependent tasks operated by each heterogeneous computing unit in the automobile heterogeneous processing platform, abstract modeling is carried out on each dependent task, and a dependent task abstract model for the automobile heterogeneous processing platform is obtained; according to the dependency task abstract model, modeling dependency task scheduling of the automobile heterogeneous platform to obtain a task scheduling model; the task scheduling model comprises a task scheduling cost function and a plurality of constraint conditions; the constraint conditions comprise a task execution position constraint, a task execution time constraint, a task round execution time constraint, a task execution sequence constraint and a task data dependency constraint; and solving the task scheduling model to obtain a task scheduling scheme. According to the method, efficient scheduling of the periodic dependency tasks in the automobile heterogeneous processing platform can be realized, and the utilization rate of computing resources is improved.

Description

Technical Field

[0001] The present application relates to the field of task scheduling, and in particular to a scheduling method and system for periodically dependent tasks of a vehicle heterogeneous central computing platform. Background Art

[0002] With the rapid development of autonomous driving and in-vehicle technology, dependent tasks in automotive electronic systems are becoming increasingly complex and their scale is also expanding. Due to space, weight, power consumption (SWaP) and cost considerations, automotive electronic and electrical architecture is gradually shifting from distributed to centralized. This trend has led to more and more dependent tasks with different priorities being integrated into a unified heterogeneous central computing platform. The priority of a task is usually closely related to its safety criticality. The higher the priority, the more serious the consequences of failure to execute the task in time. For example, a car control system may run a high-priority environmental perception task and a relatively low-priority driver voice recognition task at the same time. In this case, users can accept a certain degree of voice recognition delay, but cannot accept the problem of the environmental perception task missing its deadline.

[0003] However, traditional scheduling algorithms have difficulty effectively distinguishing and scheduling dependent tasks with different priorities. Moreover, most of the current scheduling methods for heterogeneous computing platforms are based on heuristic algorithms, which cannot ensure the global optimal solution. Summary of the invention

[0004] The purpose of this application is to provide a scheduling method and system for periodically dependent tasks in a vehicle heterogeneous central computing platform, which can achieve efficient scheduling of periodically dependent tasks in a vehicle heterogeneous central computing platform, improve the utilization rate of computing resources, and reduce the total cost of task execution.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides a method for scheduling periodically dependent tasks of a vehicle heterogeneous central computing platform, including:

[0007] Establishing an automobile heterogeneous processing platform; the automobile heterogeneous processing platform includes a plurality of heterogeneous computing units connected in communication; each heterogeneous computing unit is used to execute a plurality of computing tasks included in each of a plurality of functions having different priorities;

[0008] According to the number of dependent tasks run by each heterogeneous computing unit in the automobile heterogeneous processing platform, abstract modeling is performed on each dependent task to obtain an abstract model of dependent tasks for the automobile heterogeneous processing platform;

[0009] According to the dependent task abstract model, the dependent task scheduling of the automobile heterogeneous platform is modeled to obtain a task scheduling model; the task scheduling model includes a task scheduling cost function and multiple constraints; the constraints include task execution location constraints, task execution time constraints, task round execution time constraints, task execution order constraints and task data dependency constraints;

[0010] The task scheduling model is solved to obtain a task scheduling scheme; the task scheduling scheme includes the heterogeneous computing unit corresponding to each computing task included in each function, the start execution time and the end execution time; the heterogeneous computing unit corresponding to the computing task is a heterogeneous computing unit used to execute the computing task.

[0011] In a second aspect, the present application provides a scheduling system for periodically dependent tasks of a vehicle heterogeneous central computing platform, including:

[0012] An automobile heterogeneous processing platform building module is used to establish an automobile heterogeneous processing platform; the automobile heterogeneous processing platform includes a plurality of heterogeneous computing units connected in communication; each heterogeneous computing unit is used to execute a plurality of computing tasks included in each of a plurality of functions with different priorities;

[0013] A dependent task abstract model building module, used to perform abstract modeling on each dependent task according to the number of dependent tasks run by each heterogeneous computing unit in the automobile heterogeneous processing platform, so as to obtain an abstract model of dependent tasks for the automobile heterogeneous processing platform;

[0014] A task scheduling model building module is used to model the dependent task scheduling of the heterogeneous automobile platform according to the dependent task abstract model to obtain a task scheduling model; the task scheduling model includes a task scheduling cost function and multiple constraints; the constraints include task execution location constraints, task execution time constraints, task round execution time constraints, task execution order constraints and task data dependency constraints;

[0015] A task scheduling scheme calculation module is used to solve the task scheduling model to obtain a task scheduling scheme; the task scheduling scheme includes the heterogeneous computing unit corresponding to each computing task included in each function, the start execution time and the end execution time; the heterogeneous computing unit corresponding to the computing task is a heterogeneous computing unit used to execute the computing task.

[0016] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0017] The present application provides a method and system for scheduling periodic dependent tasks of a heterogeneous central computing platform of a vehicle, the method comprising: constructing a heterogeneous processing platform of an automobile, the platform integrating multiple heterogeneous computing units connected by communication links. Each heterogeneous computing unit is responsible for executing multiple functions, these functions have different priorities, and each function contains multiple computing tasks. In order to optimize the performance of the platform, each dependent task is abstracted according to the number of dependent tasks running in the heterogeneous computing unit, forming a dependent task abstract model dedicated to the heterogeneous processing platform of the automobile. Based on the abstract model, a dependent task scheduling model of the heterogeneous platform of the automobile is further constructed, and the model includes a task scheduling cost function and a series of constraints. These constraints comprehensively cover the requirements of task execution location, execution time, round execution time, execution order, and data dependency, ensuring the accuracy and effectiveness of scheduling. By solving the task scheduling model, a specific task scheduling scheme is obtained. The scheme specifies in detail the heterogeneous computing units to which each computing task should be assigned under each function, the start execution time of the task, and the end execution time. These computing tasks are accurately assigned to the heterogeneous computing units that can execute them efficiently. The solution proposed in this application realizes the efficient scheduling of periodically dependent tasks in the automotive heterogeneous processing platform, significantly improving the utilization of computing resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 A flowchart of a method for scheduling periodically dependent tasks of a vehicle heterogeneous central computing platform provided in one embodiment of the present application;

[0020] Figure 2 A typical automotive heterogeneous central computing platform structure diagram provided for an embodiment of the present application;

[0021] Figure 3 A typical gesture recognition flow chart provided in an embodiment of the present application;

[0022] Figure 4 A schematic diagram of a DAG representation of a typical gesture recognition provided in an embodiment of the present application;

[0023] Figure 5 A schematic diagram of the structure of a scheduling system for periodically dependent tasks of a vehicle heterogeneous central computing platform provided in one embodiment of the present application. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0026] Embodiment 1

[0027] like Figure 1 As shown, this embodiment provides a scheduling method for periodically dependent tasks of a vehicle heterogeneous central computing platform, including:

[0028] Step 101: Establishing an automobile heterogeneous processing platform; the automobile heterogeneous processing platform includes a plurality of heterogeneous computing units connected in communication; each heterogeneous computing unit is used to execute a plurality of computing tasks included in each of a plurality of functions having different priorities;

[0029] Step 102: abstractly modeling each dependent task according to the number of dependent tasks run by each heterogeneous computing unit in the automobile heterogeneous processing platform to obtain an abstract model of dependent tasks for the automobile heterogeneous processing platform;

[0030] Step 103: Modeling the dependent task scheduling of the automobile heterogeneous platform according to the dependent task abstract model to obtain a task scheduling model; the task scheduling model includes a task scheduling cost function and multiple constraints; the constraints include task execution location constraints, task execution time constraints, task round execution time constraints, task execution order constraints, and task data dependency constraints;

[0031] Step 104: Solve the task scheduling model to obtain a task scheduling plan; the task scheduling plan includes the heterogeneous computing unit corresponding to each computing task included in each function, the start execution time and the end execution time; the heterogeneous computing unit corresponding to the computing task is a heterogeneous computing unit used to execute the computing task.

[0032] In some embodiments, when executing step 101, the specific steps may be as follows:

[0033] like Figure 2As shown in the figure, in the automotive heterogeneous processing platform, CPU (general purpose processor), GPU (graphics processor), MCU (microcontroller) and other heterogeneous computing units are connected together through a high-speed bus. In particular, each CPU, GPU, and MCU here contains multiple cores. The entire heterogeneous computing platform is defined as:

[0034] P={P1,P2,...,P |P|}.

[0035] Where |P| represents the number of heterogeneous computing units interconnected by internal high-speed buses in a heterogeneous system, P i Represents a specific heterogeneous computing unit, specifically indicating that the processing cores packaged in a cluster are regarded as a heterogeneous computing unit.

[0036] In some embodiments, when executing step 102, the specific steps may be as follows:

[0037] Establishing a vehicle dependent task set T according to the number of dependent tasks run by each heterogeneous computing unit in the vehicle heterogeneous processing platform;

[0038] For each dependent task in the automobile dependent task set T, the dependent task is abstracted into a DAG to obtain a dependent task abstract model for the automobile heterogeneous processing platform.

[0039] Specifically, the abstract modeling of periodic dependent tasks of heterogeneous processing platforms of hybrid critical systems of automobiles includes the following steps:

[0040] 1) Establish the car-dependent task set T.

[0041] T={T1,T2,...,T |T|}.

[0042] Where |T| represents the number of dependent tasks in the vehicle system running on the heterogeneous processing platform described in step 1, T k Represents an independent dependent task.

[0043] 2) Abstract modeling of individual dependent tasks.

[0044] For a single dependent task T k ,by Figure 3 , Figure 4 For example, Figure 3 This is a diagram of a single dependent task. Figure 4 is with Figure 3 Correspondingly, each subtask is abstracted as a node (for example, face detection is node 12), and the subtask set is:

[0045] V={V1,V2,...,V |V|}, representing dependent task Tk All subtasks.

[0046] The dependencies between subtasks (also called data dependencies) are abstracted as directed edges, and the directed edge set E is a |V|×|V| matrix.

[0047] In this embodiment, a single dependent task T k Abstracted as DAG, it is represented as:

[0048] T k ={V,W,E,P,Prio}.

[0049] Where V = {V1, V2, ..., V |V|} represents dependent task T k All subtasks. W is a |P|×|V| matrix, W represents the worst execution time of dependent tasks on different heterogeneous computing units, W ij Represents dependent task T k Subtask N j In the heterogeneous computing unit P i The worst execution time on the task can be obtained through the operating system timer or debugging tools such as Lauterbach; if a heterogeneous computing unit does not support the subtask, its execution time is set to a positive number much larger than the normal execution time. E represents the directed edge in the DAG, which is a |V|×|V| matrix. If the subtask V j With V j' If there is a dependency between jj' = 1. P and Prio represent the period and priority of the dependent tasks.

[0050] In some embodiments, when executing step 103, the specific steps may be as follows:

[0051] The modeling of dependent task scheduling for heterogeneous automotive platforms based on the dependent task abstract model specifically includes:

[0052] 1) Find the super period of all periodic dependent tasks: the lowest common multiple of all dependent task periods. The execution times of all dependent tasks in a super period are:

[0053] {N1,N2,...,N |T|}.

[0054] 2) Establish subtask execution constraints, as follows:

[0055]

[0056] The proposed scheduling method does not consider task replication. For any dependent task T k Subtask V j, must and can only be executed by a certain heterogeneous computing unit. ijkl is a binary decision variable (can only take 0 or 1), when the dependent task T k Subtask V in the first scheduling j In the heterogeneous computing unit P i The value is 1 when the above is executed, otherwise it is 0.

[0057] 3) Establish the execution time constraints of subtasks on heterogeneous computing units, as follows:

[0058]

[0059]

[0060] In the formula, S ijkl and C ijkl Represents dependent tasks T k The lth scheduling subtask V j In the heterogeneous computing unit P i The start and end time decision variables of the execution are in the range of [0, +∞), W ijk For dependent task T k Subtask V j In the heterogeneous computing unit P i The worst execution time on .

[0061] This embodiment defines when the dependent task T k Subtask V j In the heterogeneous computing unit P i When the task is executed, its value is the actual start and end time of the subtask, otherwise the value is 0. L represents a positive number much larger than the system super period. k Subtask V j Not in heterogeneous computing unit P i If X ijkl =0, and S ijkl and C ijkl The constraint is 0; if the dependent task T k Subtask V j In the heterogeneous computing unit P i If X is executed ijkl =1, in the first formula of 3) above, L is a large positive number, which is equivalent to S ijkl and C ijkl Without constraints, the second formula in 3) above represents the dependent task T k The lth scheduling subtask V j In the heterogeneous computing unit P i The difference between the start and end time of the task on the heterogeneous computing unit P is at least equal toi If the subtask cannot be executed on the heterogeneous computing unit, because its processing time M is a large positive number relative to other processing times, it is obviously not the optimal scheduling solution and the result will be eliminated during the optimization process.

[0062] 4) Establish execution time constraints for different scheduling rounds of dependent tasks, as follows:

[0063]

[0064]

[0065]

[0066]

[0067]

[0068] Among them, S kl Represents dependent task T k The start execution time of the lth scheduling, that is, T k The start time of the first subtask executed in the lth scheduling; where C kl Represents dependent task T k The end execution time of the lth scheduling, that is, T k The third formula above represents the end execution time of the dependent task T k The first round of execution needs to start after the first execution cycle. The fourth formula above represents the dependent task T k The execution in round l needs to start after round l-1 of execution. The fifth formula above represents the dependent task T k The execution ends in the lth round, and the execution must be completed within the lth execution cycle. kl Represents dependent task T k The metric that exceeds its deadline in the lth period.

[0069] 5) Establish the execution order constraints of subtasks on heterogeneous computing units, as follows:

[0070]

[0071]

[0072] Where Y ijj'kk'll' is a binary decision variable, C ij'k'l' For dependent task T k' The l'th scheduled subtask V j'In the heterogeneous computing unit P i The start time decision variable for execution on ij'k'l' For dependent task T k' The l'th scheduled subtask V j' In the heterogeneous computing unit P i The start time decision variable for the execution on .

[0073] The proposed scheduling method does not consider task preemption, that is, once a subtask starts to execute on a heterogeneous computing unit, it will occupy the heterogeneous computing unit until the subtask is completed. The same heterogeneous computing unit can only work in a single thread, that is, it can only process one subtask at a time. Different subtasks assigned to the same heterogeneous computing unit need to make decisions on their calculation order. Define when the dependent task T k The lth scheduling subtask V j In the heterogeneous computing unit P i Prior to the dependent task T k' The l'th scheduled subtask V j' Binary decision variable Y at execution time ijj'kk'll' =1 is 1, otherwise it is 0. The above two constraints are used to constrain the end time of the previous subtask and the start time of the next subtask in two tasks that are executed successively on the same heterogeneous computing unit. In particular, k and k' can be equal, and l and l' can be equal. In other words, the above formula can not only model the execution order of subtasks with different scheduling rounds of different dependent tasks, but also model the execution order of different subtasks with the same scheduling round of a dependent task.

[0074] 6) Establish data dependency constraints for subtasks, as follows:

[0075]

[0076] Each dependent task has a data-dependent subtask that must receive the data of its predecessor task before it can start executing. and Represents dependent tasks T k In the lth scheduling, subtask V j and V j' Start execution time and end execution time.

[0077] 7) Task scheduling cost function modeling, as follows:

[0078]

[0079] In the formula, It is a weight coefficient related to the priority of the dependent task. The higher the priority, the greater the weight coefficient. Usually, the weight coefficient increases by 10 times or even higher when the priority increases by one level. The different real-time requirements of dependent tasks of different priorities are reflected by weighted summation of the time that exceeds the deadline. To represent the dependent task T k The ratio of the completion time of the lth cycle to its deadline, Delta kl To represent the dependent task T k The metric that exceeds its deadline in the lth period.

[0080] In some embodiments, when executing step 104, the specific steps may be as follows:

[0081] Select a programming language (such as C, C++, JAVA or Python, etc.) and its IDE, model the above mathematical model according to the interface requirements of a solver (such as Gurobi, SCIP, COPT), call the corresponding solver to solve it, and obtain the optimal scheduling result.

[0082] Final Scheduling Result This embodiment focuses on the solution results of the three decision variables X, S, and C. X reflects on which heterogeneous computing unit each subtask is executed, and S and C reflect the execution start and end time of each subtask on a certain heterogeneous computing unit.

[0083] Embodiment 2

[0084] like Figure 5 As shown, this embodiment provides a scheduling method for periodically dependent tasks of a vehicle heterogeneous central computing platform, including:

[0085] The automobile heterogeneous processing platform building module 501 is used to establish an automobile heterogeneous processing platform; the automobile heterogeneous processing platform includes a plurality of heterogeneous computing units connected in communication; each heterogeneous computing unit is used to execute a plurality of computing tasks included in each of a plurality of functions with different priorities;

[0086] A dependent task abstract model building module 502 is used to perform abstract modeling on each dependent task according to the number of dependent tasks run by each heterogeneous computing unit in the automobile heterogeneous processing platform, so as to obtain an abstract model of dependent tasks for the automobile heterogeneous processing platform;

[0087] The task scheduling model building module 503 is used to model the dependent task scheduling of the automobile heterogeneous platform according to the dependent task abstract model to obtain a task scheduling model; the task scheduling model includes a task scheduling cost function and multiple constraints; the constraints include task execution location constraints, task execution time constraints, task round execution time constraints, task execution order constraints and task data dependency constraints;

[0088] The task scheduling scheme calculation module 504 is used to solve the task scheduling model to obtain a task scheduling scheme; the task scheduling scheme includes the heterogeneous computing unit corresponding to each computing task included in each function, the start execution time and the end execution time; the heterogeneous computing unit corresponding to the computing task is a heterogeneous computing unit used to execute the computing task.

[0089] In summary, this application has the following technical effects:

[0090] 1) The heterogeneous computing platform periodically dependent task scheduling method provided in this application determines the optimal scheduling strategy of the system through a mixed integer linear programming method, ensures the orderly execution of each task application, and improves the overall safety of the vehicle and the utilization efficiency of computing resources.

[0091] 2) The abstraction and modeling method of the automotive heterogeneous central computing platform proposed in this application does not require a detailed internal chip layout of the heterogeneous central computing platform and has no requirements for the specific models of the heterogeneous computing units.

[0092] 3) The dependent task abstract modeling method proposed in this application comprehensively considers the heterogeneity, parallelism and priority of the system.

[0093] 4) This application establishes a mathematical model for periodically dependent task scheduling from a theoretical level, so that this method can obtain the optimal scheduling of the entire system.

[0094] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0095] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A scheduling method for periodically dependent tasks of a vehicle heterogeneous central computing platform, characterized in that: include: Establishing a heterogeneous automotive processing platform; The automotive heterogeneous processing platform includes a plurality of heterogeneous computing units connected in communication; Each heterogeneous computing unit is used to execute multiple computing tasks included in each of multiple functions with different priorities; According to the number of dependent tasks run by each heterogeneous computing unit in the automobile heterogeneous processing platform, abstract modeling is performed on each dependent task to obtain an abstract model of dependent tasks for the automobile heterogeneous processing platform; According to the dependent task abstract model, the dependent task scheduling of the automobile heterogeneous platform is modeled to obtain a task scheduling model; the task scheduling model includes a task scheduling cost function and multiple constraints; the constraints include task execution location constraints, task execution time constraints, task round execution time constraints, task execution order constraints and task data dependency constraints; The task scheduling model is solved to obtain a task scheduling scheme; the task scheduling scheme includes the heterogeneous computing unit corresponding to each computing task included in each function, the start execution time and the end execution time; the heterogeneous computing unit corresponding to the computing task is a heterogeneous computing unit used to execute the computing task.

2. The method for scheduling periodically dependent tasks of a vehicle heterogeneous central computing platform according to claim 1, characterized in that: The automotive heterogeneous processing platform is specifically: P={P1,P2,...,P |P| }; Where |P| represents the number of heterogeneous processing units interconnected by internal high-speed buses in a heterogeneous system, P i represents the i-th heterogeneous processing unit.

3. The method for scheduling periodically dependent tasks of a vehicle heterogeneous central computing platform according to claim 2, characterized in that: The heterogeneous computing unit includes an image processor, a general processor and a microcontroller.

4. The method for scheduling periodically dependent tasks of a vehicle heterogeneous central computing platform according to claim 3 is characterized in that: According to the number of dependent tasks run by each heterogeneous computing unit in the automobile heterogeneous processing platform, each dependent task is abstractly modeled to obtain an abstract model of dependent tasks for the automobile heterogeneous processing platform, which specifically includes: Establishing a vehicle dependent task set T according to the number of dependent tasks run by each heterogeneous computing unit in the vehicle heterogeneous processing platform; For each dependent task in the automobile dependent task set T, the dependent task is abstracted into a DAG to obtain a dependent task abstract model for the automobile heterogeneous processing platform.

5. The method for scheduling periodically dependent tasks of a vehicle heterogeneous central computing platform according to claim 4 is characterized in that: The dependent task abstract model is specifically: T k ={V,W,E,P,Prio}; Where V = {V1, V2, ..., V |V| } is the dependent task T k all subtasks; W is a |P|×|V| matrix, W represents the worst execution time of dependent tasks on different heterogeneous computing units, E represents the directed edge P in DAG, and Prio represents the period and priority of dependent tasks.

6. The method for scheduling periodically dependent tasks of a vehicle heterogeneous central computing platform according to claim 5, characterized in that: The task execution location constraints in the task scheduling model are specifically: Among them, X ijkl is a binary decision variable, |V k |For dependent tasks T k The number of subtasks in; N k is the number of executions of the dependent task, and l is the lth time.

7. The method for scheduling periodically dependent tasks of a vehicle heterogeneous central computing platform according to claim 6, characterized in that: The task execution time constraints in the task scheduling model are specifically: Among them, S ijkl and C ijkl Represents dependent tasks T k The lth scheduling subtask V j In the heterogeneous computing unit P i The start and end time decision variables of the execution, W ijk For dependent task T k Subtask V j In the heterogeneous computing unit P i The worst execution time on .

8. The method for scheduling periodically dependent tasks of a vehicle heterogeneous central computing platform according to claim 7, characterized in that: The task execution order constraints in the task scheduling model are specifically: Among them, Y ijj'kk'll' is a binary decision variable, C ij'k'l' For dependent task T k' The l'th scheduled subtask V j' In the heterogeneous computing unit P i The start time decision variable for execution on ij'k'l' For dependent task T k' The l'th scheduled subtask V j' In the heterogeneous computing unit P i The start time decision variable for the execution on .

9. The method for scheduling periodically dependent tasks of a vehicle heterogeneous central computing platform according to claim 8, characterized in that: The task scheduling cost function in the task scheduling model is specifically: in, is the weight coefficient related to the priority of the dependent tasks, To represent the dependent task T k The ratio of the completion time of the lth cycle to its deadline, Delta kl To represent the dependent task T k The metric that exceeds its deadline in the lth period.

10. A scheduling system for periodically dependent tasks of a vehicle heterogeneous central computing platform, characterized in that: include: Automotive heterogeneous processing platform building module, used to build automotive heterogeneous processing platform; The automotive heterogeneous processing platform includes a plurality of heterogeneous computing units connected in communication; Each heterogeneous computing unit is used to execute multiple computing tasks included in each of multiple functions with different priorities; A dependent task abstract model building module, used to perform abstract modeling on each dependent task according to the number of dependent tasks run by each heterogeneous computing unit in the automobile heterogeneous processing platform, so as to obtain an abstract model of dependent tasks for the automobile heterogeneous processing platform; A task scheduling model building module is used to model the dependent task scheduling of the automobile heterogeneous platform according to the dependent task abstract model to obtain a task scheduling model; the task scheduling model includes a task scheduling cost function and multiple constraints; the constraints include task execution location constraints, task execution time constraints, task round execution time constraints, task execution order constraints and task data dependency constraints; A task scheduling scheme calculation module is used to solve the task scheduling model to obtain a task scheduling scheme; the task scheduling scheme includes the heterogeneous computing unit corresponding to each computing task included in each function, the start execution time and the end execution time; the heterogeneous computing unit corresponding to the computing task is a heterogeneous computing unit used to execute the computing task.

Images