WO2019228285A1

WO2019228285A1 - Task scheduling method and device

Info

Publication number: WO2019228285A1
Application number: PCT/CN2019/088450
Authority: WO
Inventors: 余国生; 代成成; 董晓文
Original assignee: 华为技术有限公司
Priority date: 2018-05-28
Filing date: 2019-05-25
Publication date: 2019-12-05
Also published as: CN110543148A; CN110543148B

Abstract

Embodiments of the present application relate to the field of artificial intelligence. Disclosed are a task scheduling method and device, capable of improving the working efficiency of a vehicle-mounted device in comprehensive consideration of delay requirements and power consumption requirements in a running state. The method comprises: determining a motion state of a mobile device, and determining, according to the motion state of the mobile device, M tasks to be carried out; determining N delay-sensitive tasks in the M tasks; for the delay-sensitive tasks, computing power consumption demands of the delay-sensitive tasks; determining operators having target power consumption headroom ranges matching the power consumption demands of the delay-sensitive tasks and being capable of finishing the delay-sensitive tasks within delay-required durations of the delay-sensitive tasks as target operators for carrying out the delay-sensitive tasks; and allocating the delay-sensitive tasks to the target operators for delay-sensitive tasks.

Description

Method and device for task scheduling

Technical field

The embodiments of the present application relate to the field of artificial intelligence, and in particular, to a method and a device for task scheduling.

Background technique

Automated driving systems use technologies such as computers, modern sensing, information fusion, communication, artificial intelligence, and automatic control to achieve functions such as environmental perception, behavioral decision-making, path planning, and motion control. At present, the processing flow of the automatic driving system is: the vehicle-mounted device collects image data collected by various cameras distributed on the vehicle body, and transmits these image data to the computing device of the vehicle-mounted device. After deep learning of these image data, the surrounding environment of the vehicle body (such as: the number of pedestrians, the distance from pedestrians, etc.) is identified. Finally, decisions can be made based on the identified environment, such as controlling vehicle speed and changing paths.

In the autonomous driving scenario, the task delay is demanding, that is, the computing device of the vehicle-mounted device is required to perform the task within a strict time, such as identifying the surrounding environment of the vehicle body. Along with the demanding time delay requirements, the calculation amount of the computing unit increases greatly, which will greatly increase the energy consumption of the vehicle-mounted device. In addition, in-vehicle devices are usually sensitive to energy consumption, and the driving energy consumption should be minimized during the driving process. However, if the computing energy consumption is kept low during the driving process, it is difficult to meet the strict delay requirements of the driving process. It can be seen that the existing technology cannot simultaneously meet the requirements for energy consumption and delay in a vehicle environment.

Summary of the Invention

The embodiments of the present application provide a task scheduling method and device, which can comprehensively consider delay requirements and power consumption requirements in a driving state, and improve the working efficiency of a vehicle-mounted device.

To achieve the above purpose, the embodiments of the present application adopt the following technical solutions:

In a first aspect, a task scheduling method is disclosed. The vehicle-mounted device first determines a motion state of a mobile device, determines M tasks to be executed according to the motion state of the mobile device, and determines N delay-sensitive tasks of the M tasks. . Among them, M is an integer greater than or equal to 1, and N is an integer less than or equal to M. Further, for each delay-sensitive task of the N delay-sensitive tasks, the power consumption requirements of the delay-sensitive tasks are calculated, and the target power consumption margin range is matched with the power consumption requirements of the delay-sensitive tasks. The delay of a delay-sensitive task requires that the processor that has executed the delay-sensitive task within the time period is determined as the target processor to execute the delay-sensitive task, and then assigns the delay-sensitive task to the target processor of the delay-sensitive task. It should be noted that the target power consumption margin range is determined according to the target power consumption range of the mobile device's computing unit and the current power consumption of the computing unit; the power consumption requirements of delay-sensitive tasks are required to perform delay-sensitive tasks. The delay required for delay-sensitive tasks is the time required to execute delay-sensitive tasks.

In the task scheduling method provided in the embodiment of the present invention, the in-vehicle device can reasonably schedule tasks that need to be executed during the travel of the vehicle, which not only meets the delay requirement of the task, but also enables the computing unit to work under the target power consumption state. Considering the requirements of driving state, delay and power consumption, the working efficiency of the vehicle platform is improved.

With reference to the first aspect, in a first possible implementation manner of the first aspect, the determination of the N delay-sensitive tasks of the M tasks by the vehicle-mounted device specifically includes: determining a delay of each of the M tasks according to a motion state. Sensitivity; determine the priority level of each of the M tasks according to the delay sensitivity of each of the M tasks; the priority level is proportional to the delay sensitivity; finally, the priority of the M tasks The N tasks with higher ranks are determined as N delay-sensitive tasks.

In the embodiment of the present invention, a task with high delay sensitivity may be determined as a delay sensitive task. Further, the vehicle-mounted device first considers assigning an operator to the delay sensitive task in subsequent processing to ensure that the delay sensitive task can be executed preferentially. .

With reference to the first aspect or the first possible implementation manner of the first aspect, in the second possible implementation manner of the first aspect, the power consumption requirements of delay-sensitive tasks are calculated, and the target power consumption margin range and time are calculated. Delay-sensitive tasks that match the power requirements of the delay-sensitive tasks and that can be executed within the delay-required duration of the processor are determined to be the target processor for delay-sensitive tasks. Specific tasks include: Power requirements; for each operator of the mobile device, determine the duration of the delay-sensitive tasks performed by the operator and the target power consumption margin of the operator; determine whether the power requirements of the delay-sensitive tasks fall within the target power of the operator Whether the delay time of the delay-sensitive task is not exceeded by the delay range of the processor and the delay-sensitive task is performed; if the power consumption requirement of the delay-sensitive task falls within the target power consumption margin of the processor and the operation If the duration of the delay-sensitive task is not exceeded by the delay-sensitive task, the processor is determined to be the target processor of the delay-sensitive task.

In the embodiment of the present invention, the target power consumption range refers to that when the power consumption of the computing unit is maintained within the range, the performance of the computing unit reaches the optimal state. If the power consumption requirement of the task falls within the target power consumption margin range , It means that the computing unit can perform the task in an optimal state, thereby improving the computing power of the computing unit and further improving the working efficiency of the vehicle platform.

With reference to the second possible implementation manner of the first aspect, in a third possible implementation manner of the first aspect, the method further includes: if the power consumption requirement of the delay-sensitive task and the target power of all the computing devices of the mobile device If the consumption ranges do not match, the input data of the delay-sensitive task is split into at least two sub-data blocks; for each of the at least one sub-data block, the power consumption required to calculate the sub-data block is determined If the target power consumption margin range of one operator of the mobile device matches the power consumption, determine the operator as the target operator corresponding to the sub-data block; determine the target operator corresponding to all sub-data blocks as delay-sensitive The target computing unit corresponding to the task.

In the embodiment of the present invention, delay-sensitive tasks can be split and executed by multiple processors. This can ensure that delay-sensitive tasks can be executed quickly, ensure that the task's delay requirements are met, and will not affect the follow-up of the vehicle-mounted device. decision making.

In combination with the first aspect or any one of the first to third possible implementation manners of the first aspect, in a fourth possible implementation manner of the first aspect, delay-sensitive tasks are assigned to delay-sensitive tasks The target operator specifically includes: calculating the execution time of N delay-sensitive tasks, and determining the delay request duration of the delay-sensitive task execution time according to the execution time of the delay-sensitive task; generating a task list, and the task list is used to record N Each target delay-sensitive task corresponds to the target operator and the delay requirement corresponding to each target operator, so that the mobile device's operator determines the delay-sensitive task to be executed according to the task list and waits for the execution. Delay-sensitive tasks require that the delay-sensitive tasks to be executed are completed within the time period.

In the embodiment of the present invention, a task may be assigned to a corresponding computing unit in a task list manner, and the computing unit may perform calculation of the corresponding task according to the task list.

With reference to the first possible implementation manner of the first aspect, in a fifth possible implementation manner of the first aspect, the method provided by the embodiment of the present invention further includes: delaying each of the N delay-sensitive tasks After the sensitive task is assigned to the corresponding target computing unit, for each task of the M tasks except N delay-sensitive tasks, the task's power consumption requirement is calculated, and the target power consumption margin range is related to the task's power consumption requirement. The computing unit that is matched and capable of executing the task within the delay required time of the task is determined as the target computing unit to execute the task, and the task is allocated to the target computing unit of the delay-sensitive task. Among them, the power consumption requirement of the task is the power consumption required to execute the task, and the delay requirement of the task is the time required to execute the task.

In the embodiment of the present invention, after an operator is assigned to the delay-sensitive task, an operator may also be assigned to the non-delay-sensitive task to ensure that the non-delay-sensitive task can be executed and avoid affecting the subsequent decision-making of the vehicle-mounted device. In addition, the computing unit can perform non-time delay sensitive tasks in the best state, which improves the computing power of the computing unit and further improves the working efficiency of the vehicle platform.

With reference to the second possible implementation manner of the first aspect, in a sixth possible implementation manner of the first aspect, the duration T of the processor performing the delay-sensitive task satisfies:

Among them, Nc and Nm are the number of basic computing units inside the computing unit, f ₁ is the frequency of the core on the computing unit; f ₂ is the frequency of the memory of the computing unit, and C _i is the double-precision floating point required to perform delay-sensitive tasks. The number of operations, m _j is the frequency of memory access required to complete the delay-sensitive task; the memory access frequency is the number of times the memory is accessed per second.

In the embodiment of the present invention, the time required for each computing unit to perform a certain task can be calculated by using the foregoing formula, and then it can be determined whether the computing unit meets the delay requirement of the task.

With reference to the first aspect or any one of the first to sixth possible implementation manners of the first aspect, in a seventh possible implementation manner of the first aspect, the power consumption requirement W of the delay-sensitive task satisfies: W = n * w ^, where n is the number of basic operations required to complete the delay-sensitive task, and w ^ is the power consumption of the basic operation performed by the arithmetic unit.

In the embodiment of the present invention, the power consumption required by each computing unit to perform a certain task can be calculated by using the foregoing formula, and then it can be determined whether the computing unit meets the power consumption requirement of the task.

In a second aspect, an apparatus is disclosed, including: a determining unit for determining a motion state of a mobile device, determining M tasks to be performed according to the motion state of the mobile device, and determining N delay-sensitive tasks among the M tasks ; M is an integer greater than or equal to 1, N is an integer less than or equal to M; a calculation unit for calculating the power consumption requirements of the delay-sensitive tasks for each delay-sensitive task of the N delay-sensitive tasks; a determination unit It is also used to determine that the target power consumption margin range matches the power consumption requirements of the delay-sensitive task and that the processor capable of executing the delay-sensitive task within the delay-required duration of the delay-sensitive task is determined to execute the delay-sensitive task Target arithmetic unit; an allocation unit for assigning delay-sensitive tasks to the target arithmetic unit of delay-sensitive tasks. It should be noted that the target power consumption margin range is determined according to the target power consumption range of the mobile device's computing unit and the current power consumption of the computing unit; the power consumption requirements of delay-sensitive tasks are required to perform delay-sensitive tasks. The delay required for delay-sensitive tasks is the time required to execute delay-sensitive tasks.

The device provided in the embodiment of the present invention can reasonably schedule tasks that need to be performed during the travel of the vehicle, not only meeting the delay time required for the task, but also enabling the computing unit to work under the target power consumption state, taking into account the driving state, The requirements of delay and power consumption improve the working efficiency of the vehicle platform.

With reference to the second aspect, in a first possible implementation manner of the second aspect, the determining unit is specifically configured to determine the delay sensitivity of each of the M tasks according to the motion state; and according to each of the M tasks The delay sensitivity determines the priority level of each of the M tasks; the priority level is proportional to the delay sensitivity; the N tasks with a higher priority level among the M tasks are determined as N delays Sensitive tasks.

With reference to the second aspect or the first possible implementation manner of the second aspect, in the second possible implementation manner of the second aspect, the computing unit is specifically configured to calculate the power consumption requirements of delay-sensitive tasks; for mobile Each operator of the device determines the duration of the delay-sensitive tasks performed by the operator and the target power consumption margin range of the operator; the determination unit is specifically used to determine whether the power requirements of the delay-sensitive tasks fall within the target of the operator Within the power consumption margin and whether the delay time of the delay-sensitive task performed by the component does not exceed the delay requirement length of the delay-sensitive task; if the power consumption requirement of the delay-sensitive task falls within the target power consumption margin of the component and The execution time of the delay-sensitive task by the arithmetic unit does not exceed the delay required time of the delay-sensitive task. Then, the arithmetic unit is determined as the target arithmetic unit of the delay-sensitive task.

With reference to the second possible implementation manner of the second aspect, in a third possible implementation manner of the second aspect, the determining unit is further configured to, if the power consumption requirements of the delay-sensitive task and all the computing devices of the mobile device If the target power consumption margin ranges do not match, the input data of the delay-sensitive task is split into at least two sub-data blocks; for each sub-data block in the at least one sub-data block, the required calculation of the sub-data block is determined. If the target power consumption margin range of a computing unit of the mobile device matches the power consumption, the computing unit is determined as the target computing unit corresponding to the sub-data block; the target computing unit corresponding to all sub-data blocks is determined as Target processor for delay-sensitive tasks.

With reference to the second aspect or any one of the first to third possible implementation manners of the second aspect, in a fourth possible implementation manner of the second aspect, the allocation unit is specifically configured to calculate N delays The execution time of sensitive tasks. Determine the delay requirement of the execution time of delay-sensitive tasks according to the execution time of delay-sensitive tasks.

Generate a task list. The task list is used to record the target processor corresponding to each delay-sensitive task in the N delay-sensitive tasks and the delay requirement duration corresponding to each target processor, so that the mobile device's processor can determine it according to the task list. The delay-sensitive task to be executed and the delay-sensitive task to be executed are completed within the required time delay.

With reference to the first possible implementation manner of the second aspect, in the fifth possible implementation manner of the second aspect, the computing unit is further configured to allocate each delay-sensitive task of the N delay-sensitive tasks After giving the corresponding target computing unit, for each task of the M tasks except N delay-sensitive tasks, calculate the power consumption requirements of the task; the allocation unit is also used to compare the target power consumption margin range with the task's power The computing unit that matches the power consumption requirements and is able to complete the task within the required delay time of the task is determined as the target computing unit to execute the task, and the task is allocated to the target computing unit of the delay-sensitive task; among them, the power consumption requirement of the task is The power required to execute a task. The delay required for a task is the time required to execute the task.

With reference to the second possible implementation manner of the second aspect, in a sixth possible implementation manner of the second aspect, the duration T of the processor performing the delay-sensitive task satisfies:

With reference to the second aspect or any one of the first to sixth possible implementation manners of the second aspect, in a seventh possible implementation manner of the second aspect, the power consumption requirement of the delay-sensitive task W satisfies: W = n * w ^, where n is the number of basic operations required to complete the delay-sensitive task, and w ^ is the power consumption of the basic operation performed by the arithmetic unit.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a structural diagram of a vehicle-mounted system provided by an embodiment of the present application; FIG.

FIG. 2 is another architecture diagram of a vehicle-mounted system provided by an embodiment of the present invention; FIG.

FIG. 3 is a schematic diagram of preprocessing provided by an embodiment of the present invention; FIG.

4 is a schematic flowchart of a task scheduling method according to an embodiment of the present invention;

5 is a schematic diagram of input data of a task according to an embodiment of the present invention;

6 is a schematic diagram of high and low speeds provided by an embodiment of the present invention;

7 is a schematic flowchart of a task allocation method according to an embodiment of the present invention;

8 is a schematic diagram of a task state according to an embodiment of the present invention;

FIG. 9 is a schematic diagram of another task state according to an embodiment of the present invention; FIG.

FIG. 10 is a structural block diagram of a device according to an embodiment of the present invention; FIG.

FIG. 11 is another structural block diagram of an apparatus according to an embodiment of the present invention; FIG.

FIG. 12 is another structural block diagram of a device provided by an embodiment of the present invention.

Detailed ways

Currently, the architecture of the vehicle-mounted system is shown in Figure 1. Referring to FIG. 1, the vehicle body is provided with a monitoring device, and a vehicle-mounted device is installed inside the vehicle. The monitoring device may be a sensor, such as a camera, a laser sensor, and the like. In an autonomous driving scenario, the monitoring device collects environmental information around the vehicle body in real time, performs calculations based on the collected information (such as images taken by a camera), identifies objects around the vehicle body based on the calculation results, and provides input for subsequent decision-making. For example, according to the calculation results, the speed of the vehicle and the route of the vehicle are controlled.

The requirements for time delay in an automatic driving scenario are extremely high. This is because the surrounding environment of the vehicle body is complex, and the vehicle-mounted device needs to perform a large number of calculations quickly during high-speed driving of the vehicle, so that objects around the vehicle body can be identified in time. A large amount of calculations can cause a huge amount of energy consumption for on-board devices. However, in-vehicle devices are usually sensitive to energy consumption, and the driving process should reduce energy consumption as much as possible. It can be seen that the prior art cannot balance the requirements of vehicle-mounted devices on energy consumption and delay in a vehicle-mounted environment.

An embodiment of the present invention provides a task scheduling method. An in-vehicle device of a mobile device first determines a motion state of the mobile device, and determines M tasks to be performed according to the motion state of the mobile device. Then, N of the M tasks can also be determined. Delay-sensitive tasks. Further, for each of the above N delay-sensitive tasks, the following steps are performed: the vehicle-mounted device calculates the power consumption requirements of the delay-sensitive tasks, and compares the target power consumption margin range with the power consumption of the delay-sensitive tasks. An operator that matches the requirements and is capable of executing the delay-sensitive task within the delay required time of the delay-sensitive task determines a target processor that executes the delay-sensitive task. Further, the delay-sensitive task is assigned a corresponding target computing unit. It can be seen that, in the embodiment of the present invention, the in-vehicle device can reasonably schedule tasks that need to be performed during the travel of the vehicle, which not only meets the delay time required for the task, but also enables the computing unit to work at the target power consumption state, taking into account driving The requirements of state, delay and power consumption improve the working efficiency of the vehicle platform.

An embodiment of the present invention further provides a vehicle-mounted system. As shown in FIG. 2, the vehicle-mounted system includes: a vehicle-mounted device 10 and at least one detection device 20. The vehicle-mounted device 10 includes input / output (IO) logic 101, a scheduling module 102, and at least one computing unit 103. Specifically, the scheduling module 102 may be a central processing unit (CPU), the computing unit 103 may be an ACC (Accumulator) or a graphics processor (Graphics Processing Unit, GPU), or a neural network chip. The scheduling module 102 may allocate one or more tasks to be executed to the processor 103. The detection device 20 may be a sensing device such as a camera, a laser sensor, or a millimeter wave radar.

It should be noted that, taking a mobile device as a vehicle as an example, the camera installed on the mobile device includes the following types:

(1) Long-distance camera: It is located in front of the vehicle body and is used to identify distant targets, such as traffic lights;

(2) Medium-range camera: It is located in front of the vehicle body and is mainly used to identify lane lines, people, cars, non-motor vehicles, obstacles, etc., and can also be used for distance measurement;

(3) Close-range camera: located around the vehicle body, mainly identifying people, cars, non-motor vehicles, obstacles, and driveable areas;

In addition, other cameras can be installed on the mobile device, such as a camera located behind the vehicle body, which can be used for parking and parking; a camera installed near the driver's seat inside the vehicle, which can be used to monitor fatigue driving.

In the embodiment of the present invention, the detection device 20 collects information about the surrounding environment of the vehicle body, for example, the cameras of the vehicle body distribution capture images at different angles, breadths, detection distances, frame rates, and resolutions. The detection device 20 passes the collected information to the IO logic 101, and the IO logic 101 performs unified management and preprocessing on the received information. The preprocessing of the received information (such as images) by the IO logic 101 may be to classify the images captured by the camera. As shown in FIG. 3, each camera captures m frames of images, and the images captured by all cameras can be classified in order according to the number of frames of the images. For example, the first frame (frame0) of the images captured by all cameras is grouped, the second frame (frame1) of all images is grouped, and so on. Images are classified.

Subsequently, the IO logic 101 passes the pre-processed image to the scheduling module 102. The scheduling module 102 generates a computing task according to the information input by the IO logic 101, and sends the computing task to the processor 103 for processing. The processor 103 runs the deep nerve Network (Deep Neural Network, DNN) algorithm, and returns the calculation result to the scheduling module 102. The scheduling module 102 can make subsequent decisions according to the calculation results.

An embodiment of the present invention provides a task scheduling method, which can provide support for task scheduling of the scheduling module 102. As shown in Figure 4, the method includes the following steps:

401. The scheduling module 102 determines a current motion state of the mobile device, and determines M tasks to be executed according to the motion state of the mobile device.

It should be noted that the scheduling module 102 may receive data sent by the detection device 20 and calculate the current motion state of the mobile device according to the data sent by the detection device 20, such as the running speed of the mobile device. In addition, the mobile device in the embodiment of the present invention may be a device with a mobile function such as a vehicle or a drone. In addition, M is an integer of 1 or more.

In specific implementation, the scheduling module 102 determines the DNN tasks that need to be executed according to the vehicle sensing system algorithm, that is, the M tasks described in the embodiment of the present invention, such as: detecting lane lines, detecting pedestrians, detecting motor vehicles, and detecting non-machines. Moving cars and other tasks. In the embodiment of the present invention, each task is inputted with data (for example, an image taken by a camera) obtained by multiple detection devices 20, and different tasks are performed by different processors. For example, referring to FIG. 5, the mobile device has three tasks: Task 0, Task 1, and Task 2 under the current motion state. The input data required for the processor to execute Task 0 is taken by camera 0, camera 1, and camera 3. For images, the input data required to execute Task 1 is the images captured by camera 1 and camera 2, and the input data required to execute Task 2 is the images captured by camera 2 and camera 3.

402. The scheduling module 102 determines N delay-sensitive tasks among the M tasks.

It should be noted that the delay sensitivities of tasks are different for different movement states of mobile devices. For example, when the mobile device is traveling at high speed, the in-vehicle device 10 is required to be able to quickly identify vehicles, pedestrians, obstacles, etc. in front of the mobile device. Therefore, tasks such as "identify vehicles in front" and "identify pedestrians in front" have higher priority; in addition, related tasks that identify conditions behind mobile devices have relatively low priorities, such as "identify vehicles in the rear", "identify pedestrians in the rear", etc. task. In addition, N is an integer of M or less.

In specific implementation, the scheduling module 102 may determine the delay sensitivity of each of the M tasks according to the motion state of the mobile device. Further, since the priority level of the task is proportional to the delay sensitivity of the task, the scheduling module 102 may also determine the priority level of the task according to the delay sensitivity of each task. In addition, the delay sensitivity of the task can also be determined according to the delay requirement duration of the task. The shorter the delay requirement duration, the higher the task's delay sensitivity. For example, the delay requirements of task 1, task 2, and task 3 are 5 s, 10 s, and 8 s, respectively. Therefore, the three tasks are task 1, task 3, and task 2 in descending order of delay sensitivity, that is, According to the priority level, task 1, task 3, and task 2 are in order.

In the embodiment of the present invention, the delay required duration of a task may be an estimated time required to perform the task, and the time required to perform the task may generally be determined according to the amount of input data required for a task. For example, the input data required to perform task 1 are the images taken by camera 1 and camera 2 within 1 s. It is assumed that camera 1 and camera 2 each took 8 images within 1 s. Therefore, the input data required to perform task 1 It is 16 images. If it is specified that the processing time of an image cannot exceed 2ms, the time required to execute the task cannot exceed 2 * 16 = 32ms, so the delay required for the task is 32ms.

Subsequently, the scheduling module 102 may also arrange the M tasks in order of priority from high to low, and determine N tasks with higher priority levels as N delay-sensitive tasks.

In some embodiments, the motion state of the mobile device can be divided into high speed and low speed, and the delay requirements of tasks under different motion states are different. Specifically, when the motion state of the mobile device is high speed, it is required to be able to process tasks quickly, and the time delay of the task is relatively high; when the motion state of the mobile device is low speed, the time delay requirement of the task is relatively low.

In the embodiment of the present invention, high speed and low speed may be defined according to traffic rules in different regions, and of course, high speed and low speed may also be defined according to other factors, and the embodiment of the present invention does not limit this. FIG. 6 is an example of a low speed and a high speed provided by an embodiment of the present invention. Referring to FIG. 6, when the mobile device's speed is 80 to 120 km / h, the movement state of the mobile device is high speed; the mobile device's speed is below 80 km / h , The mobile device's motion state is low speed. Under the high-speed state of the mobile device, the processing frame rate of the computing unit 103 is high. For example, when the mobile device's speed is 80 ~ 120km / h, the computing frame rate calculated by the computing unit is not less than 30FPS, that is, the processing time of each frame. The delay does not exceed 33ms. In the low-speed state of the mobile device, the lower processing frame rate can meet the demand. For example, when the mobile device's speed is 20km / h, the calculated frame rate is 5FPS, that is, the processing time of each frame is 200ms. Further, the delay required duration of a task can be determined according to the processing delay of each frame. For example, when the mobile device is driving at a high speed, the input data required for task 1 is 10 images taken by camera 1 and camera 2. Then it can be estimated that the time required to perform task 1 is 10 * 33 = 330 ms, that is, the delay of task 1 The required time is 330ms.

Subsequently, the scheduling module 102 executes the following steps 403 to 404 for each delay-sensitive task of the N delay-sensitive tasks to determine a target processor for each delay-sensitive task. In the embodiment of the present invention, the scheduling module 102 may assign a target computing unit to each of the delay-sensitive tasks in the order of priority levels from high to low, or may perform calculations from large to large. A small order assigns a target computing unit to each of the N delay-sensitive tasks. If the priority levels of delay-sensitive tasks are the same, the target processors can be allocated in order from large to small.

In the following description of steps 403 to 404, the delay-sensitive task 1 is taken as an example, and it is described in detail that the target processor is allocated to the delay-sensitive task in the embodiment of the present invention.

403. The scheduling module 102 determines a target power consumption margin range of each computing unit 103 of the vehicle-mounted device and a time required for each calculation 103 to execute the delay-sensitive task 1.

In specific implementation, the scheduling module 102 can collect the current state of each computing unit through the following two steps to determine the target power consumption margin range of an computing unit 103, which specifically includes:

Step A: The scheduling module 102 first determines the current power consumption of the computing unit, that is, the power consumption occupied by the computing task currently being performed by the computing unit.

Step B: The scheduling module 102 determines the target power consumption margin range of the computing unit according to the current power consumption of the computing unit and the target power consumption range of the computing unit. It should be noted that the target power consumption range refers to the performance of the computing unit when the power consumption of the computing unit is maintained within this range. The scheduling module 102 can read the target power consumption range of the computing unit from the register. The target power consumption range of the computing unit can be obtained from the chip or test.

It should be noted that the scheduling module 102 can also obtain the maximum power consumption limit of the computing unit. When performing the task allocation, the scheduling module 102 needs to consider the maximum power consumption limit of the computing unit. Consumption cannot exceed this limit.

In addition, the scheduling module 102 can estimate the time T required for an operator to complete a task according to the following formula (1). In the task assignment process, not only must the operator work at the target power consumption, but also the operation must be guaranteed. The processor completes the task within the delay request time t of the task, so T does not exceed the task's delay request time t, that is, T is less than or equal to, so as to ensure that the calculation of the task can be completed.

Nc and Nm in formula (1) are the number of basic computing units inside the operator, f ₁ is the frequency of the core on the operator; f ₂ is the frequency of the memory of the operator, and C _i is the double precision required to complete the task Number of floating-point operations, m _j is the memory access frequency required to complete the delay-sensitive task; the memory access frequency is the number of times the memory is accessed per second.

In the embodiment of the present invention, the capability information of each computing unit may be recorded through the capability information table, including: a target power consumption range margin range of the computing unit and a time required for the computing unit to perform the delay-sensitive task 1. Table 1 is a possible implementation manner of the capability information table. Of course, the embodiment of the present invention does not limit the implementation manner of the capability information table. Referring to Table 1, the capability information table may also include the target power consumption range of the computing unit and the current power consumption.

Table 1

It should be noted that, in Table 1, the vehicle-mounted device should include the capability information of all the computing units (the computing unit used to calculate the DNN task). Table 1 takes the vehicle-mounted device including two computing units as an example. In addition, when assigning a target computing unit to other delay-sensitive tasks, Table 1 should include the time required for each computing unit of the vehicle-mounted device to perform the delay-sensitive task.

404. The scheduling module 102 determines a target processor for the delay-sensitive task 1.

It should be noted that the target computing unit determined for the delay-sensitive task 1 is the computing unit that performs the calculation corresponding to the delay-sensitive task. In some embodiments, the scheduling module 102 may first calculate the power consumption requirement of the delay-sensitive task 1 and estimate the delay required duration for executing the delay-sensitive task 1. By way of example, the power consumption requirements of delay-sensitive tasks and the length of delay requirements can also be recorded through the requirement information table. Table 2 is a possible implementation of the requirement information table.

Table 2

任务task	时延敏感任务1(Task 1)Delay-sensitive task 1
功耗需求Power requirements	W1W1
延迟要求时长Delay required	t1t1

In specific implementation, the scheduling module 102 may calculate a power consumption requirement W of a task (which may be a delay-sensitive task) according to formula (2).

W = n * w ^ (2)

In formula (2), n is the number of basic operations required to complete the task, and w ^ is the power consumption of the basic operation performed by the arithmetic unit. If the task is a neural network task (ie, a DNN task), n is the total multiplication and accumulation number of the neural network.

In addition, the delay required duration of a task (which may be a delay-sensitive task) refers to the deadline for completing the task at the current vehicle speed, that is, the minimum required time to complete the task. In the embodiment of the present invention, the processing time of each frame can be determined according to the current vehicle speed, and the time required to execute the task, that is, the required delay time of the task, can be further estimated based on the processing time of each frame. For example, when the mobile device is driving at a high speed, the input data required for task 1 is 10 images taken by camera 1 and camera 2. Then it can be estimated that the time required to perform task 1 is 10 * 33 = 330 ms, that is, the delay of task 1 The required time is 330ms.

Further, the scheduling module 102 traverses each computing unit 103 of the mobile device to determine whether the power consumption requirement of the delay-sensitive task falls within the target power consumption margin of the computing unit 103, and whether the computing unit 103 is able to meet the delay The delay of sensitive 1 requires the task to be completed within the time period. For example, after the scheduling module 102 determines the requirement information of the delay-sensitive task 1, it looks up the table 1 to determine the target processor of the delay-sensitive task 1. Specifically, for the arithmetic unit G0, it is first determined whether the time period T0 required by the arithmetic unit G0 to execute the delay-sensitive task 1 is less than or equal to the delay required time period t1 of the delay-sensitive task 1. If T0 is less than or equal to t1, then further judgment is made. Does the power consumption requirement of delay-sensitive task 1 fall within the range of (MW0-TW0) to (MW1-TW0), and if it falls within this range, then the computing unit G0 is determined as the target computing unit of delay-sensitive task 1. . Of course, if the time period T0 required by the processor G0 to execute the delay-sensitive task 1 is greater than the delay requirement time t1 of the delay-sensitive task 1, or the power consumption requirement of the delay-sensitive task 1 does not fall into (MW0-TW0) ~ ( MW0-TW0), it is further judged whether the arithmetic unit G1 can be used as the target arithmetic unit of the delay-sensitive task 1. For example, first, determine whether the duration T1 required by the processor G1 to execute the delay-sensitive task 1 is less than or equal to the delay request duration t1 of the delay-sensitive task 1. If T1 is less than or equal to t1, then further determine the delay-sensitive task 1. Whether the power consumption requirement falls within the range of (MW2-TW1) to (MW3-TW1), and if it falls within this range, then the computing unit G1 is determined as the target computing unit of the delay-sensitive task 1.

Judging whether the power consumption requirement of the delay-sensitive task falls within the target power consumption margin of an arithmetic unit 103 may be: if the power consumption requirement of the delay-sensitive task is 20W, the target power consumption margin range of the arithmetic unit 1 is 10W ～ 30W, the target power consumption margin of Operator 2 is 5W ～ 18W. It can be seen that the power consumption requirements of delay-sensitive tasks fall into the target power margin of Operator 1, which is the target of this delay-sensitive task. The computing unit is computing unit 1.

In some embodiments, if the target power consumption margin without the arithmetic unit 103 matches the power consumption requirement of the delay-sensitive task, the input data of the task can be split and multiple delay-executors can execute one delay-sensitive task. Calculation. Specifically, if the scheduling module 102 traverses all the arithmetic units, the power consumption requirements of the delay-sensitive tasks do not match the target power consumption margin ranges of all the arithmetic units of the mobile device, the scheduling module 102 may The input data is split into at least two sub-data blocks.

Further, for each of the at least one sub-data block, determining a power consumption required for an operator to calculate the sub-data block. Subsequently, the scheduling module 102 traverses all the arithmetic units to determine whether the power consumption matches the target power consumption margin range of the arithmetic units. If a target power consumption margin range of an operator of the mobile device matches the power consumption, it is determined that the operator is a target operator corresponding to the sub-data block. Finally, the target computing unit corresponding to all sub-data blocks is determined as the target computing unit corresponding to the delay-sensitive task.

For example, the input data for executing Task 2 is an image captured by camera 2 and camera 3, which can be divided according to the camera, the image captured by camera 2 is divided into sub data blocks 1, and the image captured by camera 3 is divided into sub data block 2. Assume that the target power consumption margin range of the computing unit G3 meets the power consumption required to calculate the sub-data block 1, and the target power consumption margin range of the computing unit G4 meets the power consumption required to calculate the sub-data block 2. Then the target corresponding to Task2 The arithmetic unit is G3, G4. In addition, the delay requirement duration corresponding to each sub-data block may be determined according to the data amount of each sub-data block, for example, according to the number of images in each sub-data block.

Of course, for special tasks, input data that does not allow the task to be split cannot be split, and the scheduling module 102 only considers whether the computing unit can meet the task's delay requirement when determining the target computing unit for the task. For example, if the input data of the delay-sensitive task 1 is not allowed to be split, the scheduling module 102 determines that the power requirement W1 of the delay-sensitive task 1 is not within the range of (MW0-TW0) to (MW1-TW0), and it is not ( In the range from MW2-TW1) to (MW3-TW1), if the time period T0 required by the processor G0 to execute the delay-sensitive task 1 is greater than the delay required time t1 of the delay-sensitive task 1, the processor G1 executes the delay-sensitive task. The required duration T1 of 1 is less than the delay required duration t1 of the delay-sensitive task 1, then it is determined that the arithmetic unit G1 is the target arithmetic unit of the delay-sensitive task 1.

The scheduling module executes steps 403 to 404 for each delay-sensitive task, so as to determine a target processor for each delay-sensitive task.

405. The scheduling module 102 assigns each delay-sensitive task to a corresponding target computing unit, and instructs the target computing unit corresponding to the delay-sensitive task to execute the delay-sensitive task within the delay-required duration of the delay-sensitive task.

In specific implementation, the scheduling module 102 generates a task list, which is used to record a target operator corresponding to each delay-sensitive task of the N delay-sensitive tasks and a delay requirement duration corresponding to each target operator. . The arithmetic unit 103 determines the delay-sensitive task to be executed according to the task list, and completes the delay-sensitive task to be executed within a delay required period of the delay-sensitive task to be executed. Table 3 below is one possible implementation of the task list.

table 3

任务task	目标运算器Target component	延迟要求时长Delay required
Task0Task0	G2G2	t0t0
Task1Task1	G1G1	t1t1
Task2Task2	G3、G4G3, G4	t2，t3t2, t3

Referring to Table 3, the arithmetic unit G2 needs to execute all calculations of the task Task0 within t0, and the arithmetic unit G1 needs to execute all calculations of the task Task1 within t1. It should be noted that if a delay-sensitive task is performed by multiple operators, that is, the data of a delay-sensitive task is split into multiple sub-data blocks, and different sub-data blocks are calculated by different operators, then Each sub-data block also has a corresponding delay request time, and the delay request time is determined according to the length of the sub-data block calculated by an operator. For example, Task2 in Table 3 is jointly executed by the arithmetic units G3 and G4, where G3 needs to calculate the sub-data block 1 of Task 2 within t2, and G4 needs to calculate the sub-data block 2 of Task 2 within t2.

406. The scheduling module 102 assigns a corresponding operator to the non-delay-sensitive tasks.

It should be noted that, after the scheduling module 102 assigns an arithmetic unit to each delay-sensitive task, it can also assign an arithmetic unit to the non-delay-sensitive task. Specifically, for each of the M tasks except the N delay-sensitive tasks (that is, the non-delay-sensitive tasks), the power consumption requirement of the task is calculated, and the target power consumption margin is calculated. An operator whose range matches the power consumption requirement of the task and is capable of executing the task within the delay request duration of the task is determined to be the target operator to execute the task, and the task is assigned to the task Target operator for delay-sensitive tasks.

Wherein, the power consumption requirement of the task is the power consumption required to execute the task, and the power consumption requirement of the non-delay-sensitive task can be calculated by referring to the above formula 2. The delay request duration of the task is the time required to execute the task. Similarly, the delay request duration of a non-delay-sensitive task can be estimated according to the amount of input data of the non-delay-sensitive task. For details, refer to step 402 above. A description of how to delay the delay-sensitive task requires a time estimation, which is not described in the embodiment of the present invention.

It should be noted that if the power consumption requirements of non-delay-sensitive tasks do not match the target power consumption range of all computing units, non-delay-sensitive tasks can be assigned to be within the delay requirements of non-delay-sensitive tasks. An operator that finishes the non-delay-sensitive task.

Of course, instead of allocating an operator for non-delay-sensitive tasks, you can wait until steps 101 to 405 are performed again to combine the input data of the non-delay-sensitive task with the data of other non-delay-sensitive tasks. The merged data determines the operator. However, it is necessary to ensure that the interval between the current time (the time at which step 406 is performed this time) and the time at which step 405 is performed again cannot exceed the delay request duration of the delay-sensitive task.

In a specific implementation, the scheduling module 102 may combine the input data of at least two non-delay-sensitive tasks into one data block.

Further, it is determined that an operator of the mobile device calculates the power consumption required for the data block, and if the target power consumption margin range of an operator of the mobile device matches the power consumption, then the A data block is allocated to the arithmetic unit, and the arithmetic unit is instructed to finish calculating the data block within a required delay time corresponding to the data block.

The delay requirement of the data block is an estimated time required to complete the calculation of the data block, and the delay requirement of the data block may be estimated according to the data amount of the data block. For example, the merged data includes 10 images. When the mobile device is driving at a high speed, the processing time of each image cannot exceed 33ms. Then it can be estimated that the required length of the data block is 10 * 33 = 330ms, that is, the delay of the data block. The required time is 330ms.

For example, suppose that the data to be calculated by executing Task1 is the image taken by camera 3, and the data to be calculated by executing Task2 is the image taken by camera 2. Task1 and Task2 are both non-delay-sensitive tasks, and the data corresponding to Task1 and Task2 are combined into A data block (ie, images taken by cameras 3 and 2). If the power consumption required to calculate the images taken by cameras 3 and 2 falls within the target power consumption margin of the processor G5, instruct the processor G5 to execute Task1 And Task2 to calculate the images taken by cameras 3 and 2.

It should be noted that merging the input data of non-delay-sensitive tasks requires the computational resources of the operator, and the merged data block should be as small as possible. For example, there are three non-delay-sensitive tasks Task4, Task5, and Task6. The data of two of the non-delay-sensitive tasks Task4 and Task5 can be merged first, and an operator is assigned to the merged data. When waiting for new computing resources to be released, an arithmetic unit can be assigned to Task6. For example, some arithmetic units have remaining power consumption after performing calculations, and Task6 can be assigned to the arithmetic unit.

In addition, no matter it is a delay-sensitive task or a non-delay-sensitive task, once the task is assigned to the computing unit, the computing unit must output the calculation result within the required delay time of the task.

In some embodiments, the scheduling module 102 may perform task allocation with reference to the method shown in FIG. 7. Specifically, referring to FIG. 7, the following steps are included:

In step S1, the scheduling module 102 sorts the tasks to be executed according to the priority order from high to low: Task0, Task1, and Task2.

In step S2, the scheduling module 102 takes out the task Task0 with the highest priority level, and calculates the power consumption requirement w0 and the delay requirement duration t0 of the task0.

In step S3, the scheduling module 102 obtains the target power consumption margin range of each operator and the time required for each operator to execute Task0.

In step S4, it is determined whether the power consumption requirement of Task0 falls within the target power consumption margin range of the computing unit, and whether the time period for which the computing unit executes Task0 is less than or equal to the delay required time period of Task0.

Step S5: If Task0 falls within a target power consumption margin range of a certain computing unit, Task0 is assigned to the computing unit, indicating that the computing unit must return the calculation result of Task0 at time t (current time) + t0.

Step S6: If Task0 does not fall within the target power consumption margin range of all the arithmetic units, the data corresponding to Task0 is split into multiple sub-data blocks, and the sub-data blocks with a large amount of calculation are preferentially allocated to the arithmetic units with more resources .

The following describes the embodiments of the present invention with reference to the accompanying drawings to provide delay-sensitive task scheduling. Referring to FIG. 8, the delay-sensitive tasks that need to be executed are Task0, Task1, and Task2 in order of priority level from high to low. The power requirements of Task0, Task1, and Task2 are 10W, 20W, and 30W, respectively. The vehicle-mounted device has two arithmetic units G0 and G1. The maximum power consumption limit of the arithmetic unit G0 is 40W, the target power consumption is 30W, the maximum power consumption limit of the arithmetic unit G1 is 60W, and the target power consumption is 40W. It is assumed that G0 and G1 both meet the delay requirements of Task0, Task1, and Task2. In the embodiment of the present invention, an arithmetic unit may be allocated to each task in order from the priority level.

For example, referring to FIG. 8, in the initial state, the occupied power consumption of the arithmetic unit G0 is 10W, the target power consumption margin is 20W, the arithmetic unit G1 is idle, and the target power consumption margin is 40W. First, the target power consumption margin of the arithmetic unit G0 is 20W, and the power consumption requirement of Task0 is 10W. The target power consumption margin of the arithmetic unit G0 can meet the 10W power consumption requirement of Task0. Therefore, Task0 can be assigned to the arithmetic unit. G0. At this time, the target power consumption margin of the arithmetic unit G0 becomes 10W, and the target power consumption margin of the arithmetic unit G1 is still 40W.

Subsequently, the target power consumption margin of the computing unit G1 can meet the 10W power consumption requirement of Task1, and Task1 can also be assigned to the computing unit G1. At this time, the target power consumption margin of the computing unit G0 is still 10W, and the computing unit G1 The target power consumption margin becomes 20W.

Finally, the power consumption requirement of Task2 is greater than the target power consumption margin of the arithmetic unit G0 of 10W, and greater than the target power consumption margin of the arithmetic unit G0 of 20W, so the data of Task2 can be split into two parts. A part of the power consumption requirement is 10W, which matches the target power consumption margin of the arithmetic unit G0 of 10W, so this part of the data can be allocated to the arithmetic unit G0 for calculation. The power consumption requirement of the other part is 20W, which matches the target power consumption margin of 20W of the arithmetic unit G1. Then, this part of data can be allocated to the arithmetic unit G1 for calculation.

The following describes embodiments of the present invention with reference to the drawings to provide scheduling for non-delay-sensitive tasks. With reference to FIG. 9, the delay-sensitive tasks that need to be executed are Task0, Task1, and Task2 in order of priority level from high to low. The power requirements of Task0, Task1, and Task2 are 10W, 20W, and 30W, respectively. The vehicle-mounted device has two processors, G0 and G1. The maximum power consumption limit of the processor G0 is 40W, the target power consumption is 30W, the maximum power consumption limit of the processor G1 is 60W, and the target power consumption is 40W. In the initial state, the occupant power consumption of the arithmetic unit G0 is 10W, the target power consumption margin is 20W, the arithmetic unit G1 is idle, and the target power consumption margin is 40W.

Referring to FIG. 9, for non-delay-sensitive tasks, tasks with lower delay requirements may be merged, and the merged tasks may be allocated to a processor. For example, the data of Task0 and Task1 can be merged into one task. The power consumption requirement of the merged data is 10W + 20W = 30W, which matches the target power consumption margin of the processor G1. Therefore, the merged task can be allocated. To the arithmetic unit G1, the target power consumption margin of the arithmetic unit G1 is 10W, and the target power consumption margin of the arithmetic unit G0 is 20W. Although the target power consumption margins of the arithmetic units G1 and G0 do not meet the power consumption requirements of Task2, since Task2 is also a non-delay-sensitive task, it is not necessary to temporarily assign an arithmetic unit to Task2, and wait until there are sufficient computing resources. Task2 assigns a component.

An embodiment of the present invention provides a device, which may be a scheduling module in an in-vehicle device according to an embodiment of the present invention, such as the scheduling module 102 of the in-vehicle device 10 shown in FIG. 2. In a case where each functional module is divided corresponding to each function, FIG. 10 shows a possible structural diagram of the foregoing communication device. As shown in FIG. 10, the device includes a determination unit 1001, a calculation unit 1002, and an allocation unit 1003.

A determining unit 1001 is configured to support the apparatus to perform step 401, step 402, and / or other processes for the technology described herein.

A computing unit 1002, configured to support the apparatus to perform step 403 in the above embodiments, and / or other processes used in the technology described herein;

An allocating unit 1003, configured to support the apparatus to perform step 405 in the above embodiments, and / or other processes for the technology described herein;

It should be noted that all relevant content of each step involved in the foregoing method embodiments can be referred to the functional description of the corresponding functional module, and will not be repeated here.

Exemplarily, when an integrated unit is used, a schematic structural diagram of a device provided in an embodiment of the present application is shown in FIG. 6. In FIG. 11, the apparatus includes a processing module 1101 and a communication module 1102. The processing module 1101 is used to control and manage the actions of the device, for example, to execute the steps performed by the determination unit 1001, the calculation unit 1002, and the distribution unit 1003, and / or other processes for performing the techniques described herein. The communication module 1102 is used to support interaction between the device and other devices. As shown in FIG. 11, the device may further include a storage module 1103. The storage module 1103 is configured to store program code and data of the device.

The task scheduling method provided by the embodiment of the present invention may be applied to the method shown in FIG. 12, and the device may be the scheduling module 102 according to the embodiment of the present invention. As shown in FIG. 12, the device may include at least one processor 1201, a memory 1202, a transceiver 1203, and a communication bus 1204.

The following describes each component of the device in detail with reference to FIG. 12:

The processor 1201 is a control center of the device, and may be a processor or a collective name of a plurality of processing elements. For example, the processor 120201 is a central processing unit (CPU), or may be an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement an embodiment of the present invention. For example, one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (Field Programmable Gate Arrays, FPGAs).

The processor 1201 can execute various functions of the device by running or executing a software program stored in the memory 1202 and calling data stored in the memory 1202.

In a specific implementation, as an embodiment, the processor 1201 may include one or more CPUs, such as CPU0 and CPU1 shown in FIG. 12.

In a specific implementation, as an embodiment, the device may include multiple processors, such as the processor 1201 and the processor 1205 shown in FIG. 12. Each of these processors can be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). A processor herein may refer to one or more devices, circuits, and / or processing cores for processing data (such as computer program instructions).

The memory 1202 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (random access memory, RAM), or other types that can store information and instructions The dynamic storage device can also be Electrically Erasable Programmable Read-Only Memory (EEPROM), Compact Disc (Read-Only Memory, CD-ROM) or other optical disk storage, optical disk storage (Including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), disk storage media or other magnetic storage devices, or can be used to carry or store desired program code in the form of instructions or data structures and can be used by a computer Any other media accessed, but not limited to this. The memory 1202 may exist independently, and is connected to the processor 1201 through a communication bus 1204. The memory 1202 may also be integrated with the processor 1201.

The memory 1202 is configured to store a software program that executes the solution of the present invention, and is controlled and executed by the processor 1201.

The transceiver 1203 uses any device such as a transceiver to communicate with other nodes in the system shown in FIG. 1, such as other relay nodes, core nodes, target nodes, or source nodes. Or used to implement communication between the device and the base station in FIG. 1. It can also be used to communicate with communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), and so on. The transceiver 1203 may include a receiving unit to implement a receiving function, and a transmitting unit to implement a transmitting function.

The communication bus 1204 may be an Industry Standard Architecture (ISA) bus, an External Device Component (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only a thick line is used in FIG. 12, but it does not mean that there is only one bus or one type of bus.

The device structure shown in FIG. 12 does not constitute a limitation on the device, and may include more or fewer parts than shown, or some parts may be combined, or different parts may be arranged.

In the embodiment of the present invention, the processor 1201 may run code in the memory 1202 to execute the methods shown in FIG. 4 and FIG. 7 of the embodiment of the present invention.

Through the description of the above embodiments, those skilled in the art can clearly understand that, for the convenience and brevity of the description, only the division of the above functional modules is used as an example. In practical applications, the above functions can be allocated according to needs. It is completed by different functional modules, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above.

In the several embodiments provided in this application, it should be understood that the disclosed database access device and method can be implemented in other ways. For example, the embodiments of the database access device described above are merely schematic. For example, the division of the modules or units is only a logical function division. In actual implementation, there may be another division manner, such as multiple units or units. Components can be combined or integrated into another device, or some features can be ignored or not implemented. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection of the database access device or unit through some interfaces, which may be electrical, mechanical or other forms.

The unit described as a separate component may or may not be physically separated, and the component displayed as a unit may be a physical unit or multiple physical units, that is, may be located in one place, or may be distributed to multiple different places. . Some or all of the units may be selected according to actual needs to achieve the objective of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each of the units may exist separately physically, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional unit.

When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a readable storage medium. Based on such an understanding, the technical solutions of the embodiments of the present application essentially or partly contribute to the existing technology or all or part of the technical solutions may be embodied in the form of a software product, which is stored in a storage medium The instructions include a number of instructions for causing a device (which can be a single-chip microcomputer, a chip, or the like) or a processor to execute all or part of the steps of the method described in the embodiments of the present application. The foregoing storage medium includes various media that can store program codes, such as a U disk, a mobile hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.

The above is only a specific implementation of this application, but the scope of protection of this application is not limited to this. Any changes or replacements within the technical scope disclosed in this application shall be covered by the scope of protection of this application. . Therefore, the protection scope of this application shall be subject to the protection scope of the claims.

Claims

A task scheduling method, comprising:

Determine the motion state of the mobile device, determine M tasks to be executed according to the motion state of the mobile device, and determine N delay-sensitive tasks among the M tasks; the M is an integer greater than or equal to 1, N is an integer less than or equal to M;

For each delay-sensitive task of the N delay-sensitive tasks, calculate the power consumption requirement of the delay-sensitive task, and match the target power consumption margin range with the power consumption requirement of the delay-sensitive task. And the processor capable of executing the delay-sensitive task within the delay-requested time of the delay-sensitive task is determined as the target processor to execute the delay-sensitive task, and the delay-sensitive task is allocated to all Describe the target computing unit for delay-sensitive tasks;

The target power consumption margin range is determined according to the target power consumption range of the mobile device's computing unit and the current power consumption of the computing unit; the power consumption requirement of the delay-sensitive task is to execute the time The power consumption required for a delay-sensitive task; the delay requirement of the delay-sensitive task is the time required to execute the delay-sensitive task.
The method according to claim 1, wherein the determining N delay-sensitive tasks out of the M tasks specifically comprises:

Determining a delay sensitivity of each of the M tasks according to the motion state;

Determining a priority level of each of the M tasks according to delay sensitivity of each of the M tasks; the priority level is proportional to the delay sensitivity;

N tasks with a higher priority among the M tasks are determined as the N delay-sensitive tasks.
The method according to claim 1 or 2, wherein the calculating the power consumption requirement of the delay-sensitive task matches the target power consumption margin range with the power consumption requirement of the delay-sensitive task and The processor capable of determining that the delay-sensitive task has been executed within the delay-requested duration of the delay-sensitive task as a target processor that specifically performs the delay-sensitive task includes:

Calculating power consumption requirements for the delay-sensitive task;

Determining, for each operator of the mobile device, a duration for which the operator performs the delay-sensitive task and a target power consumption margin range of the operator;

Determine whether the power consumption requirement of the delay-sensitive task falls within the target power consumption margin of the computing unit and whether the time duration for which the computing unit executes the delay-sensitive task does not exceed the delay-sensitive task Delay required;

If the power consumption requirement of the delay-sensitive task falls within the target power consumption margin of the computing unit and the time for which the computing unit executes the delay-sensitive task does not exceed the delay requirement of the delay-sensitive task For the duration, it is determined that the operator is a target operator of the delay-sensitive task.
The method according to claim 3, further comprising:

If the power consumption requirements of the delay-sensitive task do not match the target power consumption margin ranges of all the operators of the mobile device, the input data of the delay-sensitive task is split into at least two sub-data blocks ;

For each of the at least one sub-data block, determining a power consumption required to calculate the sub-data block, if a target power consumption margin range of an operator of the mobile device and the power consumption If they match, it is determined that the operator is a target operator corresponding to the sub-data block;

A target operator corresponding to all sub-data blocks is determined as a target operator corresponding to the delay-sensitive task.
The method according to any one of claims 1-4, wherein the target computing unit that assigns the delay-sensitive task to the delay-sensitive task specifically includes:

Calculating the execution time of the N delay-sensitive tasks, and determining the delay required duration of the execution time of the delay-sensitive tasks according to the execution time of the delay-sensitive tasks;

Generate a task list, where the task list is used to record a target operator corresponding to each delay-sensitive task in the N delay-sensitive tasks and a delay requirement duration corresponding to each target operator, so that the mobile device ’s The arithmetic unit determines a delay-sensitive task to be executed according to the task list and finishes executing the delay-sensitive task to be executed within a delay required time of the delay-sensitive task to be executed.
The method according to claim 2, further comprising:

After assigning each delay-sensitive task of the N delay-sensitive tasks to a corresponding target computing unit, for each task of the M tasks except the N delay-sensitive tasks, calculate The power consumption requirement of the task is determined by an operator that matches the target power consumption margin range with the power consumption requirement of the task and is capable of executing the task within the delay request duration of the task as being performed by the task. A target operator, and assigning the task to a target operator of the delay-sensitive task;

Wherein, the power consumption requirement of the task is the power consumption required to execute the task, and the delay required time of the task is the time required to execute the task.
The method according to claim 3, wherein a duration T of the delay-sensitive task performed by the arithmetic unit satisfies:

Among them, Nc and Nm are the number of basic computing units inside the arithmetic unit, f 1 is the frequency of the core on the arithmetic unit; f 2 is the frequency of the memory of the arithmetic unit, and C i is the double precision required to perform the delay-sensitive task. The number of floating-point operations, m j is the memory access frequency required to complete the delay-sensitive task; the memory access frequency is the number of times that the memory is accessed per second.
The method according to any one of claims 1 to 7, characterized in that the power consumption requirement W of the delay-sensitive task satisfies: W = n * w ^, where n finishes executing the delay-sensitive task. The required basic operand, w ^ is the power consumption of a basic operation performed by the arithmetic unit.
A device, comprising:

A determining unit, configured to determine a motion state of the mobile device, determine M tasks to be performed according to the motion state of the mobile device, and determine N delay-sensitive tasks among the M tasks; the M is greater than or equal to 1 An integer of N, the N is an integer less than or equal to M;

A computing unit, configured to calculate a power consumption requirement of the delay-sensitive tasks for each of the N delay-sensitive tasks;

The determining unit is further configured to match the target power consumption margin range with the power consumption requirement of the delay-sensitive task and be able to execute the delay-sensitive task within the delay-required duration of the delay-sensitive task. Is determined to be the target processor for performing the delay-sensitive task;

An allocation unit, configured to allocate the delay-sensitive task to a target operator of the delay-sensitive task;

The target power consumption margin range is determined according to the target power consumption range of the mobile device's computing unit and the current power consumption of the computing unit; the power consumption requirement of the delay-sensitive task is to execute the time The power consumption required for a delay-sensitive task; the delay requirement of the delay-sensitive task is the time required to execute the delay-sensitive task.
The device according to claim 9, wherein the determining unit is specifically configured to determine a delay sensitivity of each of the M tasks according to the motion state;

Determining the priority level of each of the M tasks according to the delay sensitivity of each of the M tasks; the priority level is proportional to the delay sensitivity;

N tasks with a higher priority among the M tasks are determined as the N delay-sensitive tasks.
The device according to claim 9 or 10, wherein:

The calculation unit is specifically configured to calculate a power consumption requirement of the delay-sensitive task; and for each operator of the mobile device, determine a time period for the operator to execute the delay-sensitive task and the operator Target power consumption margin range;

The determining unit is specifically configured to determine whether the power consumption requirement of the delay-sensitive task falls within the target power consumption margin of the computing unit and whether the time duration for which the computing unit executes the delay-sensitive task has not yet passed. Exceeds the delay requirement duration of the delay-sensitive task; if the power consumption requirement of the delay-sensitive task falls within the target power consumption margin of the processor and the processor executes the delay-sensitive task If the duration does not exceed the delay required duration of the delay-sensitive task, it is determined that the computing unit is a target computing unit of the delay-sensitive task.
The apparatus according to claim 11, wherein the determining unit is further configured to: if a power consumption requirement of the delay-sensitive task does not exceed a target power consumption margin range of all the arithmetic units of the mobile device Matching, the input data of the delay-sensitive task is split into at least two sub-data blocks;

For each of the at least one sub-data block, determining a power consumption required to calculate the sub-data block, if a target power consumption margin range of an operator of the mobile device and the power consumption If they match, it is determined that the operator is a target operator corresponding to the sub-data block;

A target operator corresponding to all sub-data blocks is determined as a target operator corresponding to the delay-sensitive task.
The apparatus according to any one of claims 9 to 12, wherein the allocation unit is specifically configured to calculate an execution time of the N delay-sensitive tasks, and determine the execution time according to the execution time of the delay-sensitive tasks. A delay request duration of the execution duration of the delay-sensitive task;

Generate a task list, where the task list is used to record a target operator corresponding to each delay-sensitive task in the N delay-sensitive tasks and a delay requirement duration corresponding to each target operator, so that the mobile device ’s The arithmetic unit determines a delay-sensitive task to be executed according to the task list and finishes executing the delay-sensitive task to be executed within a delay required time of the delay-sensitive task to be executed.
The device according to claim 10, wherein:

The computing unit is further configured to, after allocating each delay-sensitive task of the N delay-sensitive tasks to a corresponding target computing unit, remove the N delay-sensitive tasks from the M tasks For each external task, calculate the power consumption requirements of the task;

The allocating unit is further configured to: an operator that matches the target power consumption margin range with the power consumption demand of the task and is capable of executing the task within the delay request duration of the task is determined to execute the task A target operator, and assigning the task to a target operator of the delay-sensitive task;

Wherein, the power consumption requirement of the task is the power consumption required to execute the task, and the delay required time of the task is the time required to execute the task.
The apparatus according to claim 11, wherein a duration T of the delay-sensitive task performed by the computing unit satisfies:

Among them, Nc and Nm are the number of basic computing units inside the arithmetic unit, f 1 is the frequency of the core on the arithmetic unit; f 2 is the frequency of the memory of the arithmetic unit, and C i is the double precision required to perform the delay-sensitive task. Number of floating-point operations, m j is the memory access frequency required to complete the delay-sensitive task; the memory access frequency is the number of times the memory is accessed per second.
The apparatus according to any one of claims 9 to 15, wherein the power consumption requirement W of the delay-sensitive task satisfies: W = n * w ^, where n finishes executing the delay-sensitive task Of the required basic operands, w ^ is the power consumption of a basic operation performed by the arithmetic unit.