KR102542247B1 - Task calculation amount determining method and apparatus, computer readable storage medium and computer program - Google Patents

Abstract

In the present disclosure, a task calculation amount determination method, apparatus, and electronic device are provided, and relate to the field of calculation measurement evaluation technology. Methods include: controlling the target task to run in the target CPU core; Acquiring, by the target task, a CPU time slice occupancy rate in a task running period, wherein the task running period is a period from the start of running of the target task to the end of running of the target task. doing; and determining an operation amount of the target task according to the DMIPS of the target CPU core and the CPU time slice occupancy rate. includes Because of the use of the above technical means, the task calculation amount determined in the present disclosure is more true, and more accurately characterizes the CPU calculation power to be occupied by the task. Therefore, when distributing to tasks, based on the task calculation amount determined in the present disclosure, a more valuable reference can be provided for selecting the hardware type of the device.

Description

Task operation amount determination method, device, computer readable storage medium and computer program

TECHNICAL FIELD The present disclosure relates to the field of computer technology, particularly to the field of calculation and evaluation technology, and specifically to a method, apparatus, and electronic device for determining a task calculation amount.

As computer technology develops, the depth of a task gradually increases, and the computational complexity and amount of computation of the task also increase accordingly. Before distributing tasks in the device, it is necessary to consider whether or not the CPU (Central Processing Unit, central processor) computing power (i.e., computing power) of the device satisfies the computing power of the task. Therefore, before distributing tasks, the computing power for the task measurement evaluation must be performed. Currently, MACCs (multiply-accumulate operations per second) and FLOPS (floating-point operations per second) are used to characterize the amount of operation of a task, and they determine the amount of CPU computing power that a task must occupy. Not all can be characterized.

The present disclosure provides a method, apparatus, and electronic device for determining a task computation amount, and solves the technical problem that existing methods for characterizing a task computation amount cannot characterize CPU computation power to be occupied by a task.

In order to solve the above technical problem, the present disclosure is implemented as follows.

In a first aspect, the present disclosure provides a method for determining a task computation amount, the method comprising:

controlling the target task to run in the target CPU core;

Acquiring, by the target task, a CPU time slice occupancy rate in a task running period, wherein the task running period is a period from the start of running of the target task to the end of running of the target task. doing; and

determining an operation amount of the target task according to the DMIPS of the target CPU core and the CPU time slice occupancy rate; includes

Because of the use of the above technical means, the task calculation amount determined in the present disclosure is more true, and more accurately characterizes the CPU calculation power to be occupied by the task. Therefore, when distributing to tasks, based on the task calculation amount determined in the present disclosure, a more valuable reference can be provided for selecting the hardware type of the device.

Optionally, the step of obtaining, by the target task, a CPU time slice occupancy in a task running period:

obtaining, by the target CPU core, a CPU time slice in the task running period;

obtaining a CPU time slice occupied by the target task; and

determining, by the target CPU core, a CPU time slice occupied by the target task in the task running period according to the CPU time slice in the task running period and the CPU time slice occupied by the target task; includes

In the above implementation, the computational amount measurement and evaluation of the target task can be accurately evaluated by determining the CPU time slice occupancy of the target task in the task running period.

Optionally, the amount of calculation of the target task includes the amount of calculation of an image detection algorithm, the amount of calculation of a voice recognition algorithm, or the amount of calculation of a planning and control algorithm.

In the method of determining the task computation amount based on DMIPS and CPU time slice occupancy provided in the present disclosure, the CPU time slice occupancy rate can relatively truthfully and accurately characterize the CPU computing power occupancy required for task operation, so that continuous tasks Not only can it be applied, but it can also be applied to non-continuous tasks, and has universality.

Optionally, the amount of computation of the target task is the amount of computation of the image detection algorithm at a target frame rate;

The step of determining the operation amount of the target task according to the DMIPS and CPU time slice occupancy of the target CPU core:

obtaining the number of image frames processed by the image detection algorithm within the task operation period;

determining a single frame operation amount of the image detection algorithm according to the DMIPS of the target CPU core, the CPU time slice occupancy rate, the task running period, and the number of image frames; and

determining, by the image detection algorithm, an operation amount at the target frame rate according to the single frame operation amount and the target frame rate; includes

In the above implementation, it is advantageous to accurately measure and evaluate the performance of the image detection algorithm through determining the amount of computation at the target frame rate of the image detection algorithm, and therefore, when distributing the image detection algorithm, the determined image Based on the amount of computation at the target frame rate, the detection algorithm can provide a more valuable reference for device hardware type selection, thus facilitating the application of the image detection algorithm.

Optionally, the method:

The method may further include controlling an operating frequency of the target CPU core to be at a preset frequency while the target task is running in the target CPU core.

In the above practical method, by controlling the operating frequency of the target CPU core during the operation of the target task, a relatively stable operating environment is provided for the operation of the target task, thereby ensuring that the task operation amount determined in the present disclosure is more accurate. can

Optionally, controlling the target task to run in the target CPU core comprises:

and controlling the target task to run among the target CPU cores with a preset priority.

In the implementation method, by controlling the operating frequency of the target CPU core during the operation of the target task, a relatively stable operating environment is provided for the operation of the target task, thereby ensuring that the task calculation amount determined in the present disclosure is more accurate. can

Optionally, before controlling the target task to run in a target CPU core, the method comprises:

The method further includes receiving the target task transmitted from the cloud server and uploading the target task to the cloud server in advance by a user.

Optionally, after computing the amount of operations of the target task, the method:

The method may further include reporting the amount of computation of the target task to the cloud server.

In the implementation mode, by opening an open measurement evaluation environment, task design, task training, task distribution, and task calculation evaluation are made to form a closed loop. Accordingly, after the task design personnel complete the task design and task training, it is possible to obtain the evaluation result of measuring the amount of task operation in time, so that the task design personnel can easily proceed with the customized optimization for the task, so that the task design person can determine the consumption state of the task operation amount. It is advantageous to take control in time.

In a second aspect, the present disclosure provides an apparatus for determining a task computation amount, the apparatus comprising:

A first control module for controlling the target task to run in the target CPU core;

An acquisition module for obtaining a CPU time slice occupancy rate of the target task in a task running period, wherein the task running period is a period from when the target task starts running until the target task ends running , acquisition module; and

a determining module configured to determine an amount of operation of the target task according to the DMIPS of the target CPU core and the CPU time slice occupancy rate; includes

Optionally, the acquisition module:

a first obtaining submodule configured for the target CPU core to obtain a CPU time slice in the task running period;

a second acquiring submodule configured to acquire a CPU time slice occupied by the target task; and

A first determination submodule configured to determine, according to the CPU time slice of the target CPU core in the task running period and the CPU time slice occupied by the target task, the CPU time slice occupancy of the target task in the task running period ; includes

Optionally, the amount of calculation of the target task includes the amount of calculation of an image detection algorithm, the amount of calculation of a voice recognition algorithm, or the amount of calculation of a planning and control algorithm.

Optionally, the amount of computation of the target task is the amount of computation of the image detection algorithm at a target frame rate;

The confirmation module:

a second acquiring submodule configured to acquire the number of image frames processed by the image detecting algorithm within the task running period;

a second determination submodule configured to determine a single frame operation amount of the image detection algorithm according to the DMIPS of the target CPU core, the CPU time slice occupancy rate, the task running period and the number of image frames; and

a third determining submodule configured to determine, according to the single frame operation amount and the target frame rate, an operation amount at the target frame rate of the image detecting algorithm; includes

Optionally, the device:

A second control module configured to control an operating frequency of the target CPU core to be at a predetermined frequency while the target task is running in the target CPU core is included.

Optionally, the first control module specifically:

This is to control the target task to run among the target CPU cores with a preset priority.

Optionally, the device:

A receiving module for receiving the target task transmitted from the cloud server, wherein the target task has been previously uploaded to the cloud server by a user.

Optionally, the device:

and a reporting module for reporting the amount of computation of the target task to the cloud server.

In a third aspect, the present disclosure provides an electronic device, wherein the electronic device:

at least one processor; and

a memory communicatively coupled to the at least one processor, including, among them,

Instructions executable by the at least one processor are stored in the memory, and the instructions are executed by the at least one processor to cause the at least one processor to perform any one method according to the first aspect. let it do

In a fourth aspect, the present disclosure provides a non-transitory computer-readable storage medium storing computer instructions, the computer instructions for causing the computer to execute any one method according to the first aspect.

The embodiments in the above disclosure have the following advantages or advantageous effects.

In the present disclosure, when it is necessary to evaluate the amount of computation for a target task, the CPU time slice occupied by the target task during the running process is obtained by controlling the target task to run in the target CPU core, that is, the target task The amount of computation of the target task is determined according to the DMIPS of the CPU core and the CPU time slice occupied by the target task. Because of the use of the above technical means, the task calculation amount determined in the present disclosure is more true, and more accurately characterizes the CPU calculation power to be occupied by the task. Therefore, when distributing to tasks, based on the task calculation amount determined in the present disclosure, a more valuable reference can be provided for selecting the hardware type of the device.

Other effects provided by the above-described optional method will be further described by combining specific examples in the following specification.

The accompanying drawings are for a better understanding of the present solution and do not constitute limitations on the present disclosure. among them:
1 is a flowchart of a method for determining a task computation amount provided in an embodiment of the present disclosure.
2 is an exemplary processing flow chart of a method for determining a task computation amount provided by an embodiment of the present disclosure.
3 is an exemplary distribution diagram of a system for measuring and evaluating a task calculation amount provided by an embodiment of the present disclosure.
4 is an exemplary view of a task measurement evaluation closed loop provided by an embodiment of the present disclosure.
5 is an exemplary structural diagram of an apparatus for determining a task calculation amount provided in an embodiment of the present disclosure.
6 is a block diagram of an electronic device implementing a method for determining a task computation amount according to an embodiment of the present disclosure.

It should be understood that exemplary embodiments of the present disclosure are described below in connection with the drawings, and various details of the embodiments of the present disclosure are included for better understanding, which are only exemplary. Accordingly, it should be understood that those skilled in the art may make various changes and modifications to the embodiments described herein without departing from the scope and spirit of the present disclosure. Equally, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

Currently, the amount of task operations is usually characterized using MACCs and FLOPS, MACCs are for characterizing the number of theoretical multiplication accumulations, and FLOPS is for characterizing the number of floating point operations on the GPU (Graphics Processing Unit, graphics processor) side. it is for

As the depth of the task gradually increases, the computational complexity and amount of computation of the task also increase accordingly, and the CPU computational power to be occupied by the task in the course of driving also gradually increases. For example, with the development of artificial intelligence (Artificial Intelligence, abbreviated as AI), depth learning is gradually becoming a mainstream method in each field, and along with ejectors of residual networks (abbreviated as Residual Networks, ResNets), neural networks The depth of is gradually increased, and the computational complexity and amount of computation are also increased accordingly. Therefore, in order to ensure that the task can run normally in the device, before distributing the task among the devices, it is necessary to measure and evaluate whether the CPU computing power of the device can bear the load of the task's computing power, or It is necessary to measure and evaluate whether the CPU processing power satisfies the processing amount of the task.

However, none of the above-mentioned methods for characterizing the amount of computation of a task can characterize the amount of computation to be occupied when a task runs in the CPU, and it is currently impossible to truly and accurately determine the amount of computation to be occupied when a task runs in the CPU. Therefore, it is not possible to provide guidance in terms of selection of the hardware type of the device, suitability of the task, and the like.

Based on this, the present disclosure provides a task operation amount determination method for determining the amount of computation to be occupied by the target task while running in the CPU or determining the amount of computational power of the CPU to be occupied by the target task during the running process. In the present disclosure, the task calculation amount may be determined through a measurement evaluation device (eg, a terminal device such as a computer, a mobile phone, and a vehicle-mounted computer).

In the present disclosure, a target task means a task that can run in the CPU of the measurement and evaluation device, and can be a thread or a process when the target task runs in the CPU. The amount of computation of the target task means CPU computational power to be occupied by the target task during the running process.

As shown in Fig. 1, the method for determining the amount of operation of a task includes the following steps.

Step 101: Control the target task to run in the target CPU core.

Currently, CPUs generally use a multi-core (ie, physical core) architecture, and some CPUs have large and small cores. In the above step, the target task is controlled to run in the target CPU core, and therefore, the calculation amount of the target task only needs to consider the calculation power of the target CPU core, so that the calculation amount of the target task is more simply and more accurately determined. can do.

In the present disclosure, before the target task runs, the CPU core may be bound to the target task in advance so that the target task runs only on the target CPU core and not on other CPU cores. A method of binding a CPU core for a target task may implement binding of a CPU core through, for example, an Android system through the command “echo “0” > /sys/devices/system/cpu/cpuN/ online”, Among them, N denotes a CPU core to be prohibited.

In the present disclosure, the target task may be an image detection task using an image detection algorithm, and the calculation amount of the target task may be the calculation amount of the image detection algorithm. More specifically, the calculation amount of the target task may be an image detection task using an image detection algorithm. It may be an amount of computation that proceeds with detection. In a specific application, the image detection task may include, for example, face detection, face recognition, gesture recognition, lane detection, pedestrian vehicle detection, text recognition, object classification, object segmentation, and the like.

The target task may be a voice recognition task using a voice recognition algorithm, and the amount of calculation of the target task may be the amount of calculation of the voice recognition algorithm. More specifically, the amount of calculation of the target task is to perform voice recognition using a voice recognition algorithm. It may be an amount of operation. In specific applications, speech recognition tasks may include, for example, speech wake-up, semantic understanding, language translation, speech synthesis, and the like.

The target task may be a task using a planning and control algorithm, and the amount of computation of the target task may be the amount of calculation of the planning and control algorithm. In a specific application, the tasks of the planning and control algorithms may include, for example, path planning, trajectory prediction, learning enhancement, MPC (Model Predictive Control), etc. tasks.

Step 102: The target task acquires a CPU time slice occupancy rate in a task running period, and the task running period is a period from when the target task starts running to when the target task ends running.

The current working system (such as Windows, Linux, Mac OS X, etc.) allows multiple processes to run at the same time, and therefore, during the running period of the task, the target CPU core other than the target task running among the target CPU cores. During this time, other processes may possibly be running. Also, while the target task is running, it may not be running continuously, but may be running intersectively with other processes. Therefore, in the period during which the task runs, the target CPU core's computing power is not entirely occupied by the target task.

A CPU timeslice, also referred to as a "quantum" or "processor slice", is a period of CPU time that a time-slicing task system distributes to each running process microview. As you can see, a CPU time slice can relatively truthfully represent the CPU time occupied by a process (or task) running.

Thus, the CPU time slice occupancy of the target task in the task running period can relatively truthfully represent the ratio of the CPU time slice occupied by the target task running, and thus is suitable for measuring the CPU computing power occupation required for the target task running.

Based on this, in the above step, through the target task acquiring the CPU time slice occupancy in the task running period, the CPU computing power occupancy required for the target task running can be characterized relatively truthfully and accurately, so that the target task The CPU time slice occupancy rate in the task operation period may provide a basis for determining the amount of operation of the target task.

Here, the CPU time slice occupancy of the target task during the task operation period may be understood as the time slice occupancy of the target CPU core during the task operation period of the target task.

Step 103: Determine the amount of computation of the target task according to the DMIPS of the target CPU core and the CPU time slice occupancy rate.

Among them, DMIPS (Dhrystone Million Instructions executed Per Second) is a unit for measuring computer chips, and the larger the DMIPS value, the stronger the data processing capability of the CPU. MIPS (Million Instructions executed Per Second) is for calculating the processing power of the system within the same 1 second, that is, how many million instructions are executed every second. As you can see, DMIPS is a relatively truthful measure of a CPU's computational power, and DMIPS can be viewed as a unit of measure for a device's CPU's computational power. Thus, the DMIPS of the target CPU core can relatively truthfully measure the total computing power of the target CPU core.

In the present disclosure, through controlling the target task to run in the target CPU core, the DMIPS of the target CPU core can be used as the basis for determining the amount of computation of the target task, and thus the determined amount of computation of the target task is to be occupied during running of the target task. CPU computing power can be more truthfully and accurately characterized.

Since the target task's CPU time slice occupancy in the task running period is suitable for measuring the CPU computing power share required for target task running, the DMIPS of the target CPU core can relatively truthfully measure the total computing power of the target CPU core, and thus , In the above step, the amount of operation of the target task may be determined according to the DMIPS of the target CPU core and the CPU time slice occupancy of the target task in the task running period. In addition, the calculation amount of the target task thus determined can characterize the CPU calculation power to be occupied by the task more truthfully and more accurately.

For example, when the DMIPS of the target CPU core is A, the CPU time slice share of the target task in the task running period is B, and the amount of computation of the target task is C, then C = a A × b B, of which: a and b are both coefficients greater than zero.

In the present disclosure, when it is necessary to evaluate the amount of computation for a target task, the target task obtains the CPU time slice occupancy rate in the task running period by controlling the target task to run in the target CPU core, that is, The amount of computation of the target task may be determined according to the DMIPS of the target CPU core and the CPU time slice occupancy rate. Because of the use of the above technical means, the task calculation amount determined in the present disclosure is more true, and more accurately characterizes the CPU calculation power to be occupied by the task. Therefore, when distributing to tasks, based on the task calculation amount determined in the present disclosure, a more valuable reference can be provided for selecting the hardware type of the device.

What needs to be explained is that currently there is still a technology that measures and evaluates the amount of task computation based on CPU occupancy, and that technology statistics CPU occupancy through idle task timing, but this method only applies to continuous tasks. and does not have universality. As for some non-contiguous tasks, it is not possible to implement the determination of the task operation amount based on the CPU occupancy rate. Take the image detection algorithm as an example, the image detection algorithm is to process images frame by frame, and there is a lack of continuity within the task operation period, and based on the CPU occupancy rate, the amount of operation of the image detection algorithm is measured and evaluated. If so, the variance in the results is relatively large.

However, the method for determining the amount of task computation based on DMIPS and CPU time slice occupancy provided by the present disclosure can relatively truthfully and accurately characterize the CPU time slice occupancy required for task operation, so that continuous task It is applied not only to , but also to non-continuous tasks, and has universality.

Optionally, the step of obtaining, by the target task, a CPU time slice occupancy in a task running period:

obtaining, by the target CPU core, a CPU time slice in the task running period;

obtaining a CPU time slice occupied by the target task; and

and determining, by the target CPU core, the occupancy rate of the CPU time slice during the task operation period by the target task according to the CPU time slice during the task operation period and the CPU time slice occupied by the target task.

In this implementation mode, a CPU time slice in a task execution period of a target CPU core may be understood as a total CPU time slice in the period in which the target CPU core is running. Of the total CPU time slices in the period, partial time slices are for running the target task, partial time slices are for running other tasks or processes, or all time slices are for running the target task. it could be

A CPU time slice occupied by a target task can be understood as a CPU time slice occupied by the target task itself.

In this embodiment, the acquisition of the above two CPU time slices may be implemented through a CPU time slice occupancy detection module.

For example, when it is detected that the target task starts running, the CPU time slice occupancy detection module reads the time slice of /proc/stat of the target CPU core and writes it as t_cpu_start; When it is detected that the target task has finished running, the CPU time slice occupancy detection module reads the CPU time slice of /proc/stat of the target CPU core and writes it as t_cpu_end; Accordingly, the CPU time slice of the target CPU core during the task operation period is (t_cpu_end - t_cpu_start).

Correspondingly, when it is detected that the target task starts running, the CPU time slice occupancy detection module reads the time slice of /proc/[pid]/stat of the target CPU core and writes it as t_app_cpu_start; When it is detected that the target task has finished running, the CPU time slice occupancy detection module reads the CPU time slice of /proc/[pid]/stat of the target CPU core and writes it as t_app_cpu_end; Accordingly, the CPU time slice occupied by the target task is (t_app_cpu_end - t_app_cpu_start).

The pid in /proc/[pid]/stat described above indicates the id of the process and also the id of the target task.

The CPU time slice occupancy of the target task during the task running period can be obtained through the following formula:

It can be obtained through cpu_usage=(t_app_cpu_end-t_app_cpu_start)/(t_cpu_end-t_cpu_start).

Among them, cpu_usage represents the CPU time slice occupancy of the target task during the task running period.

In this implementation method, it is advantageous to accurately measure and evaluate the operation amount of the target task by determining the CPU time slice occupancy of the target task during the task operation period.

Optionally, the amount of computation of the target task is the amount of computation of the image detection algorithm at a target frame rate;

The step of determining the operation amount of the target task according to the DMIPS of the target CPU core and the CPU time slice occupancy rate is:

obtaining the number of image frames processed by the image detection algorithm within the task operation period;

determining a single frame operation amount of the image detection algorithm according to the DMIPS of the target CPU core, the CPU time slice occupancy rate, the task running period, and the number of image frames; and

and determining, by the image detection algorithm, an operation amount at the target frame rate according to the single frame operation amount and the target frame rate.

When an image detection algorithm is applied, it is generally necessary to refer to the image detection efficiency, that is, to refer to the frame rate of image detection. Based on this, in the implementation method, when the amount of calculation of the image detection algorithm needs to be calculated, the image detection algorithm may consider the amount of calculation at the target frame rate.

Referring to Fig. 2 below, an illustrative description of the processing flow chart of the above implementation manner is provided.

As shown in Fig. 2, a detection task interface 201, a CPU time slice occupancy detection module 202, an algorithm operation statistics module 203, and an operation amount analysis module 204 may be set in the measurement and evaluation device.

The image detection algorithm model 205 and the analog image 206 of a certain number of frames (eg, 500 frames) are input through the detection task interface 201, and the image detection algorithm is controlled to run in the target CPU core, so that all In order to complete the detection of the analog image, the image detection algorithm performs detection on the analog image. The period during which the image detection algorithm processes all analog images is the task operation period. The number of times each analog image is detected through an image detection algorithm may be one time or may be multiple times, and the present disclosure is not limited thereto.

Through the CPU time slice occupancy detection module 202, the CPU time slice occupancy during the running process of the image detection algorithm can be statistics, and through the algorithm operation statistics module 203, the start time, end time and The number of trips can be counted. Among them, the period from the start time to the end of the operation of the image detection algorithm is the task operation period, and the number of operations is the number of image frames processed by the image detection algorithm within the task operation period.

The CPU time slice occupancy detection module 202 may transmit the CPU time slice occupancy during the running process of the statistical image detection algorithm to the calculation amount analysis module 204, and the algorithm operation statistics module 203 may send the statistical image detection algorithm The operation start time, end time, and number of operations may be transmitted to the operation amount analysis module 204, and through the operation amount analysis module 204, the single frame operation amount of the image detection algorithm and the image detection algorithm may be performed at the target frame rate. Determine the amount of operation sequentially.

In this implementation method, the single operation amount of the image detection algorithm can be obtained by the following formula.

You can get it with dmip=duration*total_DMIPS*cpu_usage/M ,

Among them, dmip represents the amount of single operation of the image detection algorithm, duration represents the duration of the task operation, total_DMIPS represents the total computing power of the target CPU core, and cpu_usage represents the CPU time slice occupancy of the image detection algorithm during the duration of the task operation. , M represents the number of image frames processed by the image detection algorithm within the task running period.

After determining the single frame operation amount of the image detection algorithm, the image detection algorithm can obtain the operation amount at the target frame rate through the following formula, the formula is:

DMIPS= dmip*fps,

Among them, DMIPS represents the amount of operation of the image detection algorithm at the target frame rate, dmip represents the amount of single operation of the image detection algorithm, and fps represents the target frame rate.

In this implementation method, it is advantageous to accurately measure and evaluate the performance of the image detection algorithm through determining the amount of computation at the target frame rate of the image detection algorithm, and therefore, when distributing the image detection algorithm, the determined image Based on the amount of computation at the target frame rate, the detection algorithm provides a more valuable reference for device hardware type selection, thus facilitating the application of the image detection algorithm.

Optionally, the method:

The method may further include controlling an operating frequency of the target CPU core to be at a preset frequency while the target task is running in the target CPU core.

In this implementation mode, in order to prevent a change in system running frequency from affecting the determination of the computation amount of the target task, while the target task is running in the target CPU core, the running frequency of the target CPU core is locked. Thus, the operating frequency of the target CPU core is controlled to be at a preset frequency.

The preset frequency may be a preset frequency value or a preset frequency range. Through the target task running arrangement file, arrangement can be performed with respect to the running frequency of the target CPU core.

In the above practical method, by controlling the operating frequency of the target CPU core during the operation of the target task, a relatively stable operating environment is provided for the operation of the target task, thereby ensuring that the task operation amount determined in the present disclosure is more accurate. can

Optionally, controlling the target task to run in the target CPU core comprises:

and controlling the target task to run among the target CPU cores with a preset priority.

In the above implementation manner, the target task can be controlled to run in the target CPU core with a relatively high priority, so that the target task can run preferentially, and there is no need to wait for a long time, so the target task runs. To improve efficiency, the computational amount of the target task can be determined relatively quickly.

Optionally, before controlling the target task to run in a target CPU core, the method comprises:

The method further includes receiving the target task transmitted from the cloud server and uploading the target task to the cloud server in advance by a user.

In the above implementation manner, the target task may be pushed to the measurement and evaluation device via the cloud server, so that the measurement and evaluation device may receive the target task from the cloud server and determine the amount of computation of the target task.

In the implementation method, the task computing amount measurement and evaluation system is distributed to a cloud server, and an open measurement and evaluation environment is built, thereby constructing an open measurement and evaluation environment, so as to measure and evaluate efficiency of task computing amount. And it is advantageous to improve the measurement evaluation effect.

Referring to Fig. 3 below, an exemplary description of the flow chart of the above-described implementation method is provided.

As shown in FIG. 3 , the cloud server 301 may push the target task to different measurement evaluation devices 302 . The user can upload the task waiting for measurement to the cloud server 301 through the web interface 303, and after receiving the task waiting for measurement from the cloud server 301, the model of the task waiting for measurement is automatically compiled, It may also push the model of the measurement awaiting task to different measurement evaluation devices 302 .

Optionally, after computing the amount of operations of the target task, the method:

The method may further include reporting the amount of computation of the target task to the cloud server.

As shown in FIG. 3 , in the implementation method, the calculation amount of the target task determined by the measurement evaluation device 302 may be reported to the cloud server 301. Accordingly, the cloud server 301 may report the target task The amount of operation can be fed back to the user.

In the implementation method, task design, task training, task distribution, and task computation evaluation are formed in a closed loop by distributing the task calculation evaluation system to the cloud server 301 (as shown in FIG. 4 ). together). Accordingly, after task design personnel complete task design and task training, the task can be published to the cloud server 301, and the cloud server 301 immediately pushes the task to the measurement evaluation device through the measurement evaluation device. Task calculation evaluation is performed, and the task calculation evaluation device measures and evaluates the task calculation amount again to the cloud server, and task design personnel take control of the consumption status of the task calculation amount in time so that customized optimization for the task can be easily performed. advantageous to

It should be noted that the multi-optional implementation methods in the method for determining the amount of computation of a task in the present disclosure may be implemented in conjunction with each other, or may be implemented independently, and the present disclosure is not limited thereto.

The above-described embodiments of the present disclosure have the following advantages or advantageous effects.

In the present disclosure, when it is necessary to evaluate the amount of computation for a target task, the CPU time slice occupied by the target task during the running process is obtained by controlling the target task to run in the target CPU core, that is, the target task The amount of computation of the target task is determined according to the DMIPS of the CPU core and the CPU time slice occupied by the target task. Because of the use of the above technical means, the task calculation amount determined in the present disclosure is more true, and more accurately characterizes the CPU calculation power to be occupied by the task. Therefore, when distributing to tasks, based on the task calculation amount determined in the present disclosure, a more valuable reference can be provided for selecting the hardware type of the device.

In the present disclosure, an apparatus for determining an amount of operation for a task is provided, and as shown in FIG. 5 , the apparatus for determining an amount of operation for a task 400 includes:

a first control module 401 for controlling the target task to run in the target CPU core;

The acquisition module 402 for acquiring the CPU time slice occupancy rate of the target task in a task running period, wherein the task running period is a period from when the target task starts running to when the target task ends running. , the acquisition module 402; and

a determination module 403, configured to determine an operation amount of the target task according to the DMIPS of the target CPU core and the CPU time slice occupancy rate; includes

Optionally, the acquisition module 402:

a first obtaining submodule configured for the target CPU core to obtain a CPU time slice in the task running period;

a second acquiring submodule configured to acquire a CPU time slice occupied by the target task; and

A first determination submodule configured to determine, according to the CPU time slice of the target CPU core in the task running period and the CPU time slice occupied by the target task, the CPU time slice occupancy of the target task in the task running period ; includes

Optionally, the amount of calculation of the target task includes the amount of calculation of an image detection algorithm, the amount of calculation of a voice recognition algorithm, or the amount of calculation of a planning and control algorithm.

Optionally, the amount of computation of the target task is the amount of computation of the image detection algorithm at a target frame rate;

The confirmation module 403:

a second acquiring submodule configured to acquire the number of image frames processed by the image detecting algorithm within the task running period;

a second determination submodule configured to determine a single frame operation amount of the image detection algorithm according to the DMIPS of the target CPU core, the CPU time slice occupancy rate, the task running period and the number of image frames; and

a third determining submodule configured to determine, according to the single frame operation amount and the target frame rate, an operation amount at the target frame rate of the image detecting algorithm; includes

Optionally, the task calculation amount determination device 400:

and a second control module for controlling an operating frequency of the target CPU core to be at a predetermined frequency while the target task is running in the target CPU core.

Optionally, the first control module 401 specifically:

This is to control the target task to run among the target CPU cores with a preset priority.

Optionally, the task calculation amount determination device 400:

A receiving module for receiving the target task transmitted from the cloud server is further included, and the user uploads the target task to the cloud server in advance.

Optionally, the task calculation amount determination device 400:

and a reporting module for reporting the amount of computation of the target task to the cloud server.

The apparatus 400 for determining the amount of operation for a task provided in the present disclosure can implement each process implemented by the apparatus for determining the amount of task operation amount in the above-described embodiment of the method for determining the amount of operation for a task, and can achieve the same beneficial effects, avoiding redundant descriptions. To do so, it is not described in detail here.

In an embodiment of the present disclosure, an electronic device and a readable storage medium are provided in the present disclosure.

As shown in FIG. 6 , it is a block diagram of an electronic device of a method for determining a task computation amount according to an embodiment of the present disclosure. Electronic devices refer to various types of digital computers such as laptop computers, desk computers, benches, personal digital assistants, servers, blade servers, large computers, and other suitable computers. Electronic devices refer to various types of mobile devices such as personal digital processors, cellular telephones, smart phones, wearable devices and other similar computing devices. The components presented herein, their connections and relationships, and their functions are by way of example only, and are not intended to limit implementations of the disclosure described and/or claimed herein.

As shown in FIG. 6 , the electronic device includes: one or a plurality of processors 501, a memory 502, and interfaces for connecting each component including a high-speed interface and a low-speed interface. Each component is interconnected using a different bus and can also be installed on a common main plate or in other ways as required. The processor may process commands executed in the electronic device, and includes commands of image information stored in memory or displayed as a GUI in an input/output device (e.g., a display device coupled to an interface) outside the memory. . In other embodiments, if necessary, a plurality of processors, a plurality of buses, a plurality of memories, and a plurality of memories may be used together. Similarly, in connection with a plurality of electronic devices, each device may partially provide necessary operations (eg, a server array, a blade server, or a multi-processor system). One processor 501 in FIG. 6 is taken as an example.

The memory 502 is a non-transitory computer readable storage medium provided by the present disclosure. Among them, the memory stores instructions executable by at least one processor, so that the at least one processor can execute the method for determining the amount of task computation provided by the present disclosure. Computer instructions are stored in the non-transitory computer readable storage medium of the present disclosure, and the computer instructions are for causing a computer to execute the method for determining the amount of task computation provided in the present disclosure.

The memory 502 is a kind of non-transitory computer-readable storage medium, and may be used to store non-transitory software programs, non-transitory computer-executable programs and modules, and correspond to, for example, the method for determining the amount of operation of a task in an embodiment of the present disclosure. program instructions/modules (eg, the first control module 401, the acquisition module 402 and the determination module 403 as shown in FIG. 5) may be stored. The processor 501 runs non-temporary software programs, instructions and modules stored in the memory 502, and thus executes various function applications and data processing of the device for determining the amount of operation of the task, and the tasks in the above-described method embodiments. Implements the calculation amount determination method.

The memory 502 may include a program storage area and a data storage area, where the program storage area may store an operating system and application programs required for at least one function; The data storage area may store data and the like constructed according to the use of the electronic device of the data thinning method. Additionally, memory 502 may include high-speed random access memory, and may further include non-transitory memory, such as at least one magnetic disk memory, flash memory, or other non-transitory solid-state memory. In some embodiments, the memory 502 may optionally include a memory installed remotely from the processor 501, and such a remote memory may be connected to an electronic device of the method for determining the amount of task calculation through a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

An electronic device implementing the method for determining the amount of operation of a task may include: an input device 503 and an output device 504 . The processor 501, the memory 502, the input device 503, and the output device 504 may be connected through a bus or other methods, and in FIG. 6, connection through a bus is exemplified.

The input device 503 can receive input digital or character information, and can generate key signal input related to user setting and function control of an electronic device that implements a method for determining a task calculation amount, and the input device can, for example, An input device such as a touch screen, keypad, mouse, track panel, touch panel, indicator bar, one or more mouse buttons, track ball, joystick, etc. The output device 504 may include a display device, an auxiliary lighting device (eg, LED), and a tactile feedback device (eg, a vibration motor). The display device may include, but is not limited to, a liquid crystal display (LCD), a light emitting diode (LED) display, and a plasma display. In some embodiments, the display device may be a touch screen.

Various embodiments of the systems and techniques described herein may be implemented in digital electronic circuit systems, integrated circuit systems, dedicated ASICs (Special Purpose Integrated Circuits), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include implementation in one or more computer programs, and the one or more computer programs may be executed and/or interpreted in a programmable system including at least one programmable processor, The programmable processor may be a dedicated or general purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and sending data and instructions to the storage system, the at least one to one input device, and to the at least one output device.

Such computing programs (also referred to as programs, software, software applications, or code) include the machine instructions of a programmable processor, and use a programming language for high-level processes and/or objects, and/or assembly/machine language to implement such computing programs. It can run computing programs. As used herein, the terms “machine-readable medium” and “computer-readable medium” refer to any computer program product, appliance, and/or device (eg, For example, it refers to a magnetic disk, an optical disk, a memory, a programmable logic device (PLD)), and includes a machine readable medium that receives a machine command, which is a machine readable signal. The term “machine readable signal” means any signal for providing machine instructions and/or data to a programmable processor.

To provide interaction with a user, a computer may implement the systems and techniques described herein, which computer may have a display device (e.g., a cathode ray tube (CRT) or a liquid crystal display (LCD) for displaying information to a user). device) monitor); and a keyboard and a pointing device (eg, a mouse or track ball), through which a user may provide input to the computer. Other types of devices may also provide interaction with a user; for example, the feedback provided to the user may be any form of sensory feedback (eg, visual feedback, auditory feedback, or tactile feedback); An input from a user may be received in any form (sound input, voice input, or tactile input).

The systems and techniques described herein can be applied to a computing system that includes a background member (e.g., a data server), or a computing system that includes a middleware member (e.g., an application server), or a computing system that includes a front-end member (e.g., a data server). For example, a user's computer having a graphical user interface or web browser, through which the user can interact with embodiments of the systems and techniques described herein), or such background members, middleware elements, or any combination of front end elements. The elements of the system may be interconnected through any form or medium of digital data communication (eg, a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

A computer system may include a client and a server. Clients and servers are generally remote from each other and typically interact with each other through a communication network. A relationship of client and server is created through computer programs running on corresponding computers and having a client-server relationship with each other.

According to the technical solutions in the embodiments of the present disclosure, in the present disclosure, when it is necessary to measure and evaluate the amount of computation for a target task, the target task is controlled to run in the target CPU core, and the target task occupies during the running process. A CPU time slice occupancy rate is obtained, that is, an operation amount of the target task is determined according to the DMIPS of the target CPU core and the CPU time slice occupancy occupied by the target task. Because of the use of the above technical means, the task calculation amount determined in the present disclosure is more true, and more accurately characterizes the CPU calculation power to be occupied by the task. Therefore, when distributing to tasks, based on the task calculation amount determined in the present disclosure, a more valuable reference can be provided for selecting the hardware type of the device.

It should be understood that, using the flowcharts presented above in various formats, you can reorder, add, or delete steps. For example, each step described in the present disclosure may be executed in parallel, may be executed sequentially, or may be executed in a different order, as long as the technical solution of the present disclosure can achieve the desired result, the specification is not limited thereto.

The specific implementation manners described above do not constitute a limitation on the protection scope of the present disclosure. Those of ordinary skill in the art should be aware that various modifications, combinations, subcombinations, and substitutions may be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present disclosure are all included within the protection scope of the present disclosure.

Claims (19)

A method for determining the amount of operation of a task performed by a measurement and evaluation device,
The method is:
controlling the target task to run in the target CPU core;
Acquiring, by the target task, a CPU time slice occupancy rate in a task running period from the measurement evaluation device, wherein the task running period is a period from when the target task starts to run until the target task ends. which is, obtaining; and
determining an operation amount of the target task according to the DMIPS of the target CPU core and the CPU time slice occupancy rate;
A method for determining the amount of operation of a task comprising: According to claim 1,
The step of obtaining, by the target task, the CPU time slice occupancy in the task running period:
obtaining, by the target CPU core, a CPU time slice in the task running period;
obtaining a CPU time slice occupied by the target task; and
determining, by the target CPU core, a CPU time slice occupied by the target task in the task running period according to the CPU time slice in the task running period and the CPU time slice occupied by the target task;
A method for determining the amount of operation of a task comprising: According to claim 1,
The amount of calculation of the target task includes the amount of calculation of the image detection algorithm,
The method of determining the amount of task operation, characterized in that the image detection algorithm is for processing images of each one frame. According to claim 3,
The amount of calculation of the target task is the amount of calculation of the image detection algorithm at the target frame rate;
The step of determining the operation amount of the target task according to the DMIPS and CPU time slice occupancy of the target CPU core:
obtaining the number of image frames processed by the image detection algorithm within the task operation period;
determining a single frame operation amount of the image detection algorithm according to the DMIPS of the target CPU core, the CPU time slice occupancy rate, the task running period, and the number of image frames; and
determining, by the image detection algorithm, an operation amount at the target frame rate according to the single frame operation amount and the target frame rate;
A method for determining the amount of operation of a task comprising: According to claim 1,
The method is:
The method of claim 1 , further comprising controlling an operating frequency of the target CPU core to be at a preset frequency while the target task is running in the target CPU core. According to claim 1,
The step of controlling the target task to run in the target CPU core is:
and controlling the target task to run among the target CPU cores with a preset priority. According to claim 1,
Before controlling the target task to run in the target CPU core, the method:
The method of determining the amount of operation of a task, further comprising receiving the target task transmitted from a cloud server and uploading the target task to the cloud server in advance by a user. According to claim 7,
After calculating the amount of computation of the target task, the method:
The task calculation amount determination method further comprising the step of reporting the calculation amount of the target task to the cloud server. In the task calculation amount determining device,
The device is:
A first control module for controlling the target task to run in the target CPU core;
An acquisition module for obtaining, from a measurement and evaluation device, the CPU time slice occupancy of the target task in a task running period, wherein the task running period is from the start of the target task to the end of the run of the target task. period, an acquisition module; and
a determining module configured to determine an amount of operation of the target task according to the DMIPS of the target CPU core and the CPU time slice occupancy rate;
An apparatus for determining the amount of operation of a task comprising: According to claim 9,
The acquisition module:
a first obtaining submodule configured for the target CPU core to obtain a CPU time slice in the task running period;
a second acquiring submodule configured to acquire a CPU time slice occupied by the target task; and
A first determination submodule configured to determine, according to the CPU time slice of the target CPU core in the task running period and the CPU time slice occupied by the target task, the CPU time slice occupancy of the target task in the task running period ;
An apparatus for determining the amount of operation of a task comprising: According to claim 9,
The amount of calculation of the target task includes the amount of calculation of the image detection algorithm,
The image detection algorithm is for processing images of each one frame. According to claim 11,
The amount of calculation of the target task is the amount of calculation of the image detection algorithm at the target frame rate;
The confirmation module:
a second acquiring submodule configured to acquire the number of image frames processed by the image detecting algorithm within the task running period;
a second determination submodule configured to determine a single frame operation amount of the image detection algorithm according to the DMIPS of the target CPU core, the CPU time slice occupancy rate, the task running period and the number of image frames; and
a third determining submodule configured to determine, according to the single frame operation amount and the target frame rate, an operation amount at the target frame rate of the image detecting algorithm;
An apparatus for determining the amount of operation of a task comprising: According to any one of claims 9 to 12,
The device is:
and a second control module configured to control an operating frequency of the target CPU core to be at a preset frequency while the target task is running in the target CPU core. According to any one of claims 9 to 12,
The first control module specifically:
The apparatus for determining the amount of operation of a task for controlling the target task to operate among the target CPU cores with a preset priority. According to any one of claims 9 to 12,
The device is:
The apparatus for determining the amount of operation of a task further comprising a receiving module configured to receive the target task transmitted from a cloud server, and a receiving module configured to upload the target task to the cloud server in advance by a user. According to claim 15,
The device is:
and a reporting module for reporting the computational quantity of the target task to the cloud server. A non-transitory computer-readable storage medium in which computer instructions are stored,
The computer instructions are for causing the computer to execute the method according to any one of claims 1 to 8.
A non-transitory computer-readable storage medium having computer instructions stored thereon. A computer program stored in a computer readable storage medium,
The computer program, when executed by at least one processor, implements the method according to any one of claims 1 to 8,
A computer program stored on a computer readable storage medium. delete

Images