Detailed Description
Exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding, and should be considered as merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
Fig. 1 is a schematic diagram of an operation control method of a cloud mobile phone according to an embodiment of the present disclosure. The method and the device can be applied to the situations of monitoring the GPU utilization rate of the cloud mobile phone in real time and controlling the application program operated by the cloud mobile phone based on the numerical value of the GPU utilization rate. The method can be executed by an operation control device of the cloud mobile phone, and the device can be realized in a hardware and/or software mode and can be generally integrated in a cloud mobile phone server or a server cluster formed by a plurality of cloud mobile phone servers together. The cloud mobile phone server (or may also be called as a cloud mobile phone platform) is used for performing unified scheduling and operation control on a plurality of cloud mobile phones.
Accordingly, as shown in fig. 1, the method specifically may include:
s101, acquiring the utilization rate of the target GPU corresponding to the monitoring cloud mobile phone in real time.
The monitoring cloud mobile phone is a cloud mobile phone instance with a GPU monitoring program deployed. By running the GPU monitoring program on the monitoring cloud handset, the effect of real-time monitoring of the usage of a particular GPU (i.e., the target GPU) that provides graphics processing resources to the monitoring cloud handset can be achieved. The GPU monitoring program may directly use a command line tool or a monitoring tool of the GPU, or may be a GPU monitoring script obtained by custom writing, which is not limited in this embodiment.
GPU is specifically understood to be hardware dedicated to graphics processing in a cloud handset environment. In cloud handsets, the GPU is not only responsible for graphics rendering, but also can participate in computationally intensive tasks. Because there is a situation that multiple cloud mobile phone instances share one GPU, other cloud mobile phone users sharing the GPU are also affected when the GPU load is too high, monitoring the usage of the GPU is important to ensure the cloud mobile phone performance and resource optimization.
The usage rate of the target GPU may be monitored and acquired in real time at a set monitoring frequency, for example, 30s, 1 minute, or 5 minutes.
S102, when the utilization rate of the target GPU exceeds the utilization rate threshold, acquiring each associated cloud mobile phone sharing the target GPU with the monitoring cloud mobile phone.
The GPU usage threshold may be specifically understood as a performance index limit set when monitoring GPU performance, and is used to determine whether the GPU usage is in a normal or expected state.
Generally, when the usage rate of the GPU exceeds the usage rate threshold, the GPU resource is overloaded, and the situations such as jamming of the cloud mobile phone feedback screen may occur, which affects the usage experience of the cloud mobile phone user. Therefore, the overload state needs to be monitored in real time, and certain measures are taken to reduce the picture card section. The usage rate threshold may be preset according to practical situations, for example, 90% or 95%, which is not limited in this embodiment.
Specifically, in the process of acquiring the usage rate of the target GPU corresponding to the monitoring cloud mobile phone in real time, if the usage rate of the target GPU providing the GPU resource for the monitoring cloud mobile phone is found to exceed the usage rate threshold, other cloud mobile phones sharing the target GPU with the monitoring cloud mobile phone (i.e., associated cloud mobile phones) are described to be in the state of overload of GPU resource, and the associated cloud mobile phones may also have the situation that the cloud mobile phone feedback picture is blocked, at this time, the associated cloud mobile phones also need to be positioned, and the associated cloud mobile phones and the monitoring cloud mobile phones are used together as an affected cloud mobile phone set, and the running control of the application program is performed on the whole affected cloud mobile phone set.
S103, identifying each GPU high-load application in the monitoring cloud mobile phone and each associated cloud mobile phone, and performing operation control on each GPU high-load application according to operation description information of the cloud mobile phone where each GPU high-load application is located.
GPU high-load applications can be understood specifically as applications requiring a GPU to perform a large number of computation and processing tasks, such as cloud games, high-definition video conferences, online learning platforms, high-resolution movie players, animation applications, movie special effect production applications, high-definition monitoring cameras, video analysis systems, and the like, which require a GPU to perform rendering.
Specifically, after each associated cloud mobile phone sharing the target GPU with the monitoring cloud mobile phone is obtained, identifying GPU high-load application in each cloud mobile phone according to IP of the cloud mobile phone and resource use condition information in the target GPU, and according to operation description information (such as application name and version, GPU use rate, application state and error log) of the cloud mobile phone where each GPU high-load application is located, making some targeted processing measures, such as limiting the number of multitasking, avoiding running multiple high-GPU load applications at the same time, reducing the overall load of the target GPU, realizing operation control over each GPU high-load application, and optimizing the use of GPU resources.
According to the technical scheme, the real-time abnormal monitoring of the GPU utilization rate of the cloud mobile phone can be achieved, when the utilization rate of the GPU exceeds the utilization rate threshold, all the associated cloud mobile phones sharing the target GPU with the monitoring cloud mobile phone are obtained, all the GPU high-load applications are identified in the monitoring cloud mobile phone and all the associated cloud mobile phones, and according to the operation description information of the cloud mobile phone where the GPU high-load applications are located, the GPU utilization rate of the cloud mobile phone is monitored in real time, when the GPU utilization rate is abnormal, all the GPU high-load applications in the monitoring cloud mobile phone and all the associated cloud mobile phones are identified, the operation control is carried out, and through the real-time monitoring of the GPU utilization rate of the cloud mobile phone, the complicated step of manual inspection of a user is omitted, and potential problems can be found and handled quickly. The automatic operation control mechanism ensures the stable operation of the target GPU and improves the reliability of the whole operation of the cloud mobile phone.
On the basis of the embodiments, the embodiments of the invention not only can automatically control the operation of each GPU high-load application (the whole process is not perceived by a user) according to the operation description information of the cloud mobile phone where each GPU high-load application is located, but also can actively send out abnormal warning or notification so as to facilitate a system administrator or the cloud mobile phone user to timely know the use condition of GPU resources and take corresponding measures.
In a specific example, when the usage rate of the target GPU is monitored to exceed the preset usage rate threshold, which means that GPU resources are overloaded, the cloud mobile platform may issue a GPU alert to the cloud mobile user, and after the cloud mobile user receives the GPU alert, some targeted processing measures may be performed for the problem, for example, manually closing or optimizing the cloud mobile application, etc. Or the cloud mobile phone platform can also send out a GPU warning to an administrator of a merchant to which the cloud mobile phone belongs, and after receiving the GPU warning, the administrator can also make some targeted processing measures for the problem, such as directly selecting a background killing (kill) designated GPU high-load application and the like.
Furthermore, embodiments of the present disclosure primarily implement a run control scheme when the usage of the target GPU exceeds the usage threshold. In practice, when the usage rate of the target GPU is smaller than the preset low usage rate threshold, the cloud mobile phone application can be effectively controlled. In a specific example, when the usage rate of the target GPU corresponding to the monitoring cloud mobile phone is smaller than a low usage rate threshold (for example, 20% or 30%), at this time, an application program currently running in the foreground of the monitoring cloud mobile phone may be acquired, and whether the application program is in an application setting with an optimal display effect is detected, if not, the application setting of the application program may be adjusted up (for example, resolution is improved or a high consumption graphics special effect is started) at this time, so that under the condition that the target GPU is sufficient, effective utilization of GPU resources is ensured, and application usage experience of a user of the cloud mobile phone is improved,
Fig. 2 is a schematic diagram of another operation control method of a cloud mobile phone according to an embodiment of the present disclosure. The present embodiment is refined based on the above embodiments. In the embodiment, the operation of identifying each GPU high-load application in the monitoring cloud mobile phone and each associated cloud mobile phone is embodied as identifying application programs running in the foreground in the monitoring cloud mobile phone and each associated cloud mobile phone respectively, acquiring the resource occupancy rate of each application program to the target GPU, and determining the application programs with the resource occupancy rate exceeding an occupancy rate threshold value as the GPU high-load application. "
And the operation of performing operation control on each GPU high-load application according to the operation description information of the cloud mobile phone where each GPU high-load application is located is embodied in the following steps of reallocating resources in the target GPU for each GPU high-load application according to the network transmission speed between the cloud mobile phone where each GPU high-load application is located and each entity device currently establishing communication connection. "
Accordingly, as shown in fig. 2, the method specifically may include:
S201, the utilization rate of the target GPU corresponding to the monitoring cloud mobile phone is obtained in real time.
S202, when the utilization rate of the target GPU exceeds the utilization rate threshold, acquiring each associated cloud mobile phone sharing the target GPU with the monitoring cloud mobile phone.
S203, identifying application programs running in the foreground in the monitoring cloud mobile phone and each associated cloud mobile phone respectively, and acquiring the resource occupancy rate of each application program to the target GPU.
S204, determining the application program with the resource occupancy rate exceeding the occupancy rate threshold value as the GPU high-load application.
In general, applications requiring graphics rendering by a GPU, including applications such as cloud games, high definition video conferencing, and high definition movie players, consume a significant amount of GPU resources (i.e., various hardware resources such as processing cores, video memory, and memory bandwidth for performing computing tasks) when running in the foreground.
Therefore, when the utilization rate of the target GPU exceeds the utilization rate threshold, the application programs of all cloud mobile phones currently sharing the target GPU in foreground operation are identified, the resource occupancy rate of all the application programs to the target GPU is obtained, and GPU high-load applications with the resource occupancy rate exceeding the occupancy rate threshold are screened out.
S205, re-distributing resources in the target GPU for each GPU high-load application according to the network transmission speed between the cloud mobile phone where each GPU high-load application is located and each entity device which is currently in communication connection.
In particular, the network transmission speed is understood to be the rate at which data is transmitted in the network, measured in terms of the amount of data transmitted per second. In the field of cloud mobile phones, network transmission speed is important to user experience, especially in applications requiring high-resolution image display, such as cloud games, high-definition video playing, and the like. If the network transmission speed is insufficient, image delay, jam or degradation of image quality may be caused. After the GPU high-load application of each cloud mobile phone is determined, resources in the target GPU are allocated to each GPU high-load application again through a resource scheduling algorithm according to the GPU load and the network transmission speed of the application. If the network transmission speed of a cloud mobile phone is detected to be faster, the system may allocate more GPU resources to process higher-resolution images and video streams, if the network transmission speed of the cloud mobile phone is detected to be slower, in order to ensure smooth operation and user experience of cloud mobile phone applications, a data compression and coding optimization technology is adopted to reduce the transmission data amount, or the transmitted data or calculation results are cached to reduce repeated transmission, reduce occupation of network bandwidth, reduce overall load of the GPU, realize operation control of each GPU high-load application, and optimize use of GPU resources.
According to the technical scheme, the utilization rate of the target GPU corresponding to the monitoring cloud mobile phone is obtained in real time, when the utilization rate of the target GPU exceeds the utilization rate threshold, all the associated cloud mobile phones sharing the target GPU with the monitoring cloud mobile phone are obtained, application programs running on a front platform are respectively identified in the monitoring cloud mobile phones and all the associated cloud mobile phones, the resource occupancy rate of all the application programs to the target GPU is obtained, the application programs with the resource occupancy rate exceeding the occupancy rate threshold are determined to be GPU high-load application, the network transmission speed between the cloud mobile phones with the high-load application of all the GPU and all the entity equipment currently establishing communication connection is determined, real-time abnormal monitoring of the utilization rate of the GPU of the target GPU can be realized, when the utilization rate of the GPU is abnormal, all the GPU high-load front platform applications in the monitoring cloud mobile phones and all the associated cloud mobile phones are effectively identified according to the resource occupancy rate and the occupancy rate threshold, the resources in the target GPU are distributed according to the network transmission speed of all the applications, the front platform applications are guaranteed to obtain reasonable resources under the condition that network conditions are limited, the front platform applications are guaranteed to be reasonably used, the running performance of the GPU is guaranteed, the running of the cloud mobile phones is guaranteed, and the running of the cloud mobile phones can be stable, and the running performance of the GPU is guaranteed, and the user experience of the cloud mobile phones can be guaranteed.
In another optional implementation manner of this embodiment, reallocating, according to a network transmission speed between the cloud mobile phone where each GPU high load application is located and each entity device currently establishing a communication connection, resources in the target GPU for each GPU high load application includes:
acquiring the current network transmission speed between a cloud mobile phone where a current GPU high-load application is located and entity equipment which is currently in communication connection in real time, and acquiring the current frame rate of the current GPU high-load application;
And if the current frame rate is larger than the current network transmission speed, adjusting the current frame rate to be matched with the current network transmission speed.
The current frame rate of the current GPU high load application can be understood as the number of images rendered by the current GPU high load application per second by using the resources provided by the target GPU. In general, the higher the frame rate, the higher the occupancy of GPU resources at a given resolution.
It can be understood that when the current frame rate is higher than the current network transmission speed, the current network transmission speed is limited, and the image with high frame rate cannot be provided for the cloud mobile phone user to perform high-definition display immediately. At this time, even if the target GPU is highly occupied, the desired display effect is not achieved. Based on the method, the current frame rate can be adjusted to be matched with the current network transmission speed, and at the moment, the resource occupation of the current GPU high-load application to the target GPU can be effectively reduced while the original display effect of the cloud mobile phone user is ensured.
Through the arrangement, the adaptation of the frame rate and the network transmission speed is realized, the rendering pressure of the cloud mobile phone image is properly reduced under the condition of limited network conditions, meaningless waste of the GPU is avoided, and the performance of the GPU is exerted to the greatest extent.
Fig. 3 is a schematic diagram of another operation control method of a cloud mobile phone according to an embodiment of the present disclosure. The present embodiment is refined based on the above embodiments. In the embodiment, the operation of performing operation control on each GPU high-load application according to the operation description information of the cloud mobile phone where each GPU high-load application is located is embodied in the steps of acquiring priority weight information of the cloud mobile phone where each GPU high-load application is located, screening at least one target cloud mobile phone according to the order of the priority weight information from low to high, and reducing resources in the target GPU distributed to each GPU high-load application in each target cloud mobile phone by adopting a preset strategy. "
Accordingly, as shown in fig. 3, the method specifically may include:
S301, acquiring the utilization rate of the target GPU corresponding to the monitoring cloud mobile phone in real time.
S302, when the utilization rate of the target GPU exceeds the utilization rate threshold, acquiring each associated cloud mobile phone sharing the target GPU with the monitoring cloud mobile phone.
And S303, identifying each GPU high-load application in the monitoring cloud mobile phone and each associated cloud mobile phone.
S304, acquiring priority weight information of the cloud mobile phone where each GPU high-load application is located.
Priority weight information is understood to mean, in particular, the priorities and weights assigned to the different applications or services in the scheduling of resources and the management of tasks. This information helps the system decide how to allocate limited resources such as GPU, memory, storage space, or network bandwidth. The priority determines which applications or services should obtain the resources first in case of limited resources. Applications with higher priority (e.g., applications involving critical business processes or frequent user interactions) may be prioritized when allocating resources. The weight is an evaluation index when different applications have the same priority, which determines the relative share of the application in the resource allocation. For example, higher weighted applications (e.g., applications involving critical business processes or frequent user interactions) may get more resources, ensuring that the application is running stably. In cloud handsets, this setting of priorities and weights allows the system to dynamically adjust the resource allocation according to the importance of the application and the current performance requirements. For example, if a cloud handset is running a high quality graphics and visual game, the system may automatically increase the priority and weight of the game to ensure that it is able to obtain enough GPU resources to maintain a smooth gaming experience. Conversely, if the network transmission speed is insufficient to support smooth transmission of high resolution images, the system may reduce the priority and weight of applications with higher image quality requirements to ensure basic quality of service and user experience.
S305, screening out at least one target cloud mobile phone according to the order of the priority weight information from low to high.
S306, adopting a preset strategy to reduce resources in the target GPU allocated for each GPU high-load application in each target cloud mobile phone.
Specifically, after the high-load applications of each GPU are identified in the monitoring cloud mobile phone and each associated cloud mobile phone, priority weight information of the cloud mobile phone where each GPU is located is obtained and sequenced, at least one cloud mobile phone with low priority weight is screened out as a target cloud mobile phone, and resources in the distributed target GPU are differentially reduced according to the priority weights of the screened cloud mobile phones. If the screened target cloud mobile phone is more than one, the allocated GPU resources are reduced according to the priority weight differentiation of the target cloud mobile phone (for example, the display quantity allocated to a specific application is reduced or the number of GPU computing cores used by the application is limited), for high-priority applications, namely, critical applications, more GPU resources can be ensured to be still held after the allocated GPU resources are reduced, and for low-priority applications, namely, non-critical applications, the reduction quantity of resource allocation can be adaptively increased so as to ensure that the performance of the critical applications is not influenced. When the priorities of the applications are the same, the allocated GPU resources can be reduced according to the weight differentiation, for high-weight applications, namely, critical applications, more GPU resources can be ensured to be still held after the allocated GPU resources are reduced, and for low-weight applications, namely, non-critical applications, the reduction amount of resource allocation can be adaptively increased, so that the performance of the critical applications is not influenced, the stable operation of the target GPU is ensured, and the reliability of the integral operation of the cloud mobile phone is improved.
According to the technical scheme, the utilization rate of the target GPU corresponding to the monitoring cloud mobile phone is obtained in real time; the method comprises the steps of obtaining each associated cloud mobile phone sharing a target GPU with a monitoring cloud mobile phone when the utilization rate of the target GPU exceeds a utilization rate threshold value, identifying each GPU high-load application in the monitoring cloud mobile phone and each associated cloud mobile phone, obtaining priority weight information of the cloud mobile phone where each GPU high-load application is located, screening at least one target cloud mobile phone according to the order of the priority weight information from low to high, adopting a preset strategy to reduce the technical means of resources in the target GPU distributed to each GPU high-load application in each target cloud mobile phone, realizing real-time abnormal monitoring of the GPU utilization rate of the cloud mobile phone, identifying each GPU high-load foreground application in the monitoring cloud mobile phone and each associated cloud mobile phone when the utilization rate of the GPU is abnormal, reducing the resources distributed to the target GPU by the application of the target cloud mobile phone according to priority weight differentiation of each application, ensuring that the high-priority weight application can obtain enough resources under the condition of shortage of the GPU resources, maintaining smooth user experience and application performance, preventing system waste, reducing system overload or system performance caused by resource competition, and improving overall reliability of running of the cloud mobile phone, and ensuring stable running of the overall GPU.
In another optional implementation manner of this embodiment, the reducing, by using a preset policy, resources in the target GPUs allocated to the GPU high-load applications in the target cloud mobile phones includes at least one of:
Adjusting the operation mode of the GPU high-load application in the target cloud mobile phone to a very simple mode;
Closing a high-consumption graphic special effect in the GPU high-load application in the target cloud mobile phone;
and reducing the resolution or frame rate of the GPU high-load application in the target cloud mobile phone.
Specifically, under the condition of shortage of GPU resources, reducing resources in the target GPU allocated to each GPU high-load application in each target cloud mobile phone can provide additional available GPU resources for other applications by reducing the GPU resource requirements of the application, so that the reallocation of the GPU resources is realized. The reducing the GPU resource requirement may include adjusting the operation mode of the GPU high load application in the target cloud mobile phone to a very simple mode, that is, adjusting the graphic setting of the application to the minimum, reducing the complexity of graphic rendering, thereby reducing the load of the GPU, for example, closing shadows, reducing texture quality or reducing reflection effects in games, closing high consumption graphic special effects in the GPU high load application in the target cloud mobile phone, that is, closing special effects such as ray tracing, antialiasing, complex particle effects, and the like, which consume larger GPU resources, so as to reduce the workload of the GPU, reducing the resolution or frame rate of the GPU high load application in the target cloud mobile phone, that is, reducing the number of pixels rendered by the GPU by reducing the output resolution of games or applications, thereby reducing the use rate of the GPU, or reducing the number of frames required to be rendered by the GPU per second by limiting the frame rate, thereby reducing the use rate of the GPU, balancing the load, ensuring the stable operation of the GPU, and improving the reliability of the overall operation of the target GPU.
In another optional implementation manner of this embodiment, the obtaining priority weight information of the cloud mobile phone where each GPU high load application is located includes:
acquiring cloud mobile phone users to which the cloud mobile phones where the GPU high-load applications are located belong;
acquiring historical cloud mobile phone use frequency information and user grade information of each cloud mobile phone user;
According to the historical cloud mobile phone use frequency information and the user grade information, calculating to obtain priority weight information of the cloud mobile phone where each GPU high-load application is located;
the history cloud mobile phone use frequency information comprises at least one of history use duration, history cloud mobile phone use frequency and history cloud mobile phone use high-frequency time period.
The history cloud mobile phone use frequency information can be specifically understood as recording the history use information of the cloud mobile phone used by the user, and can be used for understanding the behavior mode of the user and carrying out resource allocation and optimizing service according to the history use information. The historical cloud mobile phone use frequency information comprises at least one of historical use duration, historical cloud mobile phone use frequency and historical cloud mobile phone use high-frequency time period. The historical use time length reflects the total time of the user using the cloud mobile phone, the historical use time length can influence the priority weight of the user, because the user who uses for a long time may need more stable resource allocation, the historical cloud mobile phone use frequency reflects the frequency of the user accessing the cloud mobile phone service, the user with high use frequency can be given higher priority weight, the historical cloud mobile phone use high-frequency time period reflects the use peak of the user in a specific time period, and the user can be given higher priority weight in the specific time period, so that the prediction and the resource allocation are facilitated to meet the peak demand.
Specifically, the priority weight information of the cloud mobile phone where each GPU high load application is located can be calculated through historical cloud mobile phone usage frequency information and user grade information (refer to a subscription grade or a member grade of a user, and a user with a high subscription grade or a high member grade can be given higher priority weight). After the cloud mobile phone users of the cloud mobile phones where the high-load application of each GPU belongs are obtained, historical cloud mobile phone use frequency information and user grade information of each cloud mobile phone user are obtained, the indexes are quantized, for example, the use duration and the frequency are converted into numerical values, different weights are distributed for different indexes, for example, if the user grade is higher, higher weights can be given, then weighting calculation is carried out, and priority weights are dynamically adjusted according to real-time data in the subsequent operation process so as to adapt to the continuously-changing use modes. By identifying the use habit and preference of the user, more personalized service can be provided for the user, the resource requirements of the high-value user or the high-frequency use period are ensured to be met preferentially, the resource utilization efficiency is improved, the stable operation of the target GPU is ensured, and the reliability of the integral operation of the cloud mobile phone is improved.
On the basis of the above embodiments, after the obtaining, in real time, the usage rate of the target GPU corresponding to the monitoring cloud mobile phone, the method may further include:
When the utilization rate of the target GPU is not more than the utilization rate threshold value, determining whether the utilization rate of the target GPU is in an ascending trend according to the historical utilization rate of the target GPU obtained by at least two last monitoring;
if yes, acquiring user historical behavior data of the cloud mobile phone users to which the monitoring cloud mobile phone and each associated cloud mobile phone belong;
predicting the predicted occurrence time of the GPU usage overrun of the target GPU according to the historical behavior data of each user;
and if the difference value between the estimated time and the current system time is smaller than a preset difference value threshold, respectively providing the estimated time for each cloud mobile phone user to perform autonomous decision control.
Specifically, in the process of acquiring the usage rate of the target GPU corresponding to the monitoring cloud mobile phone in real time, when the usage rate of the target GPU does not exceed the usage rate threshold, that is, is in a normal working state, determining a change trend of the usage rate of the target GPU according to the historical usage rate of the target GPU obtained by at least two last times of monitoring, if the usage rate of the target GPU is a decreasing trend, indicating that the target GPU has no overload risk, and if the usage rate of the target GPU is an increasing trend, indicating that the load of the target GPU is aggravated, and further, further judging whether the overload risk exists in the target GPU according to the user historical behavior data of the monitoring cloud mobile phone and cloud mobile phone users to which each associated cloud mobile phone belongs. According to the obtained historical behavior data of the user, the historical behavior data of the user is analyzed through a machine learning algorithm, and a prediction model is built, so that the condition possibly causing overload of the GPU is recognized in advance, and the predicted occurrence time of the occurrence of the overrun of the GPU by the target GPU is predicted. When the difference between the expected occurrence time and the current system time is smaller than a preset difference threshold, the expected occurrence time is respectively provided for each cloud mobile phone user in an abnormal early warning mode to carry out autonomous decision control, the GPU overload time is predicted by analyzing historical use data of the user in the cloud mobile phone, and early warning is sent to the user before possible overload is predicted, so that user experience can be effectively improved, and the smoothness of the equipment performance is ensured. The method is beneficial to optimizing resource allocation, reducing operation cost and system fault risk caused by overload, improving system stability, realizing effective management of GPU load and improving reliability of cloud mobile phone operation control.
Typically, if the usage rate of the target GPU is an increasing trend, the user historical behavior data of the cloud mobile phone user may be obtained, for example, the application program types run by the user, especially those applications with high demands on GPU resources, such as games or video editing software, etc., and the usage frequency and duration of these applications, and the usage condition of the GPU resources when the user uses the cloud mobile phone, including the usage duration, the usage frequency, the usage time period, etc. The method comprises the steps of dividing users into different behavior groups through cluster analysis according to system performance feedback data such as frame rate, delay and cartoon condition, identifying high-risk overload user groups, constructing a decision tree model to identify key factors causing GPU overload, performing integrated learning by combining random forest with time sequence analysis, and predicting future behavior patterns and potential GPU load peaks through analyzing the time sequence of the user behavior data to obtain the expected occurrence time of GPU overload. And judging the respective expected occurrence time and a preset difference threshold according to the high-risk overload user group, and when the expected occurrence time is smaller than the preset difference threshold, carrying out overload prompt on the high-risk overload user and the cloud mobile phone users related to the high-risk overload user, displaying the expected occurrence time, reminding the user to carry out autonomous decision control, for example, reminding the user to reduce the use of resource-intensive applications, and ensuring the stable operation of the cloud mobile phone.
On the basis of the above embodiments, after the obtaining, in real time, the usage rate of the target GPU corresponding to the monitoring cloud mobile phone, the method may further include:
When the utilization rate of the target GPU is determined not to exceed the utilization rate threshold, acquiring user historical behavior data of cloud mobile phone users of the monitoring cloud mobile phones and the associated cloud mobile phones, predicting resources required by the cloud mobile phone users of the monitoring cloud mobile phones and the associated cloud mobile phones, and simultaneously carrying out intelligent preloading of a plurality of tasks by utilizing residual resources through multithreading and asynchronous processing, or
And when the utilization rate of the target GPU is determined not to exceed the utilization rate threshold, dynamically adjusting the running mode of the GPU according to the utilization rate of the current target GPU and a preset threshold.
Specifically, in the process of acquiring the utilization rate of the target GPU corresponding to the monitoring cloud mobile phone in real time, when the utilization rate of the target GPU does not exceed the utilization rate threshold, namely, is in a normal working state, acquiring user historical behavior data of the monitoring cloud mobile phone and cloud mobile phone users to which each associated cloud mobile phone belongs, analyzing the user historical behavior data through a machine learning algorithm, screening out the cloud mobile phone to which the monitoring cloud mobile phone and each associated cloud mobile phone belong to predict high-load demand users, predicting resources required by the users, and utilizing the residual GPU resources to use multithreading and asynchronous processing to realize intelligent preloading of the resources. In the process of acquiring the utilization rate of the target GPU corresponding to the monitoring cloud mobile phone in real time, when the utilization rate of the target GPU does not exceed the utilization rate threshold, namely is in a normal working state, the running mode of the GPU can be dynamically adjusted according to the relation between the utilization rate of the current target GPU and a preset threshold, for example, when the utilization rate of the GPU is smaller than or equal to a low threshold, between the low threshold and a high threshold and larger than or equal to the high threshold according to the load condition, the running mode of the GPU can be respectively set into a low-performance mode, a medium-performance mode or a high-performance mode. The user history behavior data is used for predicting required resources, preloading is carried out according to the prediction result, and the GPU working mode is dynamically adjusted, so that the response speed and the resource utilization rate of the system can be improved, the user experience is optimized, smooth performance in resource-intensive tasks can be ensured, meanwhile, energy consumption is reduced, and the system stability is improved. In addition, the cloud mobile phone management system is beneficial to realizing load balancing, supporting high concurrency processing, enabling cloud mobile phone management to be more intelligent, reducing manual intervention requirements, reducing operation cost and improving reliability of a cloud mobile phone system.
Typically, when the target GPU is in a normal operating state, the frequency and duration of use of the applications, and the performance modes adopted by the user to use the applications, including, but not limited to, a running mode, a graphics special effect, a resolution, and a frame rate, may be obtained by obtaining user historical behavior data of the user of the cloud mobile phone, for example, application program use history of the user, particularly, high-load applications with high use frequency in the same time period of the history, such as games or video editing software, and the like. The users are divided into different behavior groups through cluster analysis to identify users possibly needing more resources, the user behaviors are analyzed through time sequences, future resource demands are predicted, key resources with high resource demands are identified through decision trees and random forest construction models, GPU asynchronous APIs are called, and the GPU tasks are recorded and synchronized through events so as to inquire the completion state of the GPU tasks and perform subsequent operations, the predicted key resources of the predicted high-load users are preloaded in the normal state or the low-load running state of the GPU, the system can respond to the operation of the users more quickly, the GPU resources are fully utilized, the waiting time of the users is shortened, the user experience is improved, the overload risk of the GPU is reduced, and the reliability of the cloud mobile phone system is improved.
Typically, when the target GPU is in a normal working state, the current utilization rate of the GPU is obtained and compared with a set threshold value to adjust the working mode, for example, the GPU can be set to be in a low-performance mode when the GPU load is lower than 20%, the GPU load is in a medium-performance mode when the GPU load is 20% -80%, and the GPU load is in a high-performance mode when the GPU load exceeds 80%, in order to ensure the stable operation of the system under the sudden high load condition, a certain proportion of GPU resources can be reserved and not allocated to user tasks, so that enough resources can be ensured to perform quick response under the sudden high load condition, the stability of the cloud mobile phone system in a short time is maintained, the system breakdown or performance reduction caused by overload is avoided, and a warning can be sent to a user to prompt the user that the current resources are tense, and the user is suggested to reduce the intensive operation of resources or optimize the application configuration.
For ease of understanding, specific application scenarios to which embodiments of the present disclosure are applicable will now be described. Fig. 4 is a general flow diagram of an embodiment suitable for use in the specific application scenario of the embodiments of the present disclosure. In the application scene, the method specifically describes the specific operation steps of monitoring the GPU utilization rate of the cloud mobile phone in real time, reporting the information of the cloud mobile phone and the GPU utilization rate to a cloud mobile phone platform when detecting that the GPU utilization rate of the cloud mobile phone exceeds a set alarm standard, informing a cloud mobile phone client to which the cloud mobile phone instance belongs by the cloud mobile phone platform, and timely sensing and processing the problem by the cloud mobile phone client.
Specifically, as shown in fig. 4, the operation control method of the cloud mobile phone is specifically divided into 7 steps, namely, deploying a GPU monitoring program, configuring an alarm standard, monitoring the GPU utilization rate of the current cloud mobile phone in real time, calling back related information to a cloud mobile phone platform, inquiring related information according to unique identification of the cloud mobile phone, returning an inquiry result, notifying all affected cloud mobile phone callbacks to affiliated merchants, and correspondingly processing cloud mobile phones with abnormal GPU utilization rates.
S401, deploying a GPU monitoring program and configuring alarm standards.
Optionally, a GPU monitor may be deployed on the cloud handset, and a GPU alert standard and a notification interface address when an abnormality occurs may be set, for example, the alert standard may be set such that the GPU usage rate within 3s is more than 90%. The notification interface address is used for recalling relevant information to the cloud mobile phone platform in an interface calling mode when the alarm standard is met.
S402, monitoring the GPU utilization rate of the current cloud mobile phone in real time.
After the deployment of the GPU monitoring program is completed, the GPU monitoring program can monitor the utilization rate of the GPU of the current cloud mobile phone in real time according to the preset monitoring frequency.
S403, callback related information to the cloud mobile phone platform.
Specifically, when the GPU monitoring program detects that the GPU usage rate of the current cloud mobile phone meets the set alarm standard, some related information is returned to the cloud mobile phone platform through the notification interface address, wherein the related information comprises a unique identifier (instance IP) of the current cloud mobile phone, the GPU usage rate, an application package name with higher load of the cloud mobile phone, and the like.
S404, inquiring related information according to the unique identification of the cloud mobile phone.
After receiving the callback notification of the GPU monitor program, the cloud mobile phone platform may query, from the instance IP of the current cloud mobile phone to the database, information such as a merchant to which the current cloud mobile phone belongs (also referred to as a cloud mobile phone client), a user ID of the current cloud mobile phone being used, other cloud mobile phones sharing the GPU with the current cloud mobile phone, and the merchant to which the current cloud mobile phone belongs.
S405, returning a query result.
The database can return the queried related information to the cloud mobile phone platform based on the query request of the cloud mobile phone platform.
S406, notifying all affected cloud mobile phone callbacks to the affiliated merchants.
Specifically, the current cloud mobile phone and other cloud mobile phones sharing the GPU with the current cloud mobile phone are both affected cloud mobile phones. The cloud mobile phone platform can inform the merchants to which the affected cloud mobile phones belong of the information of the affected cloud mobile phones with abnormal GPU utilization rate.
S407, carrying out corresponding processing on the cloud mobile phones with abnormal GPU utilization rates.
After the merchant to which the cloud mobile phone belongs receives the GPU exception notification, some targeted processing measures can be taken for the problem, such as killing the application process with higher GPU utilization rate and notifying the downstream user.
According to the technical scheme, the real-time abnormal monitoring and notifying capability of the GPU utilization rate of the cloud mobile phone can be realized, and when the GPU utilization rate on the cloud mobile phone rises, the cloud mobile phone can be timely notified to the merchant to which the cloud mobile phone belongs, so that cloud mobile phone clients of the cloud mobile phone platform can timely perceive the problem and process the problem correspondingly, the situation that the using experience of an actual cloud mobile phone user is influenced due to the fact that the clamping is avoided, and the stability of a cloud mobile phone system is improved.
As an implementation of the operation control method of each cloud mobile phone, the disclosure further provides an optional embodiment of an execution device for implementing the operation control method of each cloud mobile phone.
Fig. 5 is a block diagram of an operation control device of a cloud mobile phone according to an embodiment of the present disclosure. As shown in fig. 5, the device includes a GPU usage monitoring module 501, an associated cloud handset acquisition module 502, and a GPU high load application control module 503, wherein:
the GPU utilization rate monitoring module is used for acquiring the utilization rate of the target GPU corresponding to the monitoring cloud mobile phone in real time;
the associated cloud mobile phone acquisition module is used for acquiring each associated cloud mobile phone sharing the target GPU with the monitoring cloud mobile phone when the utilization rate of the target GPU exceeds the utilization rate threshold value;
The GPU high-load application control module is used for identifying each GPU high-load application in the monitoring cloud mobile phone and each associated cloud mobile phone, and performing operation control on each GPU high-load application according to operation description information of the cloud mobile phone where each GPU high-load application is located.
According to the technical scheme, the real-time abnormal monitoring of the GPU utilization rate of the cloud mobile phone can be achieved, when the utilization rate of the GPU exceeds the utilization rate threshold, all the associated cloud mobile phones sharing the target GPU with the monitoring cloud mobile phone are obtained, all the GPU high-load applications are identified in the monitoring cloud mobile phone and all the associated cloud mobile phones, and according to the operation description information of the cloud mobile phone where the GPU high-load applications are located, the GPU utilization rate of the cloud mobile phone is monitored in real time, when the GPU utilization rate is abnormal, all the GPU high-load applications in the monitoring cloud mobile phone and all the associated cloud mobile phones are identified, the operation control is carried out, and through the real-time monitoring of the GPU utilization rate of the cloud mobile phone, the complicated step of manual inspection of a user is omitted, and potential problems can be found and handled rapidly. The active automatic operation control mechanism ensures the stable operation of the target GPU and improves the reliability of the whole operation of the cloud mobile phone.
Based on the above embodiments, the GPU high load application control module is specifically configured to:
identifying application programs running in a foreground in the monitoring cloud mobile phone and each associated cloud mobile phone respectively, and acquiring the resource occupancy rate of each application program to the target GPU;
And determining the application program with the resource occupancy rate exceeding the occupancy rate threshold value as the GPU high-load application.
Based on the above embodiments, the GPU high load application control module is further configured to:
and re-distributing resources in the target GPU for each GPU high-load application according to the network transmission speed between the cloud mobile phone where each GPU high-load application is located and each entity device which is currently in communication connection.
Based on the above embodiments, the GPU high load application control module is further configured to:
acquiring the current network transmission speed between a cloud mobile phone where a current GPU high-load application is located and entity equipment which is currently in communication connection in real time, and acquiring the current frame rate of the current GPU high-load application;
and if the current frame rate is greater than the current network transmission speed, adjusting the current frame rate to be matched with the current network transmission speed.
Based on the above embodiments, the GPU high load application control module further includes:
The priority weight information acquisition unit is used for acquiring the priority weight information of the cloud mobile phone where each GPU high-load application is located;
the target cloud mobile phone screening unit is used for screening at least one target cloud mobile phone according to the order of the priority weight information from low to high;
the resource reduction unit is used for reducing resources in the target GPUs distributed for the GPU high-load applications in the target cloud mobile phones by adopting a preset strategy.
On the basis of the above embodiments, the resource reducing unit is specifically configured to implement at least one of the following functions:
Adjusting the operation mode of the GPU high-load application in the target cloud mobile phone to a very simple mode;
Closing a high-consumption graphic special effect in the GPU high-load application in the target cloud mobile phone;
and reducing the resolution or frame rate of the GPU high-load application in the target cloud mobile phone.
On the basis of the foregoing embodiments, the priority weight information obtaining unit is specifically configured to:
acquiring cloud mobile phone users to which the cloud mobile phones where the GPU high-load applications are located belong;
acquiring historical cloud mobile phone use frequency information and user grade information of each cloud mobile phone user;
According to the historical cloud mobile phone use frequency information and the user grade information, calculating to obtain priority weight information of the cloud mobile phone where each GPU high-load application is located;
the history cloud mobile phone use frequency information comprises at least one of history use duration, history cloud mobile phone use frequency and history cloud mobile phone use high-frequency time period.
On the basis of the above embodiments, the device further includes:
The utilization rate trend determining module is used for determining whether the utilization rate of the target GPU is in an ascending trend according to the historical utilization rate of the target GPU obtained by at least two last monitoring when the utilization rate of the target GPU is determined not to exceed the utilization rate threshold;
the historical behavior data acquisition module is used for acquiring user historical behavior data of the cloud mobile phone users to which the monitoring cloud mobile phone and each associated cloud mobile phone belong if the utilization rate of the target GPU shows an ascending trend;
the overrun time prediction module is used for predicting the predicted occurrence time of the overrun of the GPU use of the target GPU according to the historical behavior data of each user;
and the expected occurrence time providing module is used for providing the expected occurrence time for each cloud mobile phone user to perform autonomous decision control if the difference between the expected occurrence time and the current system time is smaller than a preset difference threshold.
On the basis of the above embodiments, the device further includes:
The intelligent preloading module is used for acquiring user history behavior data of the cloud mobile phone users of the monitoring cloud mobile phone and the associated cloud mobile phones when the utilization rate of the target GPU does not exceed the utilization rate threshold value, predicting resources required by the cloud mobile phone users of the monitoring cloud mobile phone and the associated cloud mobile phones, and utilizing the residual resources to simultaneously perform intelligent preloading of a plurality of tasks through multithreading and asynchronous processing, or
And the running mode adjusting module is used for dynamically adjusting the running mode of the GPU according to the current utilization rate of the target GPU and a preset threshold value when the utilization rate of the target GPU is determined not to exceed the utilization rate threshold value.
The product can execute the method provided by any embodiment of the disclosure, and has the corresponding functional modules and beneficial effects of executing the method.
In the technical scheme of the disclosure, the related processes of collecting, storing, using, processing, transmitting, providing, disclosing and the like of the personal information of the user accord with the regulations of related laws and regulations, and the public order colloquial is not violated.
According to embodiments of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.
Fig. 6 illustrates a schematic block diagram of an example electronic device 600 that may be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the disclosure described and/or claimed herein.
As shown in fig. 6, the apparatus 600 includes a computing unit 601 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 602 or a computer program loaded from a storage unit 608 into a Random Access Memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the device 600 may also be stored. The computing unit 601, ROM 602, and RAM 603 are connected to each other by a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.
Various components in the device 600 are connected to the I/O interface 605, including an input unit 606, e.g., keyboard, mouse, etc., an output unit 607, e.g., various types of displays, speakers, etc., a storage unit 608, e.g., magnetic disk, optical disk, etc., and a communication unit 609, e.g., network card, modem, wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.
The computing unit 601 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 601 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the respective methods and processes described above, for example, an operation control method of a cloud cellular phone. For example, in some embodiments, the method of controlling operation of a cloud handset may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and/or installed onto the device 600 via the ROM 602 and/or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the operation control method of the cloud cellular phone described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the operation control method of the cloud handset in any other suitable manner (e.g., by means of firmware).
That is, the utilization rate of the target GPU corresponding to the monitoring cloud mobile phone is obtained in real time;
when the utilization rate of the target GPU exceeds the utilization rate threshold, acquiring each associated cloud mobile phone sharing the target GPU with the monitoring cloud mobile phone;
identifying each GPU high-load application in the monitoring cloud mobile phone and each associated cloud mobile phone, and performing operation control on each GPU high-load application according to operation description information of the cloud mobile phone where each GPU high-load application is located.
Various implementations of the systems and techniques described here above can be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), complex Programmable Logic Devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include being implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be a special or general purpose programmable processor, operable to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user, for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback), and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a Local Area Network (LAN), a Wide Area Network (WAN), and the Internet.
The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical hosts and VPS service are overcome. The server may also be a server of a distributed system or a server that incorporates a blockchain.
Artificial intelligence is the discipline of studying the process of making a computer mimic certain mental processes and intelligent behaviors (e.g., learning, reasoning, thinking, planning, etc.) of a person, both hardware-level and software-level techniques. The artificial intelligence hardware technology generally comprises technologies such as a sensor, a special artificial intelligence chip, cloud computing, distributed storage, big data processing and the like, and the artificial intelligence software technology mainly comprises a computer vision technology, a voice recognition technology, a natural language processing technology, a machine learning/deep learning technology, a big data processing technology, a knowledge graph technology and the like.
Cloud computing (cloud computing) refers to a technical system that a shared physical or virtual resource pool which is elastically extensible is accessed through a network, resources can comprise servers, operating systems, networks, software, applications, storage devices and the like, and resources can be deployed and managed in an on-demand and self-service mode. Through cloud computing technology, high-efficiency and powerful data processing capability can be provided for technical application such as artificial intelligence and blockchain, and model training.
It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present disclosure may be performed in parallel, sequentially, or in a different order, provided that the desired results of the technical solutions provided by the present disclosure are achieved, and are not limited herein.
The above detailed description should not be taken as limiting the scope of the present disclosure. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present disclosure are intended to be included within the scope of the present disclosure.