Performance data processing method and device of graphic processing unit, equipment and medium

Abstract

The embodiment of the invention provides a performance data processing method and device for a graphic processing unit, equipment and a medium, and the method comprises the steps: obtaining an architecture type of the graphic processing unit in terminal equipment, and under the condition that the architecture type is a first preset architecture, carrying out the performance data processing of the graphic processing unit; screening out a first target program matched with a first preset architecture from a plurality of set candidate programs, sending the first target program to the terminal equipment, and determining a target data item list according to a target performance information list, a target index formula list and at least one set first target performance index, and sending the target data item list to the terminal equipment, and calculating to obtain a first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list and the first performance data. The scheme can be suitable for graphic processing units of different architecture types, the expected performance index value is accurately calculated, complex manual operation is not needed, and the universality is high.

Description

Technical Field

[0001] Embodiments of the present application relate to the field of computer technologies, and in particular, to a method, apparatus, device, and medium for processing performance data of a graphics processing unit. Background Art

[0002] With the continuous improvement of the hardware performance of terminal devices, mobile applications have become increasingly complex, and users' expectations for the performance of mobile applications are also constantly increasing. Among them, in terminal devices, the graphics processing unit is responsible for various rendering tasks to accelerate operations such as interface drawing, animation effects, and video decoding. Therefore, the performance of the graphics processing unit directly affects the fluency of mobile applications and the user experience. Especially for scenarios that rely on graphics rendering, such as live streaming, video processing, and games, the performance of the graphics processing unit is particularly crucial. With the increasing requirements of mobile applications for graphics complexity and frame rate, monitoring and optimizing the performance of the graphics processing unit has become a key step in improving application fluency and reducing power consumption, which is beneficial to subsequent code optimization and user experience improvement.

[0003] In related technologies, developers need to select a specific performance monitoring tool according to the specific architecture type of the graphics processing unit, and manually configure and operate the performance monitoring tool for performance data processing to monitor the performance status of the graphics processing unit. The operation complexity is high and the generality is poor, which needs to be improved. Summary of the Invention

[0004] Embodiments of the present application provide a method, apparatus, device, and medium for processing performance data of a graphics processing unit, which solve the problems in related technologies that it is necessary to manually select a specific performance monitoring tool according to the specific architecture type of the graphics processing unit, and manually configure and operate the performance monitoring tool for performance data processing, with high operation complexity and poor generality. It can effectively adapt to graphics processing units of different architecture types, accurately calculate the expected performance index value, without complex manual operations, and has strong generality. And the performance index value can be displayed on the set monitoring interface, so as to facilitate developers to monitor the performance status of the graphics processing unit.

[0005] In a first aspect, embodiments of the present application provide a method for processing performance data of a graphics processing unit, the method including:

[0006] Obtain the architecture type of the graphics processing unit in the terminal device. When the architecture type is a first preset architecture, screen out a first target program that matches the first preset architecture from a set of multiple candidate programs, and send the first target program to the terminal device so that the terminal device runs the first target program;

[0007] Request the terminal device to obtain target list information, so that the terminal device filters out the target list information that matches the model of the graphics processing unit from multiple sets of candidate list information based on the first target program, and feeds back the target list information, where the target list information includes a target performance information list and a target index formula list;

[0008] Determine a target data item list according to the target performance information list, the target index formula list, and at least one first target performance index set, and send the target data item list to the terminal device, so that the terminal device collects first performance data for the first performance data item in the target data item list based on the first target program, and feeds back the first performance data;

[0009] Calculate the first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list, and the first performance data.

[0010] In a second aspect, an embodiment of the present application further provides a performance data processing device for a graphics processing unit, and the device includes:

[0011] A first target program deployment module, configured to obtain the architecture type of the graphics processing unit in the terminal device, and when the architecture type is a first preset architecture, filter out a first target program that matches the first preset architecture from multiple sets of candidate programs, and send the first target program to the terminal device, so that the terminal device runs the first target program;

[0012] A list information acquisition module, configured to request the terminal device to obtain target list information, so that the terminal device filters out the target list information that matches the model of the graphics processing unit from multiple sets of candidate list information based on the first target program, and feeds back the target list information, where the target list information includes a target performance information list and a target index formula list;

[0013] A first performance data acquisition module, configured to determine a target data item list according to the target performance information list, the target index formula list, and at least one first target performance index set, and send the target data item list to the terminal device, so that the terminal device collects first performance data for the first performance data item in the target data item list based on the first target program, and feeds back the first performance data;

[0014] A first performance index calculation module, configured to calculate the first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list, and the first performance data.

[0015] In a third aspect, an embodiment of the present application further provides a performance data processing device for a graphics processing unit, the device comprising:

[0016] One or more processors;

[0017] A storage device configured to store one or more programs,

[0018] When the one or more programs are executed by the one or more processors, the one or more processors implement the performance data processing method for the graphics processing unit described in the embodiments of the present application.

[0019] In a fourth aspect, an embodiment of the present application further provides a non-volatile storage medium storing computer-executable instructions, the computer-executable instructions being configured to execute the performance data processing method for the graphics processing unit described in the embodiments of the present application when executed by a computer processor.

[0020] In a fifth aspect, an embodiment of the present application further provides a computer program product, the computer program product comprising a computer program, the computer program being stored in a computer-readable storage medium, and at least one processor of a device reads and executes the computer program from the computer-readable storage medium, so that the device executes the performance data processing method for the graphics processing unit described in the embodiments of the present application.

[0021] In the embodiments of the present application, by obtaining the architecture type of the graphics processing unit in the terminal device, when the architecture type is the first preset architecture, the first target program that matches the first preset architecture is screened out from the set of multiple candidate programs, and the first target program is sent to the terminal device so that the terminal device runs the first target program; the terminal device is requested to obtain the target list information, so that the terminal device screens out the target list information that matches the model of the graphics processing unit from the set of multiple candidate list information based on the first target program and feeds back the target list information, where the target list information includes a target performance information list and a target index formula list; a target data item list is determined according to the target performance information list, the target index formula list, and at least one first target performance index set, and the target data item list is sent to the terminal device so that the terminal device collects the first performance data of the first performance data item in the target data item list based on the first target program and feeds back the first performance data; according to the target performance information list, the target index formula list, and the first performance data, the first performance index value corresponding to the first target performance index is calculated. In the above solution, by determining the corresponding target program according to the different architecture types of the graphics processing unit, a matching target program can be deployed on the terminal device for different architecture types, so that the terminal device can feed back the relevant information for calculating the performance index based on the target program. By requesting the terminal device to obtain the target list information, the target data item list required for calculating the first target performance index can be determined according to the model of the image processing unit, so that the terminal device can effectively collect and feed back the first performance data related to the first target performance index. By calculating the first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list, and the first performance data, it can effectively adapt to the graphics processing units of different architecture types, accurately calculate the expected performance index value, without complex manual operations, and has strong versatility. And this performance index value can be used to be displayed on the set monitoring interface, which is beneficial for developers to monitor the performance status of the graphics processing unit. Description of the Drawings

[0022] Figure 1 It is a flowchart of a method for processing performance data of a graphics processing unit provided by an embodiment of the present application;

[0023] Figure 2 It is a flowchart of a method for processing performance data that includes a process of calculating index values for different architecture types of a graphics processing unit;

[0024] Figure 3 It is a flowchart of a method for processing performance data that includes a process of determining a target data item list;

[0025] Figure 4Flowchart of a performance data processing method provided by an embodiment of the present application, which includes a process of calculating a first performance index value corresponding to a first target performance index;

[0026] Figure 5 Flowchart of a performance data processing method provided by an embodiment of the present application, which includes a process of checking whether a first performance index value meets a reference index range;

[0027] Figure 6 Flowchart of a performance data processing method provided by an embodiment of the present application, which includes a process of comparing an average index value with a historical average index value;

[0028] Figure 7 Block diagram of a performance data processing device for a graphics processing unit provided by an embodiment of the present application;

[0029] Figure 8 Schematic structural diagram of a performance data processing device for a graphics processing unit provided by an embodiment of the present application. Detailed implementation manners

[0030] The following further elaborates on the embodiments of the present application in conjunction with the accompanying drawings and examples. It can be understood that the specific embodiments described herein are only used to explain the embodiments of the present application, rather than limiting the embodiments of the present application. Additionally, it should be noted that for the sake of description, only parts related to the embodiments of the present application are shown in the drawings, rather than all structures.

[0031] Terms such as "first" and "second" in the specification and claims of the present application are used to distinguish similar objects, rather than to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are usually of the same category, and do not limit the number of objects. For example, the first object can be one or multiple. In addition, "and / or" in the specification and claims indicates at least one of the connected objects, and the character " / " generally represents an "or" relationship between the associated objects before and after.

[0032] The performance data processing method for a graphics processing unit provided by an embodiment of this application can be used to deploy a matching target program in a terminal device for different architecture types of the graphics processing unit, and calculate metric values based on the performance data fed back by the terminal device, without complex manual operations, which is conducive to developers monitoring the performance status of the graphics processing unit. Related application scenarios include performance testing of the graphics processing unit, application development, etc. The several application scenarios listed above are only exemplary and explanatory. In actual applications, the performance data processing method for this graphics processing unit can also be used in other scenarios, and the embodiments of this application do not limit this.

[0033] For the performance data processing method for a graphics processing unit provided by an embodiment of this application, the execution entity of each step can be a computer device, which refers to any electronic device with data calculation, processing, and storage capabilities, such as a PC (Personal Computer). The embodiments of this application do not limit this.

[0034] Figure 1 It is a flowchart of a performance data processing method for a graphics processing unit provided by an embodiment of this application. The execution entity of this performance data processing method for the graphics processing unit can be a computer device, which can be connected to the terminal device in a wired or wireless manner. The terminal device can be a mobile phone, a tablet computer, etc. that needs to monitor the performance status of the graphics processing unit. As Figure 1 shown, it includes the following steps:

[0035] Step S101: Obtain the architecture type of the graphics processing unit in the terminal device. When the architecture type is the first preset architecture, screen out the first target program that matches the first preset architecture from multiple candidate programs set, and send the first target program to the terminal device so that the terminal device runs the first target program.

[0036] Among them, the graphics processing unit is a processor specifically used to accelerate graphics rendering and computing tasks. The architecture type of the graphics processing unit can be the Adreno architecture, the ARM Mali architecture, etc., which is not limited in this application. Since the performance data acquisition process and index calculation methods corresponding to graphics processing units of different architecture types are different, this embodiment pre-sets multiple candidate programs for deploying on the terminal device to collect performance data of the graphics processing unit and feedback index calculation information. Among them, the first preset architecture can be the ARM Mali architecture, etc. The performance data that can be collected and the index calculation formulas supported by different models of graphics processing units under this first preset architecture are different. The specific models can be Mali-G77, Mali-G78, etc., which are not limited in this application. Therefore, after the computer device determines that the graphics processing unit is the first preset architecture, it is necessary to deploy the first target program on the terminal device so that the terminal device can run the first target program, and then identify the request information or instruction information sent by the computer device based on the first target program, so as to feedback performance information, index calculation formulas and performance data that match the current model of the graphics processing unit of the terminal device. The performance information can include different counter items and different performance events set specifically. The performance data can include data representing the performance of the graphics processing unit statistically based on counters. For example, non-fragment queue active cycles, fragment queue active cycles, etc. Among them, multiple candidate list information is pre-set in the first target program. The candidate list information includes a performance information list and an index formula list. The performance information list can record data items currently set for a certain model related to representing performance. For example, counter items and performance events in different performance dimensions. The index formula list can record the index calculation formulas currently supported by a certain model.

[0037] Step S102: Request the terminal device to obtain target list information, so that the terminal device filters out the target list information that matches the model of the graphics processing unit from the multiple candidate list information set based on the first target program, and feedbacks the target list information, where the target list information includes a target performance information list and a target index formula list.

[0038] Among them, the computer device requesting the terminal device to obtain target list information can enable the terminal device to identify the model of the image processing unit based on the first target program, filter out the target list information that matches the model from the multiple candidate list information set, and feedback it to the computer device, so that the computer device can determine the target data items that need to be collected by the terminal device.

[0039] Step S103: Determine a list of target data items according to the list of target performance information, the list of target metric formulas, and at least one set first target performance metric, and send the list of target data items to the terminal device, so that the terminal device collects first performance data for the first performance data items in the list of target data items based on the first target program and feeds back the first performance data.

[0040] Among them, the first target performance metric can be determined by the developer according to the monitoring requirements of the graphics processing unit in the actual application scenario. The metric calculation formulas corresponding to different first target performance metrics are different. For example, the first target performance metric can be non-segment queue utilization rate, segment queue utilization rate, tessellator utilization rate, etc., which are not limited in this application. Therefore, it is necessary to extract the metric calculation formula corresponding to the first target performance metric from the list of target metric formulas. For example, to calculate the non-segment queue utilization rate, it can be obtained by dividing the active cycle number of the non-segment queue in the unit active cycle by the overall unit active cycle number. To calculate the tessellator utilization rate, it can be obtained by dividing the active cycle number of the tessellator in the GPU active cycle by the overall unit active cycle number, which are not limited in this application. And, there is an associated correspondence between the calculation variables in the metric calculation formula and the data items related to the performance representation of the graphics processing unit. Therefore, it is necessary to determine the target data items for which performance data needs to be collected based on the calculation variables and combine them to obtain a list of target data items. By way of example, the target data item can be the non-segment queue active cycle, segment queue active cycle, etc. counted based on a counter, which are not limited in this application. The terminal device can collect first performance data corresponding to the list of target data items based on the first target program and feed it back to the computer device.

[0041] Step S104: Calculate the first performance metric value corresponding to the first target performance metric according to the list of target performance information, the list of target metric formulas, and the first performance data.

[0042] Among them, the metric calculation formula corresponding to the first target performance metric can be extracted from the list of target metric formulas, and the performance data corresponding to each calculation variable in the metric calculation formula can be matched based on the list of target performance information, so as to calculate the first performance metric value corresponding to the first target performance metric. Optionally, the first performance metric value can be displayed on the set monitoring interface with time as the horizontal axis, which is convenient for the developer to observe the specific change trend of the performance metric value over time, and to observe the change in the operating state of the graphics processing unit caused by the terminal device receiving user operations, and to reasonably evaluate whether the current operating state of the graphics processing unit meets the expectations and whether code optimization is required.

[0043] As described above, by obtaining the architecture type of the graphics processing unit in the terminal device, when the architecture type is the first preset architecture, the first target program that matches the first preset architecture is screened out from the set of multiple candidate programs, and the first target program is sent to the terminal device so that the terminal device runs the first target program; request the terminal device to obtain the target list information, so that the terminal device screens out the target list information that matches the model of the graphics processing unit from the set of multiple candidate list information based on the first target program, and feedback the target list information, where the target list information includes the target performance information list and the target index formula list; determine the target data item list according to the target performance information list, the target index formula list, and at least one first target performance index set, and send the target data item list to the terminal device so that the terminal device collects the first performance data of the first performance data item in the target data item list based on the first target program and feedbacks the first performance data; calculate the first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list, and the first performance data. In the above solution, by determining the corresponding target program according to the different architecture types of the graphics processing unit, the target program that matches can be deployed on the terminal device for different architecture types, so that the terminal device can feedback the relevant information for calculating the performance index based on the target program. By requesting the terminal device to obtain the target list information, the target data item list required for calculating the first target performance index can be determined according to the model of the image processing unit, so that the terminal device can effectively collect and feedback the first performance data related to the first target performance index. By calculating the first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list, and the first performance data, it can effectively adapt to the graphics processing unit of different architecture types, accurately calculate the expected performance index value, without complex manual operations, with strong versatility, and this performance index value can be used to display on the set monitoring interface, which is beneficial for developers to monitor the performance status of the graphics processing unit.

[0044] Figure 2 The figure is a flowchart of a performance data processing method provided by an embodiment of the present application, which includes a process of calculating index values for different architecture types of a graphics processing unit. As Figure 2 shown, it includes the following steps (the serial numbers of the steps are only for clear description and do not aim to limit the execution order of the steps, the same below):

[0045] Step S201: Obtain the architecture type of the graphics processing unit in the terminal device.

[0046] Step S202: When the architecture type is the first preset architecture, screen out the first target program that matches the first preset architecture from the set of multiple candidate programs, and send the first target program to the terminal device so that the terminal device runs the first target program.

[0047] Step S203: Request the terminal device to obtain the target list information, so that the terminal device filters out the target list information that matches the model of the graphics processing unit from the set multiple candidate list information based on the first target program, and feeds back the target list information, where the target list information includes a target performance information list and a target index formula list.

[0048] Step S204: Determine the target data item list according to the target performance information list, the target index formula list, and at least one first target performance index set, and send the target data item list to the terminal device, so that the terminal device collects the first performance data for the first performance data item in the target data item list based on the first target program, and feeds back the first performance data.

[0049] Step S205: Calculate the first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list, and the first performance data.

[0050] Step S206: In the case where the architecture type is the second preset architecture, filter out the second target program that matches the second preset architecture from the set multiple candidate programs, and send the second target program to the terminal device, so that the terminal device runs the second target program.

[0051] Among them, the second preset architecture may be the Adreno architecture, etc. This second preset architecture does not need to distinguish specific performance information and index calculation formulas for different models. Therefore, after the computer device determines that the graphics processing unit is the second preset architecture, it is necessary to deploy the second target program on the terminal device so that the terminal device can run the second target program, and then feedback performance data based on the second target program. Among them, the computer device is pre-set with an index calculation formula corresponding to the second target performance index, which can be used to calculate the corresponding second performance index value based on the second performance data, and at least one second performance data item is pre-set in the second target program, for example, clock pulse count value, operation cycle count value, etc., which are not limited in this application.

[0052] Step S207: Receive the second performance data collected by the terminal device corresponding to the set second performance data item based on the second target program.

[0053] Among them, after the terminal device runs the second target program, it can collect the second performance data for the corresponding second performance data item based on the start instruction information sent by the computer device, and feed it back to the computer device.

[0054] Step S208: Calculate the second performance index value corresponding to the at least one second target performance index according to the index calculation formula corresponding to the set second target performance index and the second performance data.

[0055] Among them, the second target performance index can be the current clock frequency, unit utilization rate, etc. By substituting the second performance data into the index calculation formula, the second performance index value corresponding to the second performance index can be calculated accordingly. For example, the current clock frequency can be obtained by dividing the clock pulse count value by the statistical time interval. The unit utilization rate can be obtained by first dividing the running cycle count value by the statistical time interval to get the running frequency, and then dividing the running frequency by the maximum allowable running frequency. This application does not make any limitations here.

[0056] As described above, by allocating corresponding target programs to the terminal device according to different architecture types of the graphics processing unit, it is possible to effectively be compatible with different architecture types for collecting performance data and calculating index values, providing efficient support for the performance optimization and evaluation of the graphics processing unit.

[0057] Figure 3 This is a flowchart of a performance data processing method provided by an embodiment of this application, which includes a process of determining a target data item list. The target performance information list includes a performance counter list and a performance event information list. The target data item list includes a target counter list. As Figure 3 shown, it includes the following steps:

[0058] Step S301: Obtain the architecture type of the graphics processing unit in the terminal device. When the architecture type is the first preset architecture, screen out the first target program that matches the first preset architecture from the set multiple candidate programs, and send the first target program to the terminal device so that the terminal device runs the first target program.

[0059] Step S302: Request the terminal device to obtain the target list information, so that the terminal device screens out the target list information that matches the model of the graphics processing unit from the set multiple candidate list information based on the first target program, and feedback the target list information. Among them, the target list information includes a target performance information list and a target index formula list. The target performance information list includes a performance counter list and a performance event information list. The target data item list includes a target counter list.

[0060] Among them, the performance counter list may be counter items that count and statistically analyze multiple performance dimensions of the graphics processing unit. Different counter items are correspondingly associated with different performance events. The specific counter items may include non-fragment queue active cycles, fragment queue active cycles, tessellator active cycles, etc., which are not limited in this application. The performance event information list may record the association relationships between the performance events corresponding to different unit processing operations and the counter items. For example, unit activity events, non-fragment queue activity events, processing, fragment queue activity events, etc., which are not limited in this application.

[0061] Step S303: Screen out the target index formula corresponding to at least one set first target performance index from the target index formula list, extract the target performance event identifier from the performance event information list based on the variable name in the target index formula, extract the target counter item from the performance counter list according to the target performance event identifier, and combine the target counter items to obtain the target counter list.

[0062] Among them, after screening out the target index formula corresponding to at least one set first target performance index from the target index formula list, the variable name of the calculation variable that needs to substitute performance data in the target index formula can be determined. Since this variable name is different from the actual name of the counter item, it is necessary to query through the performance event information list. Based on this variable name, the corresponding target performance event identifier can be extracted from the performance event information list, and based on this target performance event identifier, the corresponding target counter item can be extracted from the performance counter list, so as to determine the target counter list for collecting the first performance data.

[0063] Step S304: Send the target counter list to the terminal device, so that the terminal device collects the first performance data for the target counter items in the target counter list based on the first target program and feeds back the first performance data.

[0064] Step S305: Calculate the first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list, and the first performance data.

[0065] As described above, by determining the target counter list for calculating the first target performance index based on the target index formula list, the performance event information list, and the performance counter list, it is possible to effectively determine the target counter item actually corresponding to the graphics processing unit based on the variable name of the target index formula, which is beneficial for the terminal device to accurately collect the first performance data for calculating the first target performance index based on the first target program.

[0066] Figure 4This is a flowchart of a performance data processing method provided by an embodiment of the present application, which includes a process of calculating a first performance index value corresponding to a first target performance index. As Figure 4 shown, the method includes the following steps:

[0067] Step S401: Obtain the architecture type of the graphics processing unit in the terminal device. When the architecture type is the first preset architecture, screen out a first target program that matches the first preset architecture from multiple set candidate programs, and send the first target program to the terminal device so that the terminal device runs the first target program.

[0068] Step S402: Request the terminal device to obtain target list information, so that the terminal device screens out target list information that matches the model of the graphics processing unit from multiple set candidate list information based on the first target program, and feedbacks the target list information, where the target list information includes a target performance information list and a target index formula list.

[0069] Step S403: Determine a target data item list according to the target performance information list, the target index formula list, and at least one set first target performance index, and send the target data item list to the terminal device so that the terminal device collects first performance data for a first performance data item in the target data item list based on the first target program, and feedbacks the first performance data.

[0070] Step S404: Screen out a target index formula corresponding to at least one set first target performance index from the target index formula list, determine a target performance data item corresponding to the variable name in the target index formula according to the target performance information list, extract target performance data corresponding to the target performance data item from the first performance data, and calculate a first performance index value corresponding to the first target performance index based on the target performance data and the target index formula.

[0071] Among them, after screening out a target index formula corresponding to at least one set first target performance index from the target index formula list, since the variable name in the target index formula is different from the actual name of the target performance data item, the target performance data item corresponding to the variable name can be queried through the target performance information list, and the target performance data corresponding to the target performance data item is extracted from the first performance data, and then the target performance data is substituted into the corresponding target index formula for calculation to obtain the corresponding first performance index value.

[0072] As described above, based on the target performance information list, the variable names of the target index formula can be effectively corresponded to match the corresponding target performance data items, and then the target performance data corresponding to the target performance data items in the first performance data can be determined, which is beneficial to accurately calculate the first performance index value corresponding to the first target performance index of the graphics processing unit under the first preset architecture, and provide reliable information for monitoring the performance status of the graphics processing unit.

[0073] Figure 5 The figure is a flowchart of a performance data processing method provided by an embodiment of the present application, which includes a process of checking whether the first performance index value meets the reference index range. As Figure 5 shown, it includes the following steps:

[0074] Step S501: Obtain the architecture type of the graphics processing unit in the terminal device. When the architecture type is the first preset architecture, screen out the first target program that matches the first preset architecture from the set multiple candidate programs, and send the first target program to the terminal device so that the terminal device runs the first target program.

[0075] Step S502: Request the terminal device to obtain the target list information, so that the terminal device screens out the target list information that matches the model of the graphics processing unit from the set multiple candidate list information based on the first target program, and feedbacks the target list information, where the target list information includes the target performance information list and the target index formula list.

[0076] Step S503: Determine the target data item list according to the target performance information list, the target index formula list, and at least one first target performance index set, and send the target data item list to the terminal device, so that the terminal device collects the first performance data for the first performance data item in the target data item list based on the first target program, and feedbacks the first performance data;

[0077] Step S504: Calculate the first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list, and the first performance data.

[0078] Step S505: Obtain the application identifier of the application program currently running on the terminal device, calculate the average index value based on the first performance index value within the preset time range, compare the average index value with the reference index range set for the corresponding application identifier, and generate the first optimization prompt information when the average index value is not within the reference index range.

[0079] Among them, by obtaining the application identifier of the application currently running on the terminal device, it is possible to determine which application is currently running on the terminal device. Since different applications have different performance requirements for the graphics processing unit, for live broadcast applications, video processing applications, game applications, etc., the graphics processing unit needs to reach the expected performance state to ensure the normal operation of the aforementioned applications. Therefore, for different applications, developers can preset the reference index range of the corresponding performance index in advance, which is used to determine whether the graphics processing unit is in the expected performance state when the terminal device runs the application. First, since the calculation of the first performance index value is related to the cycle time interval of the terminal device collecting and feeding back the first performance data. For example, the cycle time interval can be 1 s, then the first performance index value can be calculated based on the first performance data collected per second. To prevent abnormal fluctuations in the index, the average index value can be calculated based on the first performance index values within a preset time range, which plays a role in data smoothing. The preset time range can be 5 s, 10 s, etc., and this application does not make a limitation here. Then, the average index value is compared with the reference index range. If the average index value is within the reference index range, it can be considered that the graphics processing unit reaches the expected performance state. If the average index value is not within the reference index range, it can be considered that the graphics processing unit does not reach the expected performance state, and a first optimization prompt message needs to be generated to remind the developer to perform relevant code optimization. Optionally, the first optimization prompt message can be displayed on the set monitoring interface.

[0080] As described above, by obtaining the application identifier of the application currently running on the terminal device, the specific application currently running on the terminal device can be identified. By comparing the calculated average index value with the reference index range corresponding to the application identifier, it can be effectively determined whether the operating state of the graphics processing unit changes as required according to the needs of the specific application and reaches the expected performance state. When the average index value is not within the reference index range, a first optimization prompt message is generated to timely remind the developer that there may be a problem with the operating state of the graphics processing unit and further optimization is required.

[0081] Figure 6 The figure is a flowchart of a performance data processing method provided by an embodiment of this application, which includes a process of comparing an average index value with a historical average index value. As Figure 6 shown, it includes the following steps:

[0082] Step S601: Obtain the architecture type of the graphics processing unit in the terminal device. When the architecture type is the first preset architecture, screen out the first target program that matches the first preset architecture from multiple candidate programs set, and send the first target program to the terminal device so that the terminal device runs the first target program.

[0083] Step S602: Request the terminal device to obtain the target list information, so that the terminal device filters out the target list information that matches the model of the graphics processing unit from the set multiple candidate list information based on the first target program, and feedbacks the target list information, where the target list information includes a target performance information list and a target index formula list.

[0084] Step S603: Determine the target data item list according to the target performance information list, the target index formula list, and at least one set first target performance indicator, and send the target data item list to the terminal device, so that the terminal device collects the first performance data of the first performance data item in the target data item list based on the first target program, and feedbacks the first performance data;

[0085] Step S604: Calculate the first performance indicator value corresponding to the first target performance indicator according to the target performance information list, the target index formula list, and the first performance data.

[0086] Step S605: Obtain the application identifier of the application currently running on the terminal device, and calculate the average indicator value based on the first performance indicator values within a preset time range.

[0087] Step S606: Compare the average indicator value with the reference indicator range set for the corresponding application identifier. If the average indicator value is not within the reference indicator range, generate a first optimization prompt message.

[0088] Step S607: Compare the average indicator value with the historical average indicator value recorded for the corresponding application identifier, where the historical average indicator value is calculated based on the first performance indicator value corresponding to the previous run of the application on the terminal device. If the difference between the average indicator value and the historical average indicator value is not within the preset numerical range, generate a second optimization prompt message.

[0089] Among them, in order to determine whether the operating state of the graphics processing unit is stable when the terminal device runs the same application multiple times, the average indicator value calculated when the terminal device currently runs this application can be subtracted from the historical average indicator value calculated when the terminal device last ran this application to determine the difference between the two calculation results. If the difference between the average indicator value and the historical average indicator value is within the preset numerical range, it can be considered that the performance state of the graphics processing unit is within a reasonable fluctuation range. If the difference between the average indicator value and the historical average indicator value is not within the preset numerical range, it can be considered that the performance state of the graphics processing unit has abnormal fluctuations, and a second optimization prompt message can be generated to remind the developer to perform relevant code optimization. Optionally, the second optimization prompt message can be displayed on the set monitoring interface.

[0090] As described above, by comparing the calculated average metric value with the historical average metric value corresponding to the application identifier, it is possible to effectively determine whether the performance state of the graphics processing unit is stable when the terminal device runs the same application program multiple times. When the difference between the average metric value and the historical average metric value is not within the preset numerical range, a second optimization prompt message is generated to timely remind the developer that there may be a problem with the running state of the graphics processing unit and further optimization and improvement are required.

[0091] Figure 7 FIG. 4 is a structural block diagram of a performance data processing device for a graphics processing unit provided by an embodiment of the present application. The device is configured to execute the performance data processing method for the graphics processing unit provided in the above embodiment, and has functional modules and beneficial effects corresponding to the execution of the method. As Figure 7 shown, the device includes:

[0092] A first target program deployment module 101, configured to obtain the architecture type of the graphics processing unit in the terminal device. When the architecture type is the first preset architecture, screen out a first target program that matches the first preset architecture from a plurality of candidate programs set, and send the first target program to the terminal device so that the terminal device runs the first target program;

[0093] A list information acquisition module 102, configured to request the terminal device to obtain target list information, so that the terminal device screens out target list information that matches the model of the graphics processing unit from a plurality of candidate list information sets based on the first target program, and feedbacks the target list information. The target list information includes a target performance information list and a target index formula list;

[0094] A first performance data acquisition module 103, configured to determine a target data item list according to the target performance information list, the target index formula list, and at least one first target performance index set, and send the target data item list to the terminal device so that the terminal device collects first performance data for the first performance data item in the target data item list based on the first target program, and feedbacks the first performance data;

[0095] A first performance index calculation module 104, configured to calculate a first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list, and the first performance data.

[0096] As described above, by obtaining the architecture type of the graphics processing unit in the terminal device, when the architecture type is the first preset architecture, the first target program that matches the first preset architecture is screened out from the set of multiple candidate programs, and the first target program is sent to the terminal device so that the terminal device runs the first target program; request the terminal device to obtain the target list information, so that the terminal device screens out the target list information that matches the model of the graphics processing unit from the set of multiple candidate list information based on the first target program, and feedback the target list information, where the target list information includes a target performance information list and a target index formula list; determine the target data item list according to the target performance information list, the target index formula list, and at least one first target performance index set, and send the target data item list to the terminal device so that the terminal device collects the first performance data of the first performance data item in the target data item list based on the first target program, and feedback the first performance data; calculate the first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list, and the first performance data. In the above solution, by determining the corresponding target program according to the different architecture types of the graphics processing unit, the target program that matches the architecture type can be deployed on the terminal device, so that the terminal device can feedback the relevant information for calculating the performance index based on the target program. By requesting the terminal device to obtain the target list information, the target data item list required for calculating the first target performance index can be determined according to the model of the image processing unit, so that the terminal device can effectively collect and feedback the first performance data related to the first target performance index. By calculating the first performance index value corresponding to the first target performance index according to the target performance information list, the target index formula list, and the first performance data, it can effectively adapt to the graphics processing units of different architecture types, accurately calculate the expected performance index value, without complex manual operations, with strong versatility, and this performance index value can be used to be displayed on the set monitoring interface, which is beneficial for developers to monitor the performance status of the graphics processing unit.

[0097] In a possible embodiment, it further includes:

[0098] A second target program deployment module, configured to screen out a second target program that matches the second preset architecture from the set of multiple candidate programs when the architecture type is the second preset architecture, and send the second target program to the terminal device so that the terminal device runs the second target program;

[0099] A second performance data acquisition module, configured to receive the second performance data collected by the terminal device corresponding to the set second performance data item based on the second target program;

[0100] The second performance index calculation module is configured to calculate the second performance index value corresponding to the second target performance index according to the index calculation formula corresponding to at least one set second target performance index and the second performance data.

[0101] In a possible embodiment, the target performance information list includes a performance counter list and a performance event information list, the target data item list includes a target counter list, and the first performance data acquisition module 103 is further configured to:

[0102] Filter out the target index formulas corresponding to at least one set first target performance index from the target index formula list;

[0103] Extract the target performance event identifier from the performance event information list based on the variable name in the target index formula;

[0104] Extract the target counter items from the performance counter list according to the target performance event identifier, and combine the target counter items to obtain the target counter list.

[0105] In a possible embodiment, the first performance index calculation module 104 is further configured to:

[0106] Filter out the target index formulas corresponding to at least one set first target performance index from the target index formula list;

[0107] Determine the target performance data items corresponding to the variable names in the target index formula according to the target performance information list;

[0108] Extract the target performance data corresponding to the target performance data items from the first performance data, and calculate the first performance index value corresponding to the first target performance index based on the target performance data and the target index formula.

[0109] In a possible embodiment, it further includes a first optimization prompt module, which is configured to:

[0110] Obtain the application identifier of the application currently running on the terminal device;

[0111] Calculate the average index value based on the first performance index values within a preset time range, and compare the average index value with the reference index range set for the corresponding application identifier;

[0112] Generate a first optimization prompt message when the average index value is not within the reference index range.

[0113] In a possible embodiment, it further includes a second optimization prompt module, which is configured to:

[0114] Compare the average metric value with the historical average metric value recorded for the corresponding application identifier, where the historical average metric value is calculated based on the first performance metric value corresponding to the previous run of the application on the terminal device;

[0115] Generate a second optimization prompt message when the difference between the average metric value and the historical average metric value is not within a preset numerical range.

[0116] Figure 8 It is a schematic structural diagram of a performance data processing device for a graphics processing unit provided by an embodiment of the present application. As Figure 8 shown, the device includes a processor 201, a memory 202, an input device 203, and an output device 204; the number of processors 201 in the device can be one or more, Figure 8 taking one processor 201 as an example; the processor 201, memory 202, input device 203, and output device 204 in the device can be connected through a bus or other means, Figure 8 taking connection through a bus as an example. The memory 202, as a computer-readable storage medium, can be configured to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the performance data processing method of the graphics processing unit in the embodiments of the present application. The processor 201 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory 202, that is, implements the above-mentioned performance data processing method of the graphics processing unit. The input device 203 can be configured to receive input digital or character information, and generate key signal inputs related to the user settings and function controls of the device. The output device 204 can include display devices such as a display screen.

[0117] The embodiments of the present application further provide a non-volatile storage medium containing computer-executable instructions. When the computer-executable instructions are executed by a computer processor, they are configured to execute a method for processing performance data of a graphics processing unit described in the above embodiments. Specifically, the method includes: obtaining the architecture type of the graphics processing unit in the terminal device; in the case where the architecture type is the first preset architecture, screening out a first target program that matches the first preset architecture from a plurality of set candidate programs, and sending the first target program to the terminal device so that the terminal device runs the first target program; requesting the terminal device to obtain target list information, so that the terminal device filters out target list information that matches the model of the graphics processing unit from a plurality of set candidate list information based on the first target program and feeds back the target list information. The target list information includes a target performance information list and a target index formula list; determining a target data item list according to the target performance information list, the target index formula list, and at least one set first target performance indicator, and sending the target data item list to the terminal device so that the terminal device collects first performance data for the first performance data item in the target data item list based on the first target program and feeds back the first performance data; calculating a first performance indicator value corresponding to the first target performance indicator according to the target performance information list, the target index formula list, and the first performance data.

[0118] It should be noted that in the embodiments of the above-described performance data processing device of the graphics processing unit, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of the functional units are only for the convenience of mutual distinction and are not configured to limit the protection scope of the embodiments of the present application.

[0119] In some possible implementation manners, various aspects of the method provided in the present application can also be implemented in the form of a program product, which includes program code. When the program product runs on a computer device, the program code is configured to cause the computer device to execute the steps in the methods according to various exemplary implementation manners of the present application described above in this specification. For example, the computer device can execute the method for processing performance data of the graphics processing unit recorded in the embodiments of the present application. The program product can be implemented by any combination of one or more readable media.

Claims

1. A method for processing performance data of a graphics processing unit, characterized in that: include: Acquiring an architecture type of a graphics processing unit in a terminal device, and when the architecture type is a first preset architecture, selecting a first target program matching the first preset architecture from a plurality of set candidate programs, and sending the first target program to the terminal device so that the terminal device runs the first target program; Requesting the terminal device to obtain target list information, so that the terminal device screens out target list information matching the model of the graphics processing unit from a plurality of set candidate list information based on the first target program, and feeds back the target list information, wherein the target list information includes a target performance information list and a target indicator formula list; Determine a target data item list according to the target performance information list, the target indicator formula list and at least one set first target performance indicator, and send the target data item list to the terminal device, so that the terminal device collects first performance data for the first performance data item in the target data item list based on the first target program, and feeds back the first performance data; A first performance indicator value corresponding to the first target performance indicator is calculated according to the target performance information list, the target indicator formula list and the first performance data.

2. The method for processing performance data of a graphics processing unit according to claim 1, characterized in that: Also includes: In the case where the architecture type is a second preset architecture, selecting a second target program matching the second preset architecture from a plurality of set candidate programs, and sending the second target program to the terminal device so that the terminal device runs the second target program; Receiving second performance data collected by the terminal device based on the second target program and corresponding to the set second performance data item; A second performance indicator value corresponding to the second target performance indicator is calculated based on an indicator calculation formula corresponding to at least one set second target performance indicator and the second performance data.

3. The method for processing performance data of a graphics processing unit according to claim 1, characterized in that: The target performance information list includes a performance counter list and a performance event information list, the target data item list includes a target counter list, and the determining of the target data item list according to the target performance information list, the target indicator formula list and the at least one first target performance indicator set includes: Filtering out a target indicator formula corresponding to at least one first target performance indicator set from the target indicator formula list; Extracting a target performance event identifier from the performance event information list based on the variable name in the target indicator formula; A target counter item is extracted from the performance counter list according to the target performance event identifier, and the target counter items are combined to obtain a target counter list.

4. The method for processing performance data of a graphics processing unit according to claim 1, characterized in that: The calculating, according to the target performance information list, the target indicator formula list and the first performance data, a first performance indicator value corresponding to the first target performance indicator comprises: Filtering out a target indicator formula corresponding to at least one first target performance indicator set from the target indicator formula list; Determine the target performance data item corresponding to the variable name in the target indicator formula according to the target performance information list; Target performance data corresponding to the target performance data item is extracted from the first performance data, and a first performance indicator value corresponding to the first target performance indicator is calculated based on the target performance data and the target indicator formula.

5. The method for processing performance data of a graphics processing unit according to claim 1, characterized in that: After the first performance indicator value corresponding to the first target performance indicator is obtained by calculation, the method further includes: Obtaining an application identifier of an application currently running on the terminal device; Calculating an average indicator value based on the first performance indicator value within a preset time range, and comparing the average indicator value with a reference indicator range set corresponding to the application identifier; When the average index value is not within the reference index range, first optimization prompt information is generated.

6. The method for processing performance data of a graphics processing unit according to claim 5, characterized in that: After the average indicator value is calculated based on the first performance indicator value within the preset time range, the method further includes: Comparing the average index value with a historical average index value recorded corresponding to the application identification, the historical average index value being calculated based on a first performance index value corresponding to the last time the terminal device ran the application program; When the difference between the average index value and the historical average index value is not within a preset value range, second optimization prompt information is generated.

7. A performance data processing device for a graphics processing unit, characterized in that: include: a first target program deployment module, configured to obtain an architecture type of a graphics processing unit in a terminal device, and when the architecture type is a first preset architecture, screen out a first target program matching the first preset architecture from a plurality of set candidate programs, and send the first target program to the terminal device so that the terminal device runs the first target program; a list information acquisition module, configured to request the terminal device to acquire target list information, so that the terminal device screens out target list information matching the model of the graphics processing unit from a plurality of set candidate list information based on the first target program, and feeds back the target list information, wherein the target list information includes a target performance information list and a target indicator formula list; A first performance data acquisition module is configured to determine a target data item list according to the target performance information list, the target indicator formula list and at least one set first target performance indicator, and send the target data item list to the terminal device, so that the terminal device collects first performance data for the first performance data item in the target data item list based on the first target program, and feeds back the first performance data; The first performance indicator calculation module is configured to calculate a first performance indicator value corresponding to the first target performance indicator according to the target performance information list, the target indicator formula list and the first performance data.

8. A performance data processing device for a graphics processing unit, the device comprising: one or more processors; A storage device configured to store one or more programs, when the one or more programs are executed by the one or more processors, enables the one or more processors to implement the performance data processing method of the graphics processing unit described in any one of claims 1-6.

9. A non-volatile storage medium storing computer executable instructions, wherein the computer executable instructions are configured to execute the performance data processing method of a graphics processing unit according to any one of claims 1 to 6 when executed by a computer processor.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the performance data processing method of the graphics processing unit according to any one of claims 1 to 6 is implemented.

Images