Collaborative functional testing method, device, and testing platform for heterogeneous system-on-chip

Abstract

The present application provides a method, device, and test platform for collaborative function testing of heterogeneous systems on chip, and relates to the field of chip testing technology. The model of the CPU, the actual business receiving parameters when the CPU receives business data from an external business device, and the type of business data received are obtained; the actual business writing parameters when the CPU writes business data to the memory after receiving the business data are obtained; the actual business processing parameters when the GPU processes the business data after the CPU notifies the GPU to process the business data are obtained. According to the model of the CPU and the type of business data, a set of standard business comparison parameters can be determined; finally, the actual business receiving parameters, the actual business writing parameters, and the actual business processing parameters are respectively compared with the set of standard business comparison parameters to obtain the first test result of the collaborative function test of the heterogeneous system on chip, which can comprehensively and efficiently test whether there are any abnormalities in the operation of the heterogeneous system on chip in the entire workflow.

Description

Technical Field

[0001] The present application relates to the field of chip testing technology, and in particular to a collaborative functional testing method, device, and testing platform for heterogeneous on-chip systems. Background Art

[0002] With the rapid development of information technology, heterogeneous computing has shown great advantages in many fields, and heterogeneous system-on-chip has come into being. Heterogeneous system-on-chip integrates multiple computing units with different instruction sets and architectures on the same chip to achieve complex and diverse computing tasks. It can meet the stringent requirements of different application scenarios such as artificial intelligence, big data processing, and the Internet of Things for high performance, low power consumption, and high flexibility.

[0003] However, the structure of a heterogeneous system-on-chip that integrates multiple different computing units is very complex. When a heterogeneous system-on-chip is in operation, multiple computing units have unique working principles, instruction sets, hardware architectures, and complex communications and collaborative tasks between each computing unit. Traditional chip testing methods are difficult to comprehensively and efficiently test whether there are any abnormalities in the operation of a heterogeneous system-on-chip throughout the entire workflow. Summary of the invention

[0004] The present application provides a method, device, and testing platform for collaborative function testing of a heterogeneous system on chip, which are used to solve the problem in the prior art that it is difficult to comprehensively and efficiently test whether an abnormality occurs in the operation of a heterogeneous system on chip in the entire workflow.

[0005] In a first aspect, the present application provides a method for testing the collaborative function of a heterogeneous system on chip, which is applied to a test platform. The heterogeneous system on chip includes a CPU, a GPU, a memory, and a communication bus. The CPU, the GPU, and the memory are electrically connected to the communication bus respectively, and the test platform is electrically connected to the communication bus. The method provided by the present application includes:

[0006] Obtain the model of the CPU, the actual service receiving parameters when the CPU receives service data from an external service device, and the type of service data received;

[0007] Obtaining actual service write parameters for writing service data into the memory after the CPU has received the service data;

[0008] After the CPU notifies the GPU to process the business data, the actual business processing parameters used by the GPU to process the business data are obtained;

[0009] Determine a set of standard business comparison parameters based on the CPU model and the type of business data;

[0010] The actual service receiving parameters, the actual service writing parameters, and the actual service processing parameters are respectively compared with the standard service comparison parameter set to obtain a first test result of the collaborative function test of the heterogeneous system on chip.

[0011] In some implementations, the standard service comparison parameter set includes standard service reception parameters, standard service writing parameters, and standard service processing parameters, and the actual service reception parameters, the actual service writing parameters, and the service processing parameters are respectively compared with the standard service comparison parameter set to obtain the test result of the collaborative function test of the heterogeneous system on chip, including:

[0012] When the actual service receiving parameters are inconsistent with the standard service receiving parameters, it is determined that the CPU receiving data is abnormal;

[0013] When the actual business write parameters are inconsistent with the standard business write parameters, it is determined that there is an abnormality in the CPU data write;

[0014] When the actual business processing parameters are inconsistent with the standard business processing parameters, it is determined that there is an abnormality in GPU data processing;

[0015] The first test result of the cooperative function test of the heterogeneous system on chip includes at least one of an abnormality in CPU receiving data, an abnormality in CPU writing data, an abnormality in GPU reading data, and an abnormality in GPU data processing.

[0016] In some implementations, after obtaining the actual service processing parameters for processing the service data after the GPU reads the service data, the method provided by the present application further includes:

[0017] The actual business receiving parameters, the actual business writing parameters and the business processing parameters are spliced ​​and characterized to obtain the fusion parameter characteristics in the business data processing process;

[0018] The fusion parameter features are input into a pre-trained first collaborative function test model to obtain a second test result of the collaborative function test of the heterogeneous system on chip, wherein the first collaborative function test model is trained by inputting multiple first training samples into a first neural network, each first training sample includes a historical fusion parameter feature and a corresponding test label, and the test label is a normal label or an abnormal label.

[0019] In some implementations, after obtaining the actual service processing parameters for processing the service data after the GPU reads the service data, the method provided by the present application further includes:

[0020] The actual business receiving parameters, the actual business writing parameters, the actual data reading parameters and the business processing parameters are spliced ​​and characterized to obtain the fusion parameter characteristics in the business data processing process;

[0021] Extracting the correlation between the actual service receiving parameters, the actual service writing parameters and the service processing parameters, and characterizing the correlation to obtain correlation features;

[0022] Inject the associated features into the fusion parameter features to obtain the collaborative functional features of the heterogeneous on-chip system for business data processing;

[0023] The collaborative function feature is input into a pre-trained second collaborative function test model to obtain a third test result of the collaborative function test of the heterogeneous system on chip, wherein the second collaborative function test model is trained by inputting multiple second training samples into a second neural network, and each second training sample includes a historical collaborative function feature and a corresponding test label, and the test label is a normal label or an abnormal label.

[0024] In some implementations, after obtaining a third test result of the collaborative functional test of the heterogeneous system on chip, the method further includes:

[0025] When the third test result indicates that the collaborative function of the heterogeneous system-on-chip is abnormal, the collaborative function feature is input into a pre-trained abnormality cause analysis model to obtain the cause of the abnormality of the collaborative function of the heterogeneous system-on-chip, wherein the abnormality cause analysis model is obtained by inputting a plurality of third training samples into a third neural network for training, and each third training sample includes a historical collaborative function feature and a corresponding actual cause of the abnormality of the collaborative function of the heterogeneous system-on-chip in history;

[0026] Find debugging solutions for the heterogeneous system-on-chip based on the reasons why the collaborative functions of the heterogeneous system-on-chip are abnormal.

[0027] In some implementations, the business data is video data, and the GPU processes the business data including: the GPU reads the video data from the memory, performs rendering calculations on the video data, and writes the rendered video data into the memory.

[0028] In some implementations, the actual service reception parameter includes: at least one of an actual reception rate of video data and an actual number of frames of received video data;

[0029] The actual service write parameter includes: at least one of an actual data write rate at which the GPU writes video data into the memory, an actual number of frames of the written video data, an actual cache hit rate, and an actual memory bandwidth occupancy rate;

[0030] The business processing parameters include: at least one of the actual data reading rate at which the CPU reads video data from the memory, the actual number of frames of video data actually read, the actual rendering calculation time of the video data by the GPU, the actual data writing rate at which the GPU writes the rendered video data to the memory, the actual quality of the rendered video data, and the actual video memory occupancy rate corresponding to the GPU processing of the video data.

[0031] In a second aspect, the present application further provides a collaborative function test device for a heterogeneous system on chip, which is configured on a test platform. The heterogeneous system on chip includes a CPU, a GPU, a memory, and a communication bus. The CPU, the GPU, and the memory are electrically connected to the communication bus respectively, and the test platform is electrically connected to the communication bus. The device provided by the present application includes:

[0032] A data acquisition unit, used to acquire the model of the CPU, actual service receiving parameters when the CPU receives service data from an external service device, and the type of service data received;

[0033] The data acquisition unit is also used to obtain the actual business writing parameters for the CPU to write the business data into the memory after receiving the business data;

[0034] The data acquisition unit is further used to acquire actual business processing parameters of the GPU for processing the business data after the CPU notifies the GPU to process the business data;

[0035] A parameter determination unit, used to determine a set of standard service comparison parameters according to the CPU model and the type of service data;

[0036] The functional testing unit is used to compare the actual service receiving parameters, the actual service writing parameters, and the actual service processing parameters with the standard service comparison parameter set respectively to obtain the first test result of the collaborative functional test of the heterogeneous on-chip system.

[0037] In a third aspect, the present application also provides a testing platform, comprising: a processor; a memory for storing processor executable instructions; wherein the processor is configured to execute instructions to implement the collaborative functional testing method of heterogeneous on-chip systems as provided in the first aspect of the present application.

[0038] In a fourth aspect, the present application further provides a storage medium, which, when the instructions in the storage medium are executed by the processor of the test platform, enables the test platform to execute the collaborative functional testing method of the heterogeneous on-chip system provided in the first aspect of the present application.

[0039] The present application provides a method, device, and test platform for testing the collaborative function of a heterogeneous system on chip, which can obtain the model of the CPU, the actual business receiving parameters when the CPU receives business data from an external business device, and the type of business data received; obtain the actual business write parameters when the CPU writes the business data to the memory after receiving the business data; obtain the actual business processing parameters when the GPU processes the business data after the CPU notifies the GPU to process the business data. In this way, the heterogeneous system on chip can obtain the actual business receiving parameters for writing the business data received by the CPU, the actual business writing parameters for writing the data, and the actual business processing parameters for processing the business data by the GPU, and the working parameters of the heterogeneous system on chip in the workflow are fully and automatically obtained. According to the model of the CPU and the type of business data, it can be determined to obtain a set of standard business comparison parameters; finally, the actual business receiving parameters, the actual business writing parameters, and the actual business processing parameters are respectively compared with the standard business comparison parameter set to obtain the first test result of the collaborative function test of the heterogeneous system on chip. In this way, it is possible to comprehensively and efficiently test whether the operation of the heterogeneous system on chip in the entire workflow is abnormal. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0041] Figure 1 A circuit module block diagram of a heterogeneous system-on-chip provided in an embodiment of the present application;

[0042] Figure 2 One of the flow charts of the collaborative function testing method of heterogeneous system-on-chip provided in the embodiment of the present application;

[0043] Figure 3 Flow chart 2 of the collaborative function testing method of heterogeneous system-on-chip provided in an embodiment of the present application;

[0044] Figure 4 Flow chart 3 of the collaborative function testing method of heterogeneous system-on-chip provided in an embodiment of the present application;

[0045] Figure 5 A software unit block diagram of a collaborative functional testing device for a heterogeneous system-on-chip provided in an embodiment of the present application. DETAILED DESCRIPTION

[0046] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0047] Various structural schematic diagrams according to embodiments of the present disclosure are shown in the accompanying drawings. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may further design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0048] In the context of the present disclosure, when a layer / element is referred to as being "on" another layer / element, the layer / element may be directly on the other layer / element or an intervening layer / element may exist therebetween. In addition, if a layer / element is "on" another layer / element in one orientation, the layer / element may be "below" the other layer / element when the orientation is reversed.

[0049] The following is a detailed description of the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0050] The embodiment of the present application provides a method for testing the collaborative function of a heterogeneous system on chip 100, which is applied to a test platform 103, such as Figure 1 As described, the heterogeneous system on chip 100 includes a CPU 101, a GPU 102, a memory and a communication bus, wherein the CPU 101, the GPU 102 and the memory are electrically connected to the communication bus respectively, and the test platform 103 is electrically connected to the communication bus. For example, the test platform 103 may be an FPGA module. Figure 2 As shown, the method provided in the embodiment of the present application includes:

[0051] S201: Acquire the model of the CPU 101, actual service receiving parameters when the CPU 101 receives service data from an external service device, and the type of the received service data.

[0052] It is understandable that the model of CPU 101 is associated with the standard operating parameters of CPU 101 (such as standard service writing parameters, standard data receiving parameters, etc.). The type of service data (such as image type, video type, and text type, etc.) is also associated with the standard operating parameters of CPU 101 and GPU 102 (such as standard service writing parameters, standard data receiving parameters, etc.).

[0053] Exemplarily, the business data may be but is not limited to video data to be processed. Correspondingly, the external business device may be but is not limited to devices such as a mobile phone or a cloud server. The actual business reception parameters may be but are not limited to the actual data reception rate and / or the actual number of frames of the received video data.

[0054] S202: Acquire actual service write parameters for writing service data into the memory after the CPU 101 completes receiving the service data.

[0055] Exemplarily, when the business data is video data, the actual business write parameters include at least one of the actual data write rate at which CPU 101 writes video data to the memory, the actual number of frames of the written video data, the actual cache hit rate, and the actual memory bandwidth occupancy rate.

[0056] S203: Obtaining actual service processing parameters used by GPU 102 to process service data after CPU 101 notifies GPU 102 to process service data.

[0057] When the service data is video data, the processing of the service data by GPU 102 includes: GPU 102 reading the video data from the memory, performing rendering calculation on the video data, and writing the rendered video data into the memory.

[0058] Exemplarily, when the business data is video data, the actual business processing parameters include: the actual data reading rate at which GPU 102 reads video data from the memory, the actual number of frames of the read video data, the actual rendering calculation time of the video data by GPU 102, the actual data writing rate at which GPU 102 writes the rendered video data to the memory, the actual quality of the rendered video data, and at least one of the actual video memory occupancy rate corresponding to the video data processing by GPU 102.

[0059] S204: Determine a set of standard business comparison parameters according to the model of CPU 101 and the type of business data.

[0060] Exemplarily, the test platform 103 pre-stores a mapping table, which includes a one-to-one correspondence between the model of CPU 101 and the type of business data and a set of standard business comparison parameters. Further, the standard business comparison parameter set can be found from the mapping table according to the model of CPU 101 and the type of business data.

[0061] S205: Compare the actual service receiving parameters, the actual service writing parameters, and the actual service processing parameters with the standard service comparison parameter set respectively to obtain a first test result of the collaborative function test of the heterogeneous system on chip 100.

[0062] In some implementations, S205 may be specifically implemented as follows:

[0063] Step 1: When the actual service reception parameters are inconsistent with the standard service reception parameters, it is determined that there is an abnormality in the data received by the CPU 101 .

[0064] For example, when the actual service reception parameters include the data reception rate, the actual data reception rate can be compared with the standard data reception rate to see if they are consistent. If they are inconsistent, it is determined that there is an abnormality in the data received by CPU 101, and there may be a problem with the interface driver of the heterogeneous system on chip 100. For another example, when the actual service reception parameters include the number of frames of received video data, the number of frames of received video data can be compared with the number of frames of standard video data (the number of frames of video data is known in advance) to see if they are consistent. If they are inconsistent, it is determined that there is an abnormality in the data received by CPU 101, and there may be a problem with the interface driver of the heterogeneous system on chip 100.

[0065] Step 2: When the actual service write parameters are inconsistent with the standard service write parameters, it is determined that there is an abnormality in the data writing of CPU 101.

[0066] For example, the actual service write parameter may include the actual data write rate at which the CPU 101 writes video data into the memory, the number of frames of video data actually written, the actual cache hit rate, or the actual memory bandwidth occupancy rate.

[0067] When the actual data write rate at which CPU101 writes video data to the memory is inconsistent with the standard data write rate, it is determined that there is an abnormality in the data writing of CPU101; when the number of frames of video data written by CPU101 to the memory is inconsistent with the number of frames of video data received by CPU101, it is determined that there is an abnormality in the data writing of CPU101; when the actual cache hit rate at which CPU101 writes video data to the memory is inconsistent with the standard cache hit rate, it is determined that there is an abnormality in the data writing of CPU101; when the actual memory bandwidth occupancy rate at which CPU101 writes video data to the memory is inconsistent with the standard memory bandwidth occupancy rate, it is determined that there is an abnormality in the data writing of CPU101.

[0068] Step 3: When the actual business processing parameters are inconsistent with the standard business processing parameters, it is determined that there is an abnormality in the data processing of GPU 102.

[0069] For example, actual business processing parameters may include the actual data reading rate at which GPU 102 reads video data from the memory, the actual number of frames of video data read, the actual rendering calculation time of the video data by GPU 102, the actual data writing rate at which GPU 102 writes the rendered video data to the memory, the actual quality of the rendered video data, and the actual video memory occupancy rate corresponding to the video data processing by GPU 102.

[0070] When the data reading rate at which GPU 102 reads video data from the memory is inconsistent with the standard data reading rate, it is determined that there is an abnormality in the data reading of GPU 102; when the number of frames of video data actually read by GPU 102 is inconsistent with the number of frames of video data written by GPU 102, it is determined that there is an abnormality in the data reading of GPU 102; when the actual rendering calculation time of video data by GPU 102 is inconsistent with the standard rendering calculation time, it is determined that there is an abnormality in the rendering calculation of GPU 102; when the actual quality of video data rendered by GPU 102 (such as actual color accuracy and actual light and shadow effect fidelity, etc.) is inconsistent with the standard image quality (such as standard color accuracy and standard light and shadow effect fidelity, etc.), it is determined that there is an abnormality in the rendering calculation of GPU 102 (the reason may be due to defects in the image rendering algorithm model, improper parameter settings or hardware configuration mismatch); when the actual data writing rate at which GPU 102 writes rendered video data to the memory is inconsistent with the standard data writing rate, it is determined that there is an abnormality in the data writing of GPU 102.

[0071] In addition, GPU102 can also transmit the rendered video data to the display screen for display, can capture the video data displayed by the display device, and determine whether the video data displayed by the display device (such as the smoothness and clarity of the displayed video data) is abnormal. If abnormal, it may be caused by interface compatibility issues or data format conversion errors.

[0072] In addition, the heterogeneous system on chip 100 also includes DSP104, and CPU101 can also transmit the audio data in the video data to DSP104 for processing, and obtain the actual audio processing parameters of DSP104. When the actual audio processing parameters of DSP104 are inconsistent with the standard audio processing parameters, it is determined that there is an abnormality in the business processing of DSP104.

[0073] In summary, the embodiment of the present application provides a method for testing the collaborative function of a heterogeneous system-on-chip 100, which can obtain the model of CPU101, the actual business receiving parameters when CPU101 receives business data from an external business device, and the type of business data received; obtain the actual business write parameters when CPU101 writes business data to the memory after receiving the business data; obtain the actual business processing parameters when GPU102 processes the business data after CPU101 notifies GPU102 to process the business data. In this way, the heterogeneous system-on-chip 100 can obtain the actual business receiving parameters for writing business data received by CPU101, the actual business writing parameters for writing data by CPU101, and the actual business processing parameters for processing business data by GPU102, and the working parameters of the heterogeneous system-on-chip 100 in the workflow are completely and automatically obtained. According to the model of CPU101 and the type of business data, a set of standard business comparison parameters can be determined; finally, the actual business receiving parameters, the actual business writing parameters, and the actual business processing parameters are respectively compared with the set of standard business comparison parameters to obtain the first test result of the collaborative function test of the heterogeneous system-on-chip 100. In this way, it is possible to comprehensively and efficiently test whether the operation of the heterogeneous system-on-chip 100 in the entire workflow occurs abnormally.

[0074] In some embodiments, after S203, as Figure 3 As shown, the method provided in the embodiment of the present application may also include:

[0075] S301: actual business receiving parameters, actual business writing parameters and business processing parameters are concatenated and characterized to obtain fusion parameter features in the process of business data processing.

[0076] S302: Input the fusion parameter feature into the pre-trained first collaborative function test model to obtain a second test result of the collaborative function test of the heterogeneous system on chip 100.

[0077] The first collaborative function test model is obtained by inputting a plurality of first training samples into the first neural network for training, each first training sample includes a historical fusion parameter feature and a corresponding test label, and the test label is a normal label or an abnormal label. It can be understood that the reliability of the second test result obtained by S301-S302 is high.

[0078] In some embodiments, after S203, as Figure 4 As shown, the method provided in the embodiment of the present application may also include:

[0079] S401: actual business receiving parameters, actual business writing parameters, actual data reading parameters and business processing parameters are concatenated and characterized to obtain fusion parameter features in the business data processing process.

[0080] S402: extracting the association relationship between the actual service receiving parameters, the actual service writing parameters and the service processing parameters, and characterizing the association relationship to obtain association features.

[0081] It can be understood that there is a certain correlation between the actual business reception parameters, the actual business writing parameters, the actual data reading parameters and the business processing parameters. Whether there is an abnormality in the overall process of the collaborative function test of the heterogeneous on-chip system 100 is not only related to the actual business reception parameters, the actual business writing parameters, the actual data reading parameters and the business processing parameters themselves, but also to the correlation between the actual business writing parameters, the actual data reading parameters and the business processing parameters.

[0082] S403: Inject the associated features into the fusion parameter features to obtain the collaborative functional features of the heterogeneous on-chip system 100 for processing business data.

[0083] S404: Input the collaborative function feature into the pre-trained second collaborative function test model to obtain a third test result of the collaborative function test of the heterogeneous system on chip 100.

[0084] Among them, the second collaborative function test model is obtained by inputting multiple second training samples into the second neural network for training, and each second training sample includes historical collaborative function characteristics and corresponding test labels, and the test labels are normal labels or abnormal labels. It can be understood that based on the above S401-S404, not only the actual business receiving parameters, actual business writing parameters, actual data reading parameters and business processing parameters are considered, but also the correlation between the actual business writing parameters, actual data reading parameters and business processing parameters is considered, and the reliability of the third test result obtained is higher.

[0085] In some implementations, after S404, the method provided in the embodiment of the present application further includes:

[0086] S405: When the third test result indicates that the collaborative function of the heterogeneous system on chip 100 is abnormal, the collaborative function feature is input into the pre-trained abnormality cause analysis model to obtain the cause of the abnormality of the collaborative function of the heterogeneous system on chip 100.

[0087] Among them, the abnormal cause analysis model is obtained by inputting multiple third training samples into the third neural network for training, and each third training sample includes historical collaborative function characteristics and the corresponding actual reasons for the abnormalities in the collaborative functions of the heterogeneous on-chip system 100 in history.

[0088] S406: Find a debugging solution for the heterogeneous system on chip 100 according to the reason why the coordination function of the heterogeneous system on chip 100 is abnormal.

[0089] It can be understood that, since the obtained third test result has high reliability, the reliability of finding a debugging solution for the heterogeneous system-on-chip 100 is high, and the tester can be guided to debug the heterogeneous system-on-chip 100 .

[0090] In addition, the embodiment of the present application also provides a collaborative function testing device for a heterogeneous system on chip 100, which is configured on a test platform 103. It should be noted that the basic principle and technical effects of the collaborative function testing device for a heterogeneous system on chip 100 provided in the embodiment of the present application are the same as those in the above-mentioned embodiment. For the sake of brief description, for matters not mentioned in the embodiment of the present application, reference may be made to the corresponding contents in the above-mentioned embodiment. The heterogeneous system on chip 100 includes a CPU 101, a GPU 102, a memory, and a communication bus. The CPU 101, the GPU 102, and the memory are electrically connected to the communication bus, respectively, and the test platform 103 is electrically connected to the communication bus. Figure 5 As shown, the device provided in the embodiment of the present application includes a data acquisition unit, a parameter determination unit, and a function testing unit, wherein:

[0091] A data acquisition unit, used to acquire the model of the CPU 101, actual service receiving parameters when the CPU 101 receives service data from an external service device, and the type of the received service data;

[0092] The data acquisition unit is also used to acquire actual service write parameters for writing service data into the memory after the CPU 101 completes receiving the service data;

[0093] The data acquisition unit is further used to acquire actual business processing parameters of the business data processed by the GPU 102 after the CPU 101 notifies the GPU 102 to process the business data;

[0094] A parameter determination unit, used to determine a set of standard service comparison parameters according to the model of CPU 101 and the type of service data;

[0095] The functional testing unit is used to compare the actual service receiving parameters, the actual service writing parameters, and the actual service processing parameters with the standard service comparison parameter set respectively to obtain the first test result of the collaborative functional test of the heterogeneous on-chip system 100.

[0096] In some embodiments, the first test result of the collaborative function test of the heterogeneous system on chip 100 includes at least one of an abnormality in data reception by CPU 101, an abnormality in data writing by CPU 101, an abnormality in data reading by GPU 102, and an abnormality in data processing by GPU 102. The standard service comparison parameter set includes standard service reception parameters, standard service writing parameters, and standard service processing parameters. The functional test unit is specifically used to determine that there is an abnormality in data reception by CPU 101 when the actual service reception parameters are inconsistent with the standard service reception parameters; determine that there is an abnormality in data writing by CPU 101 when the actual service writing parameters are inconsistent with the standard service writing parameters; and determine that there is an abnormality in data processing by GPU 102 when the actual service processing parameters are inconsistent with the standard service processing parameters.

[0097] In some embodiments, the functional testing unit is specifically used to splice and characterize the actual business receiving parameters, the actual business writing parameters, and the business processing parameters to obtain the fusion parameter characteristics in the processing of the business data; the fusion parameter characteristics are input into the pre-trained first collaborative function test model to obtain the second test result of the collaborative function test of the heterogeneous on-chip system 100, wherein the first collaborative function test model is obtained by inputting multiple first training samples into the first neural network for training, and each first training sample includes a historical fusion parameter characteristic and a corresponding test label, and the test label is a normal label or an abnormal label.

[0098] In some embodiments, the functional testing unit is specifically used to splice and characterize the actual business receiving parameters, the actual business writing parameters, the actual data reading parameters, and the business processing parameters to obtain the fusion parameter characteristics in the business data processing process; extract the correlation relationship between the actual business receiving parameters, the actual business writing parameters, and the business processing parameters, and characterize the correlation relationship to obtain the correlation characteristics; inject the correlation characteristics into the fusion parameter characteristics to obtain the collaborative function characteristics of the heterogeneous on-chip system 100 for business data processing; input the collaborative function characteristics into the pre-trained second collaborative function test model to obtain the third test result of the collaborative function test of the heterogeneous on-chip system 100, wherein the second collaborative function test model is obtained by inputting multiple second training samples into the second neural network for training, and each second training sample includes a historical collaborative function characteristic and a corresponding test label, and the test label is a normal label or an abnormal label.

[0099] In some embodiments, the method provided in the embodiments of the present application further includes:

[0100] The abnormality cause analysis unit is used to input the collaborative function characteristics into the pre-trained abnormality cause analysis model to obtain the cause of the abnormality of the collaborative function of the heterogeneous system on chip 100 when the third test result indicates that the collaborative function of the heterogeneous system on chip 100 is abnormal, wherein the abnormality cause analysis model is trained by inputting multiple third training samples into the third neural network, and each third training sample includes historical collaborative function characteristics and the corresponding actual cause of the abnormality of the collaborative function of the heterogeneous system on chip 100 in history.

[0101] The debugging solution searching unit is used to search for a debugging solution for the heterogeneous system on chip 100 according to the reason why the cooperative function of the heterogeneous system on chip 100 is abnormal.

[0102] In some implementations, the business data is video data, and the processing of the business data by GPU 102 includes: GPU 102 reading the video data from the memory, performing rendering calculations on the video data, and writing the rendered video data into the memory.

[0103] In some implementations, the actual service reception parameter includes: at least one of an actual reception rate of video data and an actual number of frames of received video data;

[0104] The actual service write parameter includes: at least one of an actual data write rate at which the GPU 102 writes video data into the memory, an actual number of frames of the written video data, an actual cache hit rate, and an actual memory bandwidth occupancy rate;

[0105] The business processing parameters include: the actual data reading rate at which CPU101 reads video data from the memory, the actual number of frames of video data read, the actual rendering calculation time of video data by GPU102, the actual data writing rate at which GPU102 writes rendered video data to the memory, the actual quality of the rendered video data, and at least one of the actual video memory occupancy rate corresponding to the video data processing by GPU102.

[0106] In addition, an embodiment of the present application also provides a testing platform, including: a processor; a memory for storing processor executable instructions; wherein the processor is configured to execute instructions to implement a collaborative functional testing method for heterogeneous on-chip systems as provided in the above embodiments of the present application.

[0107] In addition, an embodiment of the present application further provides a storage medium. When the instructions in the storage medium are executed by the processor of the test platform, the test platform can execute the collaborative functional testing method of the heterogeneous on-chip system provided by the above embodiment of the present application.

[0108] In the above description, the technical details such as the patterning of each layer are not described in detail. However, those skilled in the art should understand that various technical means can be used to form layers, regions, etc. of desired shapes. In addition, in order to form the same structure, those skilled in the art can also design methods that are not completely the same as the methods described above. In addition, although the various embodiments are described above separately, this does not mean that the measures in the various embodiments cannot be used in combination to advantage.

[0109] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0110] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for testing the collaborative function of heterogeneous systems on a chip, characterized in that: Applied to a test platform, the heterogeneous system on chip includes a CPU, a GPU, a memory and a communication bus, the CPU, the GPU and the memory are electrically connected to the communication bus respectively, the test platform is electrically connected to the communication bus, and the method includes: Acquire the model of the CPU, actual service receiving parameters when the CPU receives service data from an external service device, and the type of the received service data, wherein the service data is video data, and the actual service receiving parameters include an actual data receiving rate of the video data; Acquire actual business write parameters of the CPU writing the business data into the memory after receiving the business data, wherein the actual business write parameters include at least one of an actual data write rate, an actual cache hit rate, and an actual memory bandwidth occupancy rate of the CPU writing the video data into the memory; After the CPU notifies the GPU to process the business data, actual business processing parameters of the GPU for processing the business data are obtained, where the actual business processing parameters include: an actual data reading rate at which the GPU reads the video data from a memory, an actual rendering calculation duration of the video data by the GPU, an actual data writing rate at which the GPU writes the rendered video data to a memory, an actual quality of the rendered video data, and an actual video memory occupancy rate corresponding to the processing of the video data by the GPU; Determining a set of standard business comparison parameters according to the model of the CPU and the type of the business data; Comparing the actual service receiving parameter, the actual service writing parameter, and the actual service processing parameter with the standard service comparison parameter set, respectively, to obtain a first test result of the collaborative function test of the heterogeneous system on chip; After the GPU reads the service data and obtains the actual service processing parameters for processing the service data, the method further includes: The actual service receiving parameter, the actual service writing parameter, the actual data reading parameter and the service processing parameter are spliced ​​and characterized to obtain the fusion parameter characteristics in the process of processing the service data; Extracting the association relationship between the actual service receiving parameter, the actual service writing parameter and the service processing parameter, and characterizing the association relationship to obtain an association feature; Injecting the associated feature into the fusion parameter feature to obtain a collaborative functional feature of the heterogeneous system-on-chip processing of the business data; The collaborative function feature is input into a pre-trained second collaborative function test model to obtain a third test result of the collaborative function test of the heterogeneous system on chip, wherein the second collaborative function test model is obtained by inputting multiple second training samples into a second neural network for training, each of the second training samples includes a historical collaborative function feature and a corresponding test label, and the test label is a normal label or an abnormal label.

2. The method according to claim 1, characterized in that The standard service comparison parameter set includes a standard service reception parameter, a standard service writing parameter, and a standard service processing parameter, and the actual service reception parameter, the actual service writing parameter, and the service processing parameter are respectively compared with the standard service comparison parameter set to obtain a test result of the collaborative function test of the heterogeneous system on chip, including: When the actual service reception parameter is inconsistent with the standard service reception parameter, determining that there is an abnormality in the CPU receiving data; When the actual service write parameter is inconsistent with the standard service write parameter, determining that an abnormality exists in the CPU data write; When the actual service processing parameter is inconsistent with the standard service processing parameter, determining that an abnormality exists in the GPU data processing; The first test result of the cooperative functional test of the heterogeneous system on chip includes at least one of an abnormality in the CPU receiving data, an abnormality in the CPU writing data, an abnormality in the GPU reading data, and an abnormality in the GPU processing data.

3. The method according to claim 1, characterized in that After the GPU reads the service data and obtains the actual service processing parameters for processing the service data, the method further includes: The actual service receiving parameter, the actual service writing parameter and the service processing parameter are spliced ​​and characterized to obtain the fusion parameter characteristics in the process of processing the service data; The fusion parameter feature is input into a pre-trained first collaborative function test model to obtain a second test result of the collaborative function test of the heterogeneous system on chip, wherein the first collaborative function test model is obtained by inputting multiple first training samples into a first neural network for training, each of the first training samples includes a historical fusion parameter feature and a corresponding test label, and the test label is a normal label or an abnormal label.

4. The method according to claim 1, characterized in that: After obtaining the third test result of the collaborative functional test of the heterogeneous system on chip, the method further includes: In the case where the third test result indicates that the collaborative function of the heterogeneous system-on-chip is abnormal, the collaborative function feature is input into a pre-trained abnormality cause analysis model to obtain the cause of the abnormality of the collaborative function of the heterogeneous system-on-chip, wherein the abnormality cause analysis model is obtained by inputting a plurality of third training samples into a third neural network for training, and each of the third training samples includes a historical collaborative function feature and a corresponding actual cause of the abnormality of the collaborative function of the heterogeneous system-on-chip in history; According to the reason why the cooperative function of the heterogeneous system on chip is abnormal, a debugging solution for the heterogeneous system on chip is found.

5. The method according to any one of claims 1 to 4, characterized in that: The business data is video data, and the GPU processes the business data including: the GPU reads the video data from the memory, performs rendering calculation on the video data, and writes the rendered video data into the memory.

6. A collaborative functional testing device for heterogeneous system-on-chip, characterized in that: The heterogeneous system-on-chip is configured on a test platform, wherein the heterogeneous system-on-chip includes a CPU, a GPU, a memory, and a communication bus, wherein the CPU, the GPU, and the memory are electrically connected to the communication bus respectively, and the test platform is electrically connected to the communication bus. The device includes: a data acquisition unit, used to acquire the model of the CPU, actual service reception parameters when the CPU receives service data from an external service device, and the type of the received service data, wherein the service data is video data, and the actual service reception parameters include an actual data reception rate of the video data; The data acquisition unit is further used to acquire actual business write parameters of the CPU writing the business data into the memory after receiving the business data, wherein the actual business write parameters include at least one of an actual data write rate, an actual cache hit rate, and an actual memory bandwidth occupancy rate of the CPU writing the video data into the memory; The data acquisition unit is further used to acquire actual business processing parameters of the business data processed by the GPU after the CPU notifies the GPU to process the business data, the actual business processing parameters including: an actual data reading rate at which the GPU reads the video data from the memory, an actual rendering calculation time of the video data by the GPU, an actual data writing rate at which the GPU writes the rendered video data to the memory, an actual quality of the rendered video data, and an actual video memory occupancy rate corresponding to the processing of the video data by the GPU; A parameter determination unit, used to determine a set of standard service comparison parameters according to the model of the CPU and the type of the service data; A functional testing unit, configured to compare the actual service receiving parameter, the actual service writing parameter, and the actual service processing parameter with the standard service comparison parameter set, respectively, to obtain a first test result of the collaborative functional test of the heterogeneous system on chip; The functional testing unit is specifically used to splice and characterize the actual business receiving parameters, the actual business writing parameters, the actual data reading parameters and the business processing parameters to obtain fusion parameter features in the process of processing the business data; extract the association relationship between the actual business receiving parameters, the actual business writing parameters and the business processing parameters, and characterize the association relationship to obtain association features; inject the association features into the fusion parameter features to obtain the collaborative function features of the heterogeneous on-chip system for processing the business data; input the collaborative function features into the pre-trained second collaborative function test model to obtain the third test result of the collaborative function test of the heterogeneous on-chip system, wherein the second collaborative function test model is obtained by inputting multiple second training samples into the second neural network for training, and each of the second training samples includes a historical collaborative function feature and a corresponding test label, and the test label is a normal label or an abnormal label.

7. A test platform, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the collaborative functional testing method of a heterogeneous system on chip as described in any one of claims 1 to 5.

8. A storage medium, characterized in that: When the instructions in the storage medium are executed by a processor of the test platform, the test platform is enabled to execute the collaborative functional testing method for heterogeneous systems on a chip as described in any one of claims 1 to 5.

Images