Test vector generation method and device
Through the neural network model, the test vector length selection of the test vectors is optimized, and the test vectors that meet preset conditions are generated, which solves the problem of determining the length of the test vectors in integrated circuit tests, improves the test efficiency and accuracy, and reduces engineer dependence.
Patent Information
- Application Number
- CN202411113452.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-14
- Publication Date
- 2024-11-08
AI Technical Summary
In integrated circuit testing, it is difficult for the prior art to quickly and efficiently determine the length of the test vector corresponding to different test modes, resulting in difficult to balance the fault coverage rate and test costs. Relying on engineering experience leads to a large number of iterations, low efficiency and uncertain results.
A pre-trained neural network model is used to generate a fault coverage network set based on different test vector lengths under different test modes, select a combination of test vectors that meet the preset constraints, and generate a test vector with the optimal length combination result through the ATPG tool.
It realizes more accurate test vector generation, reduces test costs, improves test efficiency, automates the test vector selection process, reduces engineers' energy and time dependence, and achieves the best balance between test cost and fault coverage.
Smart Images

Figure CN118914816A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of integrated circuit testing, specifically to the field of automatic test vector generation, and more specifically to a test vector generation method and device. Background Art
[0002] In integrated circuit (IC) testing, automatic test pattern generation (ATPG) is an important step to ensure the normal function of the chip. For complex system-on-chip (SoC), how to achieve sufficiently high fault coverage with the lowest possible test cost is a major challenge in testing. In the prior art, the compression mode (COMP) is the first choice for generating test vectors due to its advantage of low test cost. However, with the increasing complexity of SoC chips, a single compression mode often cannot achieve a high fault coverage; while the full scan mode (FULLSCAN) can achieve high coverage, but the test cost is high. Therefore, engineers usually use the incremental test method to generate test vectors. First, a certain length of COMP mode test vector is used to cover most of the faults, and then a certain length of FULLSCAN mode test vector is used on this basis to cover those faults that are difficult to cover with the compression mode, thereby further improving the fault coverage at a limited test cost.
[0003] At present, in incremental testing, the test vector lengths corresponding to different test modes can only be selected based on the experience of engineers, such as the test vector lengths of COMP mode and FULLSCAN mode, and there is no unified method for determining the test vector lengths of COMP mode and FULLSCAN mode. When using conventional experience to select the test vector lengths of different test modes, in order to balance the test cost and fault coverage as much as possible, it is necessary to repeatedly try the fault coverage effects under different test vector lengths, so that it often takes multiple iterations to confirm the final test vector length, which is inefficient and the combination selection result of the test vector is not necessarily the optimal solution under the current test cost. And when more test modes with different advantages and disadvantages appear, the conventional experience selection method will require more iteration cycles and the energy and time of engineers. Over-reliance on the engineering experience of engineers cannot guarantee the test efficiency and effect of integrated circuits. Summary of the invention
[0004] In order to solve at least one of the above-mentioned technical problems caused by selecting test vector lengths corresponding to different test modes in incremental testing in the prior art, the present disclosure provides a test vector generation method and device.
[0005] According to a first aspect of the present disclosure, a test vector generation method is provided, including: generating a first length combination set based on different first test vector lengths under different test modes; processing the first length combination set using a pre-trained neural network model corresponding to an object to be tested to generate a fault coverage network set; selecting a test vector combination corresponding to a preset constraint condition in the fault coverage network set to obtain an optimal length combination result; and based on the object to be tested, calling an ATPG tool to generate a test vector corresponding to the optimal length combination result.
[0006] In the embodiments of the present disclosure, since the test vector lengths corresponding to different test modes are selected in the incremental test, they can only rely on engineering experience. After multiple iterations, it is impossible to quickly determine the test vector lengths corresponding to different test modes. By implementing the embodiments of the present disclosure, a first length combination set is generated based on different first test vector lengths in different test modes; the pre-trained neural network model corresponding to the object to be tested is used to process the above first length combination set to generate a fault coverage network set; the test vector combination corresponding to the preset constraint conditions in the above fault coverage network set is selected to obtain the optimal length combination result; and based on the above object to be tested, the ATPG tool is called to generate the test vector corresponding to the above optimal length combination result. By introducing the neural network model and optimizing the test vector length selection process for different test modes, the test vector can be generated more accurately, with high prediction accuracy, and the test cost can be effectively reduced. It only needs to be iterated once when training the neural network model, ensuring that the proportion of test vectors with the lowest fault coverage and the lowest test cost can be selected in one iteration, greatly improving the test efficiency. Directly use the pre-trained neural network model to predict and select the optimal test vector length combination under constraints, that is, the optimal length combination result, use low-precision test vectors to train the neural network, and use the trained network to obtain a high-precision fault coverage network, which can accurately help engineers select the optimal test vector combination and automate the entire test vector selection and generation process, greatly reducing engineers' energy, time and experience dependence, and reasonably utilizing the advantages of different test modes to achieve the best balance between test cost and fault coverage.
[0007] According to an embodiment of the present disclosure, the above-mentioned test mode includes a first test mode and a second test mode; the above-mentioned first test vector lengths based on different test modes generate a first length combination set, including: taking the first test vector length corresponding to the above-mentioned first test mode as the first step length, traversing the first test mode range, and generating several first lengths; on the basis of determining each of the above-mentioned first lengths, taking the first test vector length corresponding to the above-mentioned second test mode as the second step length, traversing the second test mode range, and generating several second lengths; and obtaining the first length combination set according to each of the above-mentioned first lengths and the corresponding second lengths.
[0008] According to an embodiment of the present disclosure, the training process of the above-mentioned neural network model includes: generating a number of test vector length combinations based on different second test vector lengths under different test modes; generating corresponding fault coverage test data under each of the above-mentioned test vector length combinations using the above-mentioned ATPG tool according to the above-mentioned object to be tested and each of the above-mentioned test vector length combinations; preprocessing the above-mentioned fault coverage test data to obtain a test data file; and training a pre-built neural network model based on the above-mentioned test data file.
[0009] According to an embodiment of the present disclosure, the above-mentioned ATPG tool is used according to the above-mentioned object to be tested and each of the above-mentioned test vector length combinations to generate corresponding fault coverage test data under each of the above-mentioned test vector length combinations, including: based on the above-mentioned object to be tested, generating a working directory environment corresponding to each of the above-mentioned test vector length combinations; configuring an ATPG configuration file corresponding to each of the above-mentioned working directory environments; and based on each of the above-mentioned ATPG configuration files, calling the above-mentioned ATPG tool to obtain corresponding fault coverage test data under each of the above-mentioned test vector length combinations.
[0010] According to an embodiment of the present disclosure, based on each of the above-mentioned ATPG configuration files, the above-mentioned ATPG tool is called to obtain the corresponding fault coverage test data under each of the above-mentioned test vector length combinations, including: based on each of the above-mentioned ATPG configuration files, the above-mentioned ATPG tool is called to generate the test log files of the above-mentioned different test modes under each of the above-mentioned working directory environments; and the two-dimensional sequence test data in the test log files of the above-mentioned different test modes under each of the above-mentioned working directory environments are extracted to obtain the corresponding fault coverage test data under each of the above-mentioned test vector length combinations.
[0011] According to an embodiment of the present disclosure, the above-mentioned fault coverage test data includes test vector length information and fault coverage information; the above-mentioned preprocessing of the above-mentioned fault coverage test data to obtain a test data file includes: integrating the test vector length information and fault coverage information corresponding to the above-mentioned different test modes under each of the above-mentioned test vector length combinations to obtain the integrated information data under each of the above-mentioned test vector length combinations; adding a new identification bit to the integrated information data under each of the above-mentioned test vector length combinations to identify the test mode corresponding to the above-mentioned integrated information data; and storing the integrated information data and the corresponding identification bit under each of the above-mentioned test vector length combinations into a file of a preset format to obtain the above-mentioned test data file.
[0012] According to an embodiment of the present disclosure, the above-mentioned different test modes include at least two of: COMP mode, FULLSCAN mode, OPMISR mode, and ELASTIC mode.
[0013] According to an embodiment of the present disclosure, the above-mentioned preset constraint conditions include: at least one of: a coverage condition, a test cost condition, and a vector length condition of the above-mentioned different test modes.
[0014] According to an embodiment of the present disclosure, the above-mentioned neural network model includes at least one of a long short-term memory network model, a gated recurrent unit network model, and a recurrent neural network model.
[0015] According to an embodiment of the present disclosure, the above method is automatically executed using a preset program script; the above program script includes at least one of shell, python, tcl, and c++.
[0016] A second aspect of the present disclosure provides a test vector generation device, which includes: a first combination module, which is used to generate a first length combination set based on different first test vector lengths under different test modes; a model prediction module, which is used to process the first length combination set using a pre-trained neural network model corresponding to the object to be tested to generate a fault coverage network set; a combination selection module, which is used to select the test vector combination corresponding to the preset constraint conditions in the fault coverage network set to obtain the optimal length combination result; and a test vector generation module, which is used to call the ATPG tool based on the object to be tested to generate a test vector corresponding to the optimal length combination result.
[0017] According to an embodiment of the present disclosure, the above-mentioned test mode includes a first test mode and a second test mode; the above-mentioned first combination module includes: a first length submodule, used to take the first test vector length corresponding to the above-mentioned first test mode as the first step length, traverse the first test mode range, and generate a number of first lengths; a second length submodule, used to take the first test vector length corresponding to the above-mentioned second test mode as the second step length, traverse the second test mode range, and generate a number of second lengths based on determining each of the above-mentioned first lengths; and a collection submodule, used to obtain a first length combination set according to each of the above-mentioned first lengths and the corresponding second lengths.
[0018] According to an embodiment of the present disclosure, the above-mentioned device also includes: a second combination module, which is used to generate a number of test vector length combinations based on different second test vector lengths under different test modes; a test data generation module, which is used to generate corresponding fault coverage test data under each of the above-mentioned test vector length combinations according to the above-mentioned object to be tested and each of the above-mentioned test vector length combinations using the above-mentioned ATPG tool; a preprocessing module, which is used to preprocess the above-mentioned fault coverage test data to obtain a test data file; and a model training module, which is used to train a pre-built neural network model based on the above-mentioned test data file.
[0019] According to an embodiment of the present disclosure, the above-mentioned test data generation module includes: an environment generation submodule, which is used to generate a working directory environment corresponding to each of the above-mentioned test vector length combinations based on the above-mentioned object to be tested; a file configuration submodule, which is used to configure an ATPG configuration file corresponding to each of the above-mentioned working directory environments; and a tool calling submodule, which is used to call the above-mentioned ATPG tool based on each of the above-mentioned ATPG configuration files to obtain the corresponding fault coverage test data under each of the above-mentioned test vector length combinations.
[0020] According to an embodiment of the present disclosure, the tool calling submodule includes: a log generating unit, which is used to call the ATPG tool based on each of the ATPG configuration files, and generate test log files of the different test modes in each of the working directory environments; and a data extracting unit, which is used to extract two-dimensional sequence test data in the test log files of the different test modes in each of the working directory environments, and obtain the corresponding fault coverage test data under each of the test vector length combinations.
[0021] According to an embodiment of the present disclosure, the above-mentioned fault coverage test data includes test vector length information and fault coverage information; the above-mentioned preprocessing module includes: a data integration submodule, used to integrate the test vector length information and fault coverage information corresponding to the above-mentioned different test modes under each of the above-mentioned test vector length combinations, and obtain the integrated information data under each of the above-mentioned test vector length combinations; a mode identification submodule, used to add an identification bit to the integrated information data under each of the above-mentioned test vector length combinations to identify the test mode corresponding to the above-mentioned integrated information data; and a file generation submodule, used to store the integrated information data and the corresponding identification bit under each of the above-mentioned test vector length combinations into a file of a preset format to obtain the above-mentioned test data file.
[0022] According to an embodiment of the present disclosure, the above-mentioned different test modes include at least two of: COMP mode, FULLSCAN mode, OPMISR mode, and ELASTIC mode.
[0023] According to an embodiment of the present disclosure, the above-mentioned preset constraint conditions include: at least one of: a coverage condition, a test cost condition, and a vector length condition of the above-mentioned different test modes.
[0024] According to an embodiment of the present disclosure, the above-mentioned neural network model includes at least one of a long short-term memory network model, a gated recurrent unit network model, and a recurrent neural network model.
[0025] According to an embodiment of the present disclosure, the above method is automatically executed using a preset program script; the above program script includes at least one of shell, python, tcl, and c++.
[0026] A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.
[0027] The fourth aspect of the present disclosure further provides a computer-readable storage medium having a computer program or instructions stored thereon, which implements the steps of the above method when the above computer program or instructions are executed by a processor.
[0028] The fifth aspect of the present disclosure further provides a computer program product, including a computer program or instructions, which implement the steps of the above method when the above computer program or instructions are executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] The above contents and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:
[0030] Figure 1 A diagram schematically shows an application scenario of a method and device for generating a test vector according to an embodiment of the present disclosure;
[0031] Figure 2 A flow chart schematically shows a method for generating a test vector according to an embodiment of the present disclosure;
[0032] Figure 3 A length combination flow chart of a method for generating a test vector according to an embodiment of the present disclosure is schematically shown;
[0033] Figure 4 A model training flow chart of a method for generating a test vector according to an embodiment of the present disclosure is schematically shown;
[0034] Figure 5 Another flow chart of a method for generating a test vector according to an embodiment of the present disclosure is schematically shown;
[0035] Figure 6 A flowchart of selecting a test vector combination of a method for generating a test vector according to an embodiment of the present disclosure is schematically shown;
[0036] Figure 7 A configuration test flow chart of a method for generating a test vector according to an embodiment of the present disclosure is schematically shown;
[0037] Figure 8 A diagram schematically showing test data generation of a method for generating a test vector according to an embodiment of the present disclosure;
[0038] Fig. 9 A flowchart of calling an ATPG tool for a method for generating a test vector according to an embodiment of the present disclosure is schematically shown;
[0039] Fig.10 A diagram schematically shows an example of a test log file of a method for generating a test vector according to an embodiment of the present disclosure;
[0040] Fig.11 A data preprocessing flow chart of a method for generating a test vector according to an embodiment of the present disclosure is schematically shown;
[0041] Fig.12 A data integration diagram schematically illustrates a method for generating a test vector according to an embodiment of the present disclosure;
[0042] Fig.13 A result comparison diagram schematically shows a method for generating a test vector according to an embodiment of the present disclosure;
[0043] Fig.14 A structural block diagram schematically shows a test vector generating device according to an embodiment of the present disclosure; and
[0044] Fig.15 A block diagram of an electronic device suitable for implementing a method for generating a test vector according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION
[0045] 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 the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.
[0046] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.
[0047] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.
[0048] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).
[0049] At present, there is no unified method for how to select test vectors for incremental testing. The existing idea of incremental testing is to first generate a relatively long COMP mode test vector, and then generate a COMP mode test vector of a certain length based on the result. On this basis, a relatively long FULLSCAN mode vector is generated, and finally, based on the result, a FULLSCAN mode test vector of a certain length is generated based on the compression mode test vector of a certain length. This method has undergone many iterations. If the test cost and fault coverage shown by the results are still unsatisfactory, it often requires multiple iterations to determine the optimal test vector length, which is time-consuming and inefficient. A lot of computing resources and time will also be wasted during multiple iterations. If the test cost and fault coverage cannot be effectively balanced, it will also lead to a waste of test resources.
[0050] Specifically, an embodiment of the present disclosure provides a test vector generation method, the method comprising: generating a first length combination set based on different first test vector lengths under different test modes; processing the first length combination set using a pre-trained neural network model corresponding to an object to be tested to generate a fault coverage network set; selecting a test vector combination corresponding to a preset constraint condition in the fault coverage network set to obtain an optimal length combination result; and based on the object to be tested, calling an ATPG tool to generate a test vector corresponding to the optimal length combination result.
[0051] It should be noted that the test vector generation method and device provided in the embodiments of the present disclosure can be applied to the field of integrated circuit testing, the field of automatic test vector generation, or any technical field other than the above-mentioned technical fields. The application field of the test vector generation method and device provided in the embodiments of the present disclosure is not specifically limited.
[0052] An embodiment of the present disclosure provides a method for generating a test vector.
[0053] Figure 1 The application scenario diagram of the test vector generation method and device according to the embodiment of the present disclosure is schematically shown.
[0054] like Figure 1 As shown, the application scenario 100 according to this embodiment may include integrated circuit (IC) testing, chip testing, and SoC testing. The network 104 is used to provide a medium for a communication link between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.
[0055] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only examples).
[0056] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.
[0057] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the terminal devices 101, 102, and 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.
[0058] It should be noted that the test vector generation method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the test vector generation device provided in the embodiment of the present disclosure can generally be set in the server 105. The test vector generation method provided in the embodiment of the present disclosure can also be executed by a server or a server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the test vector generation device provided in the embodiment of the present disclosure can also be set in a server or a server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105.
[0059] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.
[0060] The following will be based on Figure 1 The scene described by Figure 2~Figure 13 The method for generating the test vector of the disclosed embodiment is described in detail.
[0061] Figure 2 A flowchart of a method for generating a test vector according to an embodiment of the present disclosure is schematically shown.
[0062] like Figure 2 As shown, the generation of the test vector of this embodiment includes operations S210 to S240, and the generation of the test vector can be automatically performed.
[0063] In operation S210 , a first length combination set is generated based on different first test vector lengths in different test modes.
[0064] In operation S220, the first length combination set is processed using a pre-trained neural network model corresponding to the object to be tested to generate a fault coverage network set.
[0065] In operation S230, a test vector combination corresponding to a preset constraint condition is selected from the fault coverage network set to obtain an optimal length combination result.
[0066] In operation S240 , based on the object to be tested, an ATPG tool is called to generate a test vector corresponding to the optimal length combination result.
[0067] According to an embodiment of the present disclosure, a method for generating a test vector is automatically executed using a preset program script; the program script includes at least one of shell, python, tcl, and c++.
[0068] In operation S210, different test modes include: at least two of COMP mode, FULLSCAN mode, OPMISR mode, and ELASTIC mode. Two combined test modes are usually selected in integrated circuit testing. In the embodiment disclosed in the present invention, two test modes are mainly used for explanation. A test vector length combination for incremental testing of test mode A and test mode B is selected, and a test vector is generated based on the selected optimal length combination result. It should be noted that the two test modes are only used as an example. When it is necessary to generate a test vector in combination with more than two test modes, the test vector generation method disclosed in the present invention can still be used to select the test vector lengths corresponding to different test modes, determine the test vector combination that meets the corresponding constraints, and generate a test vector corresponding to the optimal length combination result.
[0069] COMP mode is a compression mode that analyzes the circuit structure to generate test codes, thereby helping designers test the entire circuit. This test mode is based on the FAN algorithm, which generates test codes for combinational circuits by analyzing the circuit structure. Through the COMP test mode, most faults in the circuit can be effectively detected and covered, thereby improving the reliability and quality of the circuit.
[0070] FULLSCAN mode is full scan mode, which is a type of scan test. In scan testing, storage cells (such as registers) in the circuit are converted into controllable and observable units and connected into one or more shift registers, namely scan chains. Full scan mode means that all registers in the circuit are included in the scan chain, which can improve the fault coverage of the test, but may come at the expense of area and speed.
[0071] OPMISR mode is a test mode that uses the on-chip multi-input discrimination register. OPMISR technology is used to compress test responses to reduce the amount of test data and test time. In test mode, OPMISR can compress the test responses on the scan chain and output the compressed data, thereby reducing the requirements for the cache capacity of the automatic test equipment.
[0072] ELASTIC mode is the embedded logic analyzer mode, which is a test mode that combines scan test technology and embedded logic analyzer functions. This mode embeds a scan chain inside the chip, allowing the capture of changes in internal signals while the chip is running, thereby achieving real-time monitoring and analysis of chip behavior. ELASTIC mode not only provides high-resolution capture capabilities for internal chip signals, but also improves test efficiency and accuracy through integrated test configuration and automated test scripts. This mode is suitable for integrated circuit test scenarios that require high precision and high efficiency.
[0073] It is worth noting that the first test vector length is a high-precision test vector length value preset under different test modes, and the first length combination set is combination data of different preset test vector length values under different test modes.
[0074] In operation S220, the neural network model that has been pre-trained before the object to be tested is used to predict the fault coverage results of any finite pair of test vector combinations of the object to be tested, and output them in a specific format for easy reading. Using the trained model, test vector combinations and fault coverage information of different test modes with arbitrary precision can be extracted. For example, test mode A is sampled every a vector length, and test mode B is sampled every b vector length, and the values of a and b are much smaller than the test vector length selected in the model training stage. A high-precision fault coverage network is obtained, and the fault coverage network can also be exported to the corresponding file for engineers to better read the information.
[0075] The neural network model includes at least one of a long short-term memory network model, a gated recurrent unit network model, and a recurrent neural network model. The object to be tested is a digital logic circuit in a SoC chip.
[0076] In operation S230, the trained model is used to predict the optimal test vector combination based on preset constraints, wherein the preset constraints include at least one of a fault coverage condition, a test cost condition, and a vector length condition of different test modes.
[0077] Preset corresponding constraints and select the optimal test vector combination in the fault coverage network. For example, given a fault coverage requirement (coverage condition), the data will be traversed in the fault coverage network to select the option with the lowest test cost to achieve this fault coverage. For example, given a test vector length limit (vector length condition), the test vector combination that can achieve the highest fault coverage under the test vector length limit will be selected. Taking the test modes of COMP mode and FULLSCAN mode as an example, the fault coverage to be achieved is set to 98.35%, and the combination with the lowest test cost under this fault coverage will be automatically provided: COMP11400+FULLSCAN16, with a predicted coverage of 98.36%.
[0078] In operation S240, based on the test vector combination corresponding to the selected optimal length combination result, a new working directory is created through a shell script, and the ATPG configuration file is modified according to the final test vector combination, and the ATPG tool is called to generate the final test vector. Taking the test mode as COMP mode and FULLSCAN mode as an example, the optimal length combination result is COMP11400+FULLSCAN16, and the ATPG tool is automatically called by the python script to generate the test vector of COMP11400+FULLSCAN16, and the actual fault coverage result is 98.36%, and the total number of test cycles is 8,434,187. The number of test cycles can be mapped to the test cost.
[0079] It should be noted that ATPG, or Automatic Test Pattern Generation, is a test technology for digital circuits. It is used to detect faults in digital circuits by automatically generating test vectors. These test vectors are based on certain algorithms and are designed to detect all possible faults in the circuit by analyzing the description of the circuit. ATPG has a wide range of applications, including but not limited to integrated circuits, microprocessors, and other digital logic devices. By using ATPG, the need for testing can be taken into account during the design phase, thereby improving the circuit's design for testability (DFT) and ensuring that the circuit can be effectively tested after production, ensuring that its functions and performance meet the design requirements. The ATPG process includes steps such as error selection, finding the initial input vector, determining the transmission vector set, and comparing the results. Through these steps, ATPG can generate test vectors for specific fault models. These vectors can expose the signals inside the circuit, allowing testers to fully test the circuit without directly accessing the internal nodes of the circuit.
[0080] By introducing a neural network model, the test vector length selection process of the test mode is optimized, ensuring that the proportion of test vectors that meet the fault coverage requirements and have the lowest test cost can be selected in one iteration, greatly improving the test efficiency. The entire test vector selection process is automated, reducing the reliance on engineer experience, and making reasonable use of the advantages of different test modes to achieve the best balance between test cost and fault coverage.
[0081] Figure 3 The length combination flow chart of the method for generating a test vector according to an embodiment of the present disclosure is schematically shown.
[0082] like Figure 3 As shown, according to an embodiment of the present disclosure, the test mode includes a first test mode and a second test mode; in operation S210, based on different first test vector lengths in different test modes, a first length combination set is generated, including operations S310-S330.
[0083] In operation S310, a first test vector length corresponding to a first test pattern is used as a first step length, a first test pattern range is traversed, and a plurality of first lengths are generated.
[0084] In operation S320, based on the determination of each first length, the second test pattern range is traversed with the first test vector length corresponding to the second test pattern as a second step length to generate a plurality of second lengths.
[0085] In operation S330, a first length combination set is obtained according to each first length and the corresponding second length.
[0086] In the embodiment of the present disclosure, taking the first test mode as COMP mode, the second test mode as FULLSCAN mode, the first step length as 200, the first test mode range as 0 to 15000, the second step length as 16, the second test mode range as 16 to 400 as an example, the COMP mode test vector length is based on every 200 lengths, traversing from 0 to 15000, and continuing to generate FULLSCAN mode test vectors with every 16 lengths at each base point, traversing from 16 to 400, thereby generating 75×25 test vector length combinations, i.e., the first length combination set. A fault coverage result can be obtained by inputting each test vector length combination into the trained neural network model, and a 75×25 high-precision fault coverage network is generated using the trained neural network model, and the results are uniformly output to an Excel spreadsheet.
[0087] Figure 4 A model training flow chart of a method for generating a test vector according to an embodiment of the present disclosure is schematically shown;
[0088] Figure 5Another flowchart of a method for generating a test vector according to an embodiment of the present disclosure is schematically shown.
[0089] like Figure 5 As shown, according to an embodiment of the present disclosure, before operation S210, the neural network is also trained, such as Figure 4 As shown, the training process of the neural network model includes operations S410-S440.
[0090] In operation S410, a plurality of test vector length combinations are generated based on different second test vector lengths in different test modes;
[0091] In operation S420, according to the object to be tested and each test vector length combination, the ATPG tool is used to generate corresponding fault coverage test data under each test vector length combination.
[0092] In operation S430, the fault coverage test data is pre-processed to obtain a test data file.
[0093] In operation S440, a pre-built neural network model is trained based on the test data file.
[0094] According to an embodiment of the present disclosure, when different test modes include a first test mode and a second test mode, in operation S410, based on different second test vector lengths in different test modes, several test vector length combinations are generated, including: taking the second test vector length corresponding to the first test mode as the third step length, traversing the third test mode range, and generating several third lengths; on the basis of determining each third length, taking the second test vector length corresponding to the second test mode as the fourth step length, generating several fourth lengths; and taking each third length and the corresponding fourth length as each test vector length combination. In operation S420, an ATPG tool is used to generate a finite pair of test vector combinations using different test modes, and test results such as the current coverage fault number and fault coverage rate are obtained.
[0095] Taking the first test mode as COMP mode and the second test mode as FULLSACN mode as an example, the shell script is run to call the ATPG tool in parallel, and a limited number of test vector length combinations of COMP mode and FULLSACN mode are generated at the same time. For the test vector length of the COMP mode, from the third test mode range (such as 0 to 12000), every third step length (such as 1000) is taken as a basic point, and on this basis, the FULLSCAN mode test vector with a vector length of the fourth step length (such as 400) is generated. The test vector length combinations are (COMP1000, FULLSCAN400), (COMP2000, FULLSCAN400), (COMP3000, FULLSCAN400), ..., (COMP12000, FULLSCAN400), a total of 12 groups of test vector length combinations. Run the shell script to call the ATPG tool in parallel to generate COMP1000+FULLSCAN400, COMP2000+FULLSCAN400, COMP3000+FULLSCAN400, ..., COMP12000+FULLSCAN400, a total of 12 groups of combination test vectors.
[0096] The second test vector length is a low-precision test vector length value preset under different test modes. Therefore, the fault coverage test data of the low-precision test vector combination is used to train the neural network model, and the high-precision test vector combination is input into the trained neural network to predict a high-precision fault coverage network. Figure 6 The flowchart of selecting a test vector combination of the test vector generation method according to the embodiment of the present disclosure is schematically shown. The test vector combination selection process includes operations S410-S440 and S210-S230. Figure 6 It is shown that by using the test data of a limited number of low-precision test vector length combinations to train the neural network, a neural network model that can effectively predict the fault coverage of any high-precision test vector length combination is obtained. Based on the high-precision fault coverage prediction of the neural network model, given the corresponding constraints, the selection of the optimal test vector length combination can be efficiently realized, the optimal test vector length combination, that is, the optimal length combination result, is selected, and the test vector length selection process is accelerated, thereby greatly improving the test efficiency.
[0097] In operation S440, the structure of the neural network model is xHZy. A neural network model of xHZy is built using a program script, where x is a multi-dimensional input, including information such as the test length of different test modes; y is the fault coverage; H is a hidden layer of the neural network; and Z is a fully connected layer, which is used to map the output of H to y. The number of neurons in the hidden layer of the neural network, the learning rate and other related parameters are set, and the CSV data in the n test data files are put into the model for a limited number of training times.
[0098] Taking the test modes of COMP mode and FULLSCAN mode as an example, use Pytorch to build an x-256-256-y long short-term memory network (LSTM) model, where x is a 3D input [COMP length, FULLSCAN length, flag], the flag is 0 or 1, indicating COMP mode or FULLSCAN mode; y is the fault coverage; the number of hidden layer neurons is set to 256, and the learning rate is set to 0.0001. Use the above 12 sets of sequence data to train the network, each set of data is trained 30,000 times, and the training time only takes 1-2 hours.
[0099] Figure 7 A configuration test flow chart of a method for generating a test vector according to an embodiment of the present disclosure is schematically shown.
[0100] like Figure 7 As shown, according to an embodiment of the present disclosure, in operation S420, according to the object to be tested and each test vector length combination, using the ATPG tool, corresponding fault coverage test data under each test vector length combination is generated, including operations S710-S730.
[0101] In operation S710, a working directory environment corresponding to each test vector length combination is generated based on the object to be tested.
[0102] In operation S720 , an ATPG configuration file corresponding to each working directory environment is configured.
[0103] In operation S730 , based on each ATPG configuration file, an ATPG tool is called to obtain corresponding fault coverage test data under each test vector length combination.
[0104] According to an embodiment of the present disclosure, taking the use of a shell script to call an ATPG tool to obtain test data as an example, the test mode in the shell script is set to A, B, and a finite number of test vector lengths are selected, that is, the test vector length combinations are (x1, y1), (x2, y2), ..., (xn, yn), and the shell script is run in parallel to generate n groups of working directory environments. Each combination of test vectors will correspond to a different working directory environment, and the corresponding ATPG configuration files in each environment are configured, so as to facilitate calling the ATPG tool in the corresponding ATPG configuration file and obtaining the fault coverage test data under the test vector length combination in the respective working directory environment.
[0105] Figure 8 A diagram schematically shows a test data generation method according to an embodiment of the present disclosure for generating a test vector.
[0106] like Figure 8 As shown, according to an embodiment of the present disclosure, in operation S730, based on each ATPG configuration file, the ATPG tool is called to obtain the corresponding fault coverage test data under each test vector length combination, including operations S810-S820.
[0107] In operation S810 , based on each ATPG configuration file, the ATPG file is called to generate a test log file of a different test mode in each working directory environment.
[0108] In operation S820, two-dimensional sequence test data in the test log files of different test modes under each working directory environment is extracted to obtain corresponding fault coverage test data under each test vector length combination.
[0109] Take using shell script to call ATPG tool to obtain test data as an example. Fig. 9 The flowchart of calling the ATPG tool of the test vector generation method according to the embodiment of the present disclosure is schematically shown. Fig. 9 As shown, for each ATPG configuration file, the ATPG tool is called by a shell script to generate test data under the corresponding test mode, wherein the test mode in the shell script is set to test mode A and test mode B. In operation S810, the ATPG tool is called in different working directory environments, and the ATPG tool generates corresponding test data in the working directory according to the test mode and test vector length in the configuration file. These test data will be automatically exported by the ATPG tool to the corresponding log file, that is, test log files with different test modes are generated in different working directory environments. Fig.10 The following schematically shows an example of a test log file of a method for generating a test vector according to an embodiment of the present disclosure. Fig.10 As shown, in the same working directory, different test modes will generate different test log files, that is, the test log files of test mode A and test mode B will be generated in the same working directory. Taking the log information example of the process of generating test vectors by the ATPG tool, it can be seen that the test log file contains key information such as the length of the test vector, the faults detected, and the fault coverage currently achieved. Focus on the Eff column under Tests and the Tmode column under ATcov. These two sequences are the test vector length and fault coverage information of concern. These two sequences are two-dimensional sequence test data to be extracted. In operation S820, the generated Log file is processed using a shell script to extract the test data of the two columns of Eff and Tmode. This information is the two-dimensional sequence information that changes with the increase of the test vector length of a certain test mode. Two sets of two-dimensional sequence test data will be extracted in the working directory of the test vector combination (x1, y1), that is, the corresponding fault coverage test data under the test vector combination (x1, y1) is obtained.
[0110] Fig.11 A data preprocessing flow chart of a method for generating a test vector according to an embodiment of the present disclosure is schematically shown.
[0111] like Fig.11 As shown, according to an embodiment of the present disclosure, the fault coverage test data includes test vector length information and fault coverage rate information; in operation S430, the fault coverage test data is pre-processed to obtain a test data file, including operations S1110-S1130.
[0112] In operation S1110, test vector length information and fault coverage information corresponding to different test modes under each test vector length combination are integrated to obtain integrated information data under each test vector length combination.
[0113] In operation S1120, a flag is added to the integrated information data under each test vector length combination to identify the test mode corresponding to the integrated information data.
[0114] In operation S1130, the integrated information data and the corresponding identification bits under each test vector length combination are stored in a file of a preset format to obtain a test data file.
[0115] According to an embodiment of the present disclosure, Fig.12 The data integration diagram of the test vector generation method according to the embodiment of the present disclosure is schematically shown. Fig.12As shown, in the working directory of the same test vector length combination, the data of the two sets of test vector length information and fault coverage information are integrated together, and a new identification bit is added to identify the test mode, and finally the data is saved in a CSV file, where p and q represent the current fault coverage of test mode A and test mode B respectively, and the first bit of p and q is the identification bit, identification bit 0 represents test mode A, and identification bit 1 represents test mode B. The test vector combination data in different working directories are saved in files with different preset formats (such as: CSV files), and a total of n test data files (such as: CSV data files) are obtained.
[0116] Taking the 12 groups of test vector length combinations in the above operation S410 embodiment as an example, a shell script is used to automatically collect and integrate all test data in the generated test log file, and save the data of different combinations into different CSV files respectively, and finally obtain 12 CSV files.
[0117] By implementing the embodiments of the present disclosure, a first length combination set is generated based on different first test vector lengths under different test modes; a pre-trained neural network model corresponding to the object to be tested is used to process the first length combination set to generate a fault coverage network set; a test vector combination corresponding to the preset constraint conditions in the fault coverage network set is selected to obtain the optimal length combination result; and based on the object to be tested, an ATPG tool is called to generate a test vector corresponding to the optimal length combination result. By introducing a neural network model and optimizing the test vector length selection process of different test modes, a test vector can be generated more accurately, with high prediction accuracy, and the test cost can be effectively reduced. Only one iteration is required when training the neural network model, ensuring that the proportion of test vectors with the lowest fault coverage and the lowest test cost can be selected in one iteration, greatly improving the test efficiency. The optimal test vector length combination, that is, the optimal length combination result, is directly selected under the prediction and constraint conditions through a pre-trained neural network model, without iteration, automating the entire test vector selection and generation process, greatly reducing the energy, time and experience dependence of engineers, and reasonably utilizing the advantages of different test modes to achieve the best balance between test cost and fault coverage.
[0118] To further illustrate the practical application effect of the present invention, taking the test modes COMP and FULLSCAN as examples, the test results of the test vector generation method of the present invention and the traditional test vector generation method are compared. Fig.13 The result comparison diagram of the test vector generation method according to the embodiment of the present disclosure is schematically shown, as shown in FIG. Fig.13As shown, when performing ATPG incremental testing on a SoC chip module in the traditional method, multiple iterations are required, and finally the combination of COMP 10000 + FULLSCAN 100 is selected, with a fault coverage of 98.35% and a test cycle number of 15,661,251 (the test cycle number can be mapped to the test cost). Using the test vector generation method, the SoC chip module is subjected to ATPG incremental testing, and finally the combination of COMP11400+FULLSCAN16 is selected, with a fault coverage of 98.36% and a test cycle number of 8,434,187. On the premise of obtaining the same coverage, the introduction of the neural network model in the present disclosure can save half the test cost compared with the traditional incremental test method, and the system has high prediction accuracy, which can effectively reduce the test cost. The entire process can be automatically completed by only one iteration when training the neural network, which greatly reduces the energy and time of engineers and improves test efficiency.
[0119] Based on the above test vector generation method, the present disclosure also provides a test vector generation device. Fig.14 The device is described in detail.
[0120] Fig.14 The structural block diagram of a test vector generation device according to an embodiment of the present disclosure is schematically shown.
[0121] like Fig.14 As shown, the test vector generating device 1400 of this embodiment includes a first combining module 1410 , a model prediction module 1420 , a combination selecting module 1430 and a test vector generating module 1440 .
[0122] The first combination module 1410 is used to generate a first length combination set based on different first test vector lengths in different test modes. In one embodiment, the first combination module 1410 can be used to perform the operation S210 described above, which will not be described in detail here.
[0123] The model prediction module 1420 is used to process the first length combination set using the pre-trained neural network model corresponding to the object to be tested to generate a fault coverage network set. In one embodiment, the model prediction module 1420 can be used to perform the operation S220 described above, which will not be repeated here.
[0124] The combination selection module 1430 is used to select the test vector combination corresponding to the preset constraint condition in the fault coverage network set to obtain the optimal length combination result. In one embodiment, the combination selection module 1430 can be used to perform the operation S230 described above, which will not be repeated here.
[0125] The test vector generation module 1440 is used to call the ATPG tool based on the object to be tested to generate a test vector corresponding to the optimal length combination result. In one embodiment, the test vector generation module 1440 can be used to perform the operation S240 described above, which will not be repeated here.
[0126] The test vector generating device 1400 further includes a second combining module 1450 , a test data generating module 1460 , a preprocessing module 1470 and a model training module 1480 .
[0127] The second combination module 1450 is used to generate a plurality of test vector length combinations based on different second test vector lengths in different test modes. In one embodiment, the second combination module 1450 can be used to perform the operation S410 described above, which will not be described in detail here.
[0128] The test data generation module 1460 is used to generate corresponding fault coverage test data under each test vector length combination according to the object to be tested and each test vector length combination using the ATPG tool. In one embodiment, the test data generation module 1460 can be used to perform the operation S420 described above, which will not be repeated here.
[0129] The preprocessing module 1470 is used to preprocess the fault coverage test data to obtain a test data file. In one embodiment, the preprocessing module 1470 can be used to perform the operation S430 described above, which will not be described in detail here.
[0130] The model training module 1480 is used to train the pre-built neural network model based on the test data file. In one embodiment, the model training module 1480 can be used to perform the operation S440 described above, which will not be repeated here.
[0131] According to an embodiment of the present disclosure, the test mode includes a first test mode and a second test mode; the first combination module 1410 includes: a first length submodule, used to traverse the first test mode range with the first test vector length corresponding to the first test mode as the first step length, and generate a plurality of first lengths; a second length submodule, used to traverse the second test mode range with the first test vector length corresponding to the second test mode as the second step length on the basis of determining each first length, and generate a plurality of second lengths; and a collection submodule, used to obtain a first length combination set according to each first length and the corresponding second length.
[0132] According to an embodiment of the present disclosure, the test data generation module 1460 includes: an environment generation submodule, which is used to generate a working directory environment corresponding to each test vector length combination based on the object to be tested; a file configuration submodule, which is used to configure the ATPG configuration file corresponding to each working directory environment; and a tool calling submodule, which is used to call the ATPG tool based on each ATPG configuration file to obtain the corresponding fault coverage test data under each test vector length combination.
[0133] According to an embodiment of the present disclosure, a tool calling submodule includes: a log generating unit, which is used to call the ATPG tool based on each ATPG configuration file to generate a test log file of a different test mode in each working directory environment; and a data extracting unit, which is used to extract two-dimensional sequence test data in the test log file of the different test mode in each working directory environment to obtain the corresponding fault coverage test data under each test vector length combination.
[0134] According to an embodiment of the present disclosure, fault coverage test data includes test vector length information and fault coverage information; the preprocessing module 1470 includes: a data integration submodule, used to integrate the test vector length information and fault coverage information corresponding to different test modes under each test vector length combination, and obtain the integrated information data under each test vector length combination; a mode identification submodule, used to add an identification bit to the integrated information data under each test vector length combination to identify the test mode corresponding to the integrated information data; and a file generation submodule, used to store the integrated information data and the corresponding identification bit under each test vector length combination into a file of a preset format to obtain a test data file.
[0135] According to an embodiment of the present disclosure, different test modes include at least two of: COMP mode, FULLSCAN mode, OPMISR mode, and ELASTIC mode.
[0136] According to an embodiment of the present disclosure, the preset constraint conditions include: at least one of: coverage conditions, test cost conditions, and vector length conditions of different test modes.
[0137] According to an embodiment of the present disclosure, the neural network model includes at least one of a long short-term memory network model, a gated recurrent unit network model, and a recurrent neural network model.
[0138] According to an embodiment of the present disclosure, the method is automatically executed using a preset program script; the program script includes at least one of shell, python, tcl, and c++.
[0139] According to an embodiment of the present disclosure, any multiple modules among the first combination module 1410, the model prediction module 1420, the combination selection module 1430, the test vector generation module 1440, the second combination module 1450, the test data generation module 1460, the preprocessing module 1470 and the model training module 1480 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the first combination module 1410, the model prediction module 1420, the combination selection module 1430, the test vector generation module 1440, the second combination module 1450, the test data generation module 1460, the preprocessing module 1470 and the model training module 1480 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging circuits, or implemented in any one of the three implementation methods of software, hardware and firmware, or in an appropriate combination of any of them. Alternatively, at least one of the first combination module 1410, the model prediction module 1420, the combination selection module 1430, the test vector generation module 1440, the second combination module 1450, the test data generation module 1460, the preprocessing module 1470 and the model training module 1480 can be at least partially implemented as a computer program module, which can perform the corresponding function when the computer program module is executed.
[0140] Fig.15 A block diagram of an electronic device suitable for implementing a method for generating a test vector according to an embodiment of the present disclosure is schematically shown.
[0141] like Fig.15 As shown, the electronic device 1500 according to an embodiment of the present disclosure includes a processor 1501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1502 or a program loaded from a storage part 1508 to a random access memory (RAM) 1503. The processor 1501 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 1501 may also include an onboard memory for caching purposes. The processor 1501 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.
[0142] In RAM 1503, various programs and data required for the operation of electronic device 1500 are stored. Processor 1501, ROM 1502 and RAM 1503 are connected to each other through bus 1504. Processor 1501 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 1502 and / or RAM 1503. It should be noted that the program can also be stored in one or more memories other than ROM 1502 and RAM 1503. Processor 1501 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in one or more memories.
[0143] According to an embodiment of the present disclosure, the electronic device 1500 may further include an input / output (I / O) interface 1505, which is also connected to the bus 1504. The electronic device 1500 may further include one or more of the following components connected to the I / O interface 1505: an input portion 1506 including a keyboard, a mouse, etc.; an output portion 1507 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 1508 including a hard disk, etc.; and a communication portion 1509 including a network interface card such as a LAN card, a modem, etc. The communication portion 1509 performs communication processing via a network such as the Internet. A drive 1510 is also connected to the I / O interface 1505 as needed. A removable medium 1511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1510 as needed, so that a computer program read therefrom is installed into the storage portion 1508 as needed.
[0144] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.
[0145] According to an embodiment of the present disclosure, a computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus, or a device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 1502 and / or RAM 1503 described above and / or one or more memories other than ROM 1502 and RAM 1503.
[0146] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the test vector generation method provided by the embodiment of the present disclosure.
[0147] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 1501. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.
[0148] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 1509, and / or installed from the removable medium 1511. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.
[0149] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 1509, and / or installed from the removable medium 1511. When the computer program is executed by the processor 1501, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, means, module, unit, etc. described above can be implemented by a computer program module.
[0150] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).
[0151] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.
[0152] It will be appreciated by those skilled in the art that the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations and / or combinations are not explicitly described in the present disclosure. In particular, the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways without departing from the spirit and teachings of the present disclosure. All of these combinations and / or combinations fall within the scope of the present disclosure.
[0153] The embodiments of the present disclosure are described above. However, these embodiments are only for the purpose of illustration and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. The scope of the present disclosure is defined by the attached claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present disclosure.
Claims
1. A test vector generation method, characterized in that: The method comprises: Generate a first length combination set based on different first test vector lengths in different test modes; Using a pre-trained neural network model corresponding to the object to be tested, processing the first length combination set to generate a fault coverage network set; Selecting a combination of test vectors corresponding to the preset constraint conditions in the fault coverage network set to obtain an optimal length combination result; and Based on the object to be tested, an ATPG tool is called to generate a test vector corresponding to the optimal length combination result.
2. The method according to claim 1, characterized in that The test mode includes a first test mode and a second test mode; The generating of the first length combination set based on different first test vector lengths in different test modes includes: Taking the first test vector length corresponding to the first test mode as the first step length, traversing the first test mode range to generate a plurality of first lengths; On the basis of determining each of the first lengths, taking the first test vector length corresponding to the second test pattern as a second step length, traversing the second test pattern range to generate a plurality of second lengths; and A first length combination set is obtained according to each of the first lengths and the corresponding second length.
3. The method according to claim 1, characterized in that The training process of the neural network model includes: Based on different second test vector lengths in different test modes, generating a plurality of test vector length combinations; According to the object to be tested and each combination of the test vector lengths, using the ATPG tool, generating corresponding fault coverage test data under each combination of the test vector lengths; Preprocessing the fault coverage test data to obtain a test data file; and Based on the test data file, a pre-built neural network model is trained.
4. The method according to claim 3, characterized in that The generating, according to the object to be tested and each of the test vector length combinations, corresponding fault coverage test data under each of the test vector length combinations by using the ATPG tool includes: Based on the object to be tested, generating a working directory environment corresponding to each of the test vector length combinations; Configuring an ATPG configuration file corresponding to each of the working directory environments; and Based on each of the ATPG configuration files, the ATPG tool is called to obtain corresponding fault coverage test data under each of the test vector length combinations.
5. The method according to claim 4, characterized in that The step of calling the ATPG tool based on each of the ATPG configuration files to obtain the corresponding fault coverage test data under each of the test vector length combinations includes: Based on each of the ATPG configuration files, calling the ATPG tool to generate test log files of the different test modes in each of the working directory environments; and Extract the two-dimensional sequence test data in the test log files of the different test modes under each of the working directory environments to obtain the corresponding fault coverage test data under each of the test vector length combinations.
6. The method according to claim 3, characterized in that The fault coverage test data includes test vector length information and fault coverage rate information; The preprocessing of the fault coverage test data to obtain a test data file includes: Integrate the test vector length information and fault coverage information corresponding to the different test modes under each of the test vector length combinations to obtain integrated information data under each of the test vector length combinations; adding a new identification bit to the integrated information data under each of the test vector length combinations to identify the test mode corresponding to the integrated information data; and The integrated information data and the corresponding identification bits under each of the test vector length combinations are stored in a file of a preset format to obtain the test data file.
7. The method according to claim 1, characterized in that The different test modes include: at least two of: COMP mode, FULLSCAN mode, OPMISR mode, and ELASTIC mode.
8. The method according to claim 1, characterized in that The preset constraint conditions include: at least one of: a coverage condition, a test cost condition, and a vector length condition of the different test modes.
9. The method according to any one of claims 1 to 6, characterized in that: The neural network model includes at least one of a long short-term memory network model, a gated recurrent unit network model, and a recurrent neural network model.
10. The method according to any one of claims 1 to 6, characterized in that: The method is automatically executed using a preset program script; the program script includes at least one of shell, python, tcl, and c++.
11. A test vector generating device, characterized in that: The device comprises: A first combination module, used for generating a first length combination set based on different first test vector lengths in different test modes; A model prediction module, used to process the first length combination set using a pre-trained neural network model corresponding to the object to be tested, so as to generate a fault coverage network set; A combination selection module is used to select a combination of test vectors corresponding to the preset constraint conditions in the fault coverage network set to obtain an optimal length combination result; and The test vector generation module is used to call the ATPG tool based on the object to be tested to generate a test vector corresponding to the optimal length combination result.
12. The device according to claim 11, characterized in that The device also includes: A second combination module, used to generate a plurality of test vector length combinations based on different second test vector lengths in different test modes; A test data generation module, used to generate corresponding fault coverage test data under each test vector length combination using the ATPG tool according to the object to be tested and each test vector length combination; A preprocessing module, used for preprocessing the fault coverage test data to obtain a test data file; and The model training module is used to train a pre-built neural network model based on the test data file.
Citation Information
Cited By
Multi-channel chip packaging test method and system
CN119619814A
A multi-channel chip packaging test method and system
CN119619814B
