Signal simulation test method and apparatus for stacked chips

By building a signal simulation test platform and a multi-task attention mechanism chip performance analysis model, using multi-layer structure information and chip operating condition parameters, the problem of insufficient accuracy of stacked chip signal simulation test is solved, and efficient and accurate signal transmission performance is achieved and timing integrity analysis.

CN118569202BActive Publication Date: 2024-11-05SHENZHEN ZSIPAK TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411037401.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-11-05
Estimated Expiration
2044-07-31

AI Technical Summary

Technical Problem

The existing stacked chip signal simulation test methods are difficult to accurately evaluate the signal transmission performance and timing integrity under multi-layer structures and high-density wiring, resulting in inaccurate test results.

Method used

By building a signal simulation test platform, using multi-layer structure information, wiring density information and a variety of chip operating environment parameters, signal simulation, data fusion, Fourier transform and feature extraction are performed, combined with the multi-task attention mechanism chip performance analysis model , generate signal transmission performance and timing integrity analysis results.

Benefits of technology

It improves the accuracy of stacked chip signal simulation test, reduces manual intervention and testing costs, enhances the stability and reliability of test data, and can accurately simulate signal characteristics in different chip operating conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118569202B_ABST
    Figure CN118569202B_ABST
Patent Text Reader

Abstract

The present application relates to the field of chip testing technology, and discloses a signal simulation test method and device for stacked chips. The method includes: constructing a signal simulation test platform and performing a signal simulation test on a target stacked chip to obtain a plurality of first test signal data; performing signal fusion to obtain a second test signal data; performing Fourier transform and data set classification to obtain transmission rate data and timing delay data; constructing a chip working condition transmission rate relationship curve and performing feature extraction to obtain a plurality of chip working condition transmission rate features; constructing a chip working condition timing delay relationship curve and performing feature extraction to obtain a plurality of chip working condition timing delay features; performing stacked chip signal simulation test performance analysis through a multi-task attention mechanism chip performance analysis model to obtain a signal transmission performance analysis result and a timing integrity analysis result. The present application improves the accuracy of signal simulation test of stacked chips.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of chip testing technology, and in particular to a signal simulation testing method and device for stacked chips. Background Art

[0002] As a new integrated circuit packaging technology, stacked chips have the advantages of high integration, small size and low power consumption, and are therefore increasingly widely used in chip design.

[0003] However, due to the multi-layer structure and high-density wiring of stacked chips, the impact of signal transmission performance and timing integrity on performance becomes more important. Therefore, an efficient and accurate signal simulation test method and device are needed to evaluate the signal performance of stacked chips under different working conditions. Summary of the invention

[0004] The present application provides a signal simulation test method and device for stacked chips, and the present application improves the accuracy of signal simulation test for stacked chips.

[0005] In a first aspect, the present application provides a signal simulation test method for stacked chips, the signal simulation test method for stacked chips comprising:

[0006] Construct a signal simulation test platform and perform signal simulation tests on target stacked chips respectively through various chip operating environment parameters to obtain a plurality of first test signal data for each chip operating environment parameter;

[0007] Performing signal fusion on the plurality of first test signal data by a preset strong classifier to obtain second test signal data of each chip operating environment parameter;

[0008] Performing Fourier transform on the second test signal data to obtain target chip test data of each chip working environment parameter, and performing data set classification on the target chip test data to obtain transmission rate data and timing delay data of each chip working environment parameter;

[0009] Constructing a corresponding chip working condition transmission rate relationship curve according to the transmission rate data and the multiple chip working condition environment parameters, and performing curve feature extraction and feature combination weighting on the chip working condition transmission rate relationship curve to obtain multiple chip working condition transmission rate features;

[0010] Constructing a corresponding chip operating condition timing delay relationship curve according to the timing delay data and the multiple chip operating condition environment parameters, and performing curve feature extraction and feature combination weighting on the chip operating condition timing delay relationship curve to obtain multiple chip operating condition timing delay features;

[0011] The transmission rate characteristics of the multiple chip operating conditions and the timing delay characteristics of the multiple chip operating conditions are input into a preset multi-task attention mechanism chip performance analysis model to perform stacked chip signal simulation test performance analysis to obtain signal transmission performance analysis results and timing integrity analysis results.

[0012] In a second aspect, the present application provides a signal simulation test device for stacked chips, the signal simulation test device for stacked chips comprising:

[0013] A test module is used to build a signal simulation test platform and perform signal simulation tests on target stacked chips respectively through a variety of chip operating environment parameters to obtain a plurality of first test signal data for each chip operating environment parameter;

[0014] A fusion module, used for performing signal fusion on the plurality of first test signal data through a preset strong classifier to obtain second test signal data of each chip operating environment parameter;

[0015] A transformation module, used to perform Fourier transformation on the second test signal data to obtain target chip test data of each chip working environment parameter, and perform data set classification on the target chip test data to obtain transmission rate data and timing delay data of each chip working environment parameter;

[0016] A first extraction module is used to construct a corresponding chip working condition transmission rate relationship curve according to the transmission rate data and the multiple chip working condition environment parameters, and perform curve feature extraction and feature combination weighting on the chip working condition transmission rate relationship curve to obtain multiple chip working condition transmission rate features;

[0017] A second extraction module is used to construct a corresponding chip operating condition timing delay relationship curve according to the timing delay data and the multiple chip operating condition environment parameters, and perform curve feature extraction and feature combination weighting on the chip operating condition timing delay relationship curve to obtain multiple chip operating condition timing delay features;

[0018] The analysis module is used to input the transmission rate characteristics of the multiple chip working conditions and the timing delay characteristics of the multiple chip working conditions into a preset multi-task attention mechanism chip performance analysis model to perform stacked chip signal simulation test performance analysis to obtain signal transmission performance analysis results and timing integrity analysis results.

[0019] The third aspect of the present application provides a computer device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the computer device executes the above-mentioned signal simulation test method for stacked chips.

[0020] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned signal simulation test method for stacked chips.

[0021] In the technical solution provided by the present application, the signal simulation test of the target stacked chip is realized through the signal simulation test platform by using multi-layer structure information, wiring density information and various chip working environment parameters. This automated method can improve test efficiency, reduce manual intervention and reduce test costs. Through the comprehensive calculation of signal simulation function, environmental parameter influence function and signal noise function, the signal characteristics under different chip working conditions can be accurately simulated, including signal amplitude, attenuation, frequency, phase, etc. This helps to obtain accurate test data. Using a strong classifier to fuse the first test signal data can reduce the impact of noise on the test results, improve signal quality, and enhance the stability and reliability of test data. Through Fourier transform and data set classification, transmission rate data and timing delay data can be extracted from the second test signal data. These data can be used to analyze the signal transmission performance and timing integrity of the stacked chip. By constructing a chip working condition transmission rate relationship curve and a timing delay relationship curve, and performing feature extraction, combination and weighting, multiple chip working condition transmission rate features and timing delay features can be generated. These features can be input into a multi-task attention mechanism chip performance analysis model to perform stacked chip signal simulation test performance analysis, thereby improving the accuracy of the signal simulation test of the stacked chip. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0023] Figure 1 A schematic diagram of an embodiment of a signal simulation test method for stacked chips in an embodiment of the present application;

[0024] Figure 2 Schematic diagram of an embodiment of a signal simulation test device for stacked chips in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The embodiment of the present application provides a signal simulation test method and device for stacked chips. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 , an embodiment of the signal simulation test method of the stacked chip in the embodiment of the present application includes:

[0027] Step 101, constructing a signal simulation test platform and performing signal simulation tests on target stacked chips respectively through a variety of chip operating environment parameters to obtain a plurality of first test signal data for each chip operating environment parameter;

[0028] It is understandable that the execution subject of the present application may be a signal simulation test device for stacked chips, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0029] Specifically, a signal simulation test platform is constructed to obtain the multi-layer structure information and wiring density information of the target stacked chip. Based on the multi-layer structure information and wiring density information, a variety of chip operating environment parameters, such as temperature, voltage, and frequency, are set. These parameters simulate various environmental conditions that the chip may encounter in actual use. Through the signal simulation test platform, the target stacked chip is subjected to signal simulation testing according to a variety of chip operating environment parameters. During the simulation test, the platform calculates the ideal output signal data under each chip operating environment parameter through a signal simulation function. The signal simulation function takes into account multiple factors such as signal amplitude, attenuation coefficient, signal frequency, and signal phase, and also includes the interaction terms between chip operating environment parameters and time, so as to more accurately simulate the ideal signal output of the chip under specific working conditions. In order to further improve the accuracy of the test, an environmental parameter influence function and a signal noise function are introduced. The environmental parameter influence function is used to calculate the environmental parameter influence coefficient of each chip operating environment parameter, and the influence of environmental parameters on the signal is comprehensively evaluated through linear influence coefficients and interactive influence coefficients. The signal noise function calculates the noise signal data under each chip operating environment parameter. This function takes into account factors such as noise intensity, random noise function, noise frequency, and noise phase constant. By simulating the noise that may occur in the actual working environment, the test results are closer to the actual situation. The total signal output function combines the ideal output signal data, environmental parameter influence coefficients, and noise signal data to generate multiple first test signal data under each chip operating environment parameter. The total signal output function not only takes into account the influence of ideal signals and environmental parameters, but also integrates the influence of noise signals, so that the performance of stacked chips under various environmental conditions can be more comprehensively evaluated.

[0030] Step 102: performing signal fusion on a plurality of first test signal data by a preset strong classifier to obtain second test signal data of each chip operating environment parameter;

[0031] Specifically, noise is removed from each first test signal data, and background noise and irrelevant interference are removed through signal processing technology, such as digital filter or waveform analysis technology, to retain the true characteristics of the signal and obtain standard test signal data. Each standard test signal data is input into a preset strong classifier, which is composed of multiple convolutional threshold recurrent networks, and is a composite network structure that combines the characteristics of convolutional neural network (CNN) and recurrent neural network (RNN). The convolutional network part is responsible for performing convolution operations and signal data encoding conversion on the standard test signal data, which can not only extract feature information in the signal data, but also convert this information into a coding form that is easier to process and analyze. The convolution operation captures the local dependency features of the signal through the filter group, and the signal data encoding conversion converts these features into a new data representation. The signal encoding conversion data is modulated and analyzed by the threshold recurrent network part, and the signal pattern and change trend are identified and analyzed to obtain modulation analysis signal data. The time series analysis capability of the recurrent network is used to understand the changes of signal data over time, so as to extract deeper signal features and laws. Obtain the weight ratio data of multiple convolutional gated recurrent networks, and fuse the modulated analytical signal data generated by each network according to the weight ratio data to achieve the final integration of the signal data. The fusion process is based on the network weights, and the contribution of each network output will be adjusted according to its weight ratio to ensure that the result of signal fusion contains all important signal features while removing redundancy and noise. Finally, the second test signal data of each chip operating environment parameter is obtained.

[0032] Step 103: Perform Fourier transform on the second test signal data to obtain target chip test data of each chip working environment parameter, and perform data set classification on the target chip test data to obtain transmission rate data and timing delay data of each chip working environment parameter;

[0033] Specifically, the second test signal data is subjected to Fourier transform, and the target chip test data under each chip working environment parameter is obtained by converting the signal in the time domain to the frequency domain. The target chip test data is input into the preset data set classification model, and the signal data is further analyzed by machine learning technology. The data set classification model calculates the cluster center of the target chip test data by cluster analysis method, and identifies the main groups in the data, namely the first transmission rate cluster center and the first timing delay cluster center. These two cluster centers represent the typical characteristics of the transmission rate and timing delay in the signal data, respectively. The cluster center offset correction calculation is performed on the target chip test data and the first transmission rate cluster center and the first timing delay cluster center, respectively, to improve the classification accuracy and reliability. By comparing the difference between the target chip test data and the first transmission rate cluster center and the first timing delay cluster center, the position of the cluster center is adjusted, so as to obtain the second transmission rate cluster center and the second timing delay cluster center. The correction process ensures that the cluster center more accurately reflects the main characteristics in the data set. Based on the corrected second transmission rate cluster center and second timing delay cluster center, the target chip test data is finally classified into different data sets. The test data is divided into different categories according to the characteristics of transmission rate and timing delay, and the specific transmission rate data and timing delay data under each chip working environment parameters are obtained.

[0034] Step 104: construct a corresponding chip working condition transmission rate relationship curve according to the transmission rate data and multiple chip working condition environment parameters, and perform curve feature extraction and feature combination weighting on the chip working condition transmission rate relationship curve to obtain multiple chip working condition transmission rate features;

[0035] Specifically, the corresponding chip working condition transmission rate relationship curve is constructed according to the transmission rate data and the chip working condition environmental parameters, and the change trend of the transmission rate under different working conditions is revealed in a graphical way. Through the relationship curve, it is obtained how the transmission rate changes with the change of environmental parameters such as temperature, voltage, and frequency. The curve feature extraction is performed on the chip working condition transmission rate relationship curve, and the key features in the curve are identified and extracted, such as slope, area under the curve, peak value, etc. These initial working condition transmission rate characteristics reflect the basic law of the transmission rate changing with environmental parameters. By analyzing these curves, multiple initial working condition transmission rate characteristics describing the transmission rate change characteristics are obtained. The initial working condition transmission rate characteristics are subjected to probability distribution calculation and information entropy weight calculation to evaluate the uncertainty and information content of each feature. By calculating the probability distribution of each feature, the first feature probability distribution is obtained, and the information entropy weight is calculated by distribution to reflect the amount of information of each feature in describing the change of the transmission rate. The larger the information entropy weight, the richer the information contained in the feature, and the more important it is for understanding the change of the transmission rate. By performing standard deviation and mean calculation on the initial working condition transmission rate characteristics, and then calculating the coefficient of variation weight, the stability and range of change of each feature are evaluated. The coefficient of variation weight measures the relative degree of change of the feature by combining the standard deviation and the mean, giving a quantitative metric to the importance of the feature. The information entropy weight and the coefficient of variation weight are linearly combined to obtain the comprehensive weight of each feature, which takes into account both the amount of information of the feature and the stability of its change. The final weight of each initial working condition transmission rate feature is obtained by weighted averaging the weights, and these weighted features are further linearly combined to obtain a set of chip working condition transmission rate features that comprehensively consider the amount of information and stability.

[0036] Step 105: construct a corresponding chip operating condition timing delay relationship curve according to the timing delay data and multiple chip operating condition environment parameters, and perform curve feature extraction and feature combination weighting on the chip operating condition timing delay relationship curve to obtain multiple chip operating condition timing delay features;

[0037] Specifically, according to the timing delay data and different chip working environment parameters, the corresponding chip working condition timing delay relationship curve is constructed, and the data is visualized to reveal how the timing delay changes with the change of environmental parameters. The relationship curve can intuitively show the changing trend of the timing delay. By analyzing the shape, slope, peak value and other aspects of the curve, the curve feature extraction of the timing delay relationship curve is performed to obtain multiple initial working condition features that describe the change of the timing delay. The probability distribution operation is performed on the initial working condition timing delay feature, and then the probability distribution of each feature is calculated to evaluate the distribution characteristics of the feature. Based on the probability distribution, the information entropy weight of each feature is calculated. The calculation of the information entropy weight reflects the amount of information of each feature in describing the change of the timing delay. The standard deviation and mean operation are performed on the features to calculate the coefficient of variation weight of each feature. The coefficient of variation weight provides a quantitative evaluation of the discrete degree of the feature by comparing the distribution range and central tendency of the feature value. The information entropy weight and the coefficient of variation weight are linearly combined to obtain the comprehensive weight of each initial working condition timing delay feature. By comprehensively considering the amount of information and stability, a comprehensive weight is assigned to each feature. Based on the comprehensive weights, all initial operating condition timing delay features are weighted and linearly combined, and finally a set of features that comprehensively reflect the chip operating condition timing delay performance is obtained.

[0038] Step 106: Input the transmission rate characteristics of multiple chip working conditions and the timing delay characteristics of multiple chip working conditions into a preset multi-task attention mechanism chip performance analysis model to perform stacked chip signal simulation test performance analysis to obtain signal transmission performance analysis results and timing integrity analysis results.

[0039] Specifically, multiple chip working condition transmission rate features and multiple chip working condition timing delay features are input into a preset multi-task attention mechanism chip performance analysis model, wherein the multi-task attention mechanism chip performance analysis model includes: an input layer, an attention mechanism layer, a first chip performance analysis task network, and a second chip performance analysis task network. These features are encoded through the input layer and converted into a form that can be processed by the model, that is, an input transmission rate feature vector and an input timing delay feature vector. The input feature vector is weighted by the attention mechanism layer to identify which features are more critical to the analysis of chip performance, and then the influence of these features is strengthened by weighting, so that the model can automatically focus on the most important information, thereby improving the accuracy and efficiency of the analysis, thereby obtaining a first chip performance evaluation vector and a second chip performance evaluation vector. The evaluation vectors are processed by the first chip performance analysis task network and the second chip performance analysis task network respectively. The two task networks are responsible for different analysis tasks respectively. The first network uses a random forest algorithm to process the first chip performance evaluation vector to obtain an analysis result of the signal transmission performance; and the second network also uses a random forest algorithm to process the second chip performance evaluation vector and outputs an analysis result of the timing integrity. As an integrated learning method, random forest can provide highly accurate prediction results by constructing multiple decision trees and summarizing their prediction results. It is suitable for processing complex nonlinear relationships.

[0040] In the embodiment of the present application, the signal simulation test of the target stacked chip is realized through the signal simulation test platform by using the multi-layer structure information, wiring density information and various chip working environment parameters. This automated method can improve the test efficiency, reduce manual intervention and reduce the test cost. Through the comprehensive calculation of the signal simulation function, the environmental parameter influence function and the signal noise function, the signal characteristics under different chip working conditions can be accurately simulated, including signal amplitude, attenuation, frequency, phase, etc. This helps to obtain accurate test data. The first test signal data is fused using a strong classifier to reduce the impact of noise on the test results, improve the signal quality, and enhance the stability and reliability of the test data. Through Fourier transform and data set classification, the transmission rate data and timing delay data can be extracted from the second test signal data. These data can be used to analyze the signal transmission performance and timing integrity of the stacked chip. By constructing the chip working condition transmission rate relationship curve and the timing delay relationship curve, and performing feature extraction, combination and weighting, multiple chip working condition transmission rate features and timing delay features can be generated. These features can be input into the multi-task attention mechanism chip performance analysis model to perform stacked chip signal simulation test performance analysis, thereby improving the accuracy of the signal simulation test of the stacked chip.

[0041] In a specific embodiment, the process of executing step 101 may specifically include the following steps:

[0042] (1) Build a signal simulation test platform and obtain the multi-layer structure information and wiring density information of the target stacked chip;

[0043] (2) Setting a variety of chip operating environment parameters based on the multi-layer structure information and wiring density information, where each chip operating environment parameter includes: temperature, voltage, and frequency;

[0044] (3) Through the signal simulation test platform, signal simulation tests are performed on the target stacked chips according to various chip working environment parameters, and the ideal output signal data of each chip working environment parameter is calculated through the signal simulation function, where the signal simulation function is: , Indicates the signal amplitude under the chip working environment parameter P, represents the attenuation coefficient, Indicates the signal frequency under the chip working environment parameter P, Indicates the signal phase under the chip operating environment parameter P, represents the interaction term between chip operating environment parameter P and time t, Represents ideal output signal data;

[0045] (4) The environmental parameter influence coefficient of each chip operating condition environmental parameter is calculated through the environmental parameter influence function, where the environmental parameter influence function is: , represents the operating environment parameters of the i-th chip, represents the linear influence coefficient of the i-th chip operating environment parameter, represents the interaction coefficient between the i-th and j-th chip operating environment parameters, Indicates the environmental parameter influence coefficient;

[0046] (5) Calculate the noise signal data of each chip operating environment parameter through the signal noise function, where the signal noise function is: , It indicates the noise intensity under the chip working environment parameter P. represents the random noise function at time t, Indicates the noise frequency under the chip operating environment parameter P, represents the noise phase constant;

[0047] (6) Generate multiple first test signal data of each chip operating environment parameter according to the ideal output signal data, the environmental parameter influence coefficient and the noise signal data through the total signal output function, wherein the total signal output function is: , Represents the first test signal data.

[0048] Specifically, a highly precise signal simulation test platform is constructed to obtain the multi-layer structure information and wiring density information of the target stacked chip. Based on this information, various chip working environment parameters, such as temperature, voltage and frequency, are set to ensure the authenticity and complexity of the test environment, thereby simulating various situations that the chip may encounter in actual use. The ideal output signal data of the target stacked chip under different working environment parameters are calculated through the signal simulation function. The signal simulation function takes into account multiple factors such as signal amplitude, attenuation coefficient, signal frequency, signal phase, and the interaction term between the chip working environment parameters and time, so as to accurately simulate the ideal signal output of the chip under specific environmental conditions. For example, when the test platform simulates high temperature conditions, the signal simulation function can calculate the corresponding ideal output signal data according to the influence of temperature on the signal attenuation coefficient and signal phase, which helps to understand the change of chip performance under extreme temperature conditions. The environmental parameter influence coefficient of each chip working environment parameter is calculated through the environmental parameter influence function, and the noise signal data under each environmental parameter is calculated through the signal noise function. The environmental parameter influence function and the signal noise function work together to simulate the uncertainty factors in the actual environment, such as the noise generated by temperature and voltage fluctuations, and provide a more realistic test environment for signal simulation testing. The ideal output signal data, environmental parameter influence coefficient and noise signal data are integrated through the total signal output function to obtain the first test signal data under each chip working environment parameter. This process not only reflects the performance of the chip under ideal conditions, but also takes into account the comprehensive influence of environmental factors and random noise on the signal, so that the performance of the chip in practical applications can be more accurately evaluated. For example, suppose that at a certain voltage and frequency, the target stacked chip needs to work in a high temperature environment. The test platform first sets the corresponding high temperature environmental parameters according to the multi-layer structure information and wiring density information of the chip. The signal simulation function is used to calculate the ideal output signal data according to these environmental parameters, and the environmental parameter influence coefficient and noise signal data are calculated at the same time. If the attenuation coefficient increases under high temperature conditions, the signal amplitude decreases, and the noise signal becomes more significant due to voltage fluctuations. By analyzing these data, the signal transmission performance and timing integrity of the chip in a high temperature environment are evaluated, so that the chip design can be adjusted and optimized as necessary to ensure its reliability and performance in practical applications.

[0049] In a specific embodiment, the process of executing step 102 may specifically include the following steps:

[0050] (1) performing noise removal on each of the plurality of first test signal data to obtain standard test signal data for each of the first test signal data;

[0051] (2) inputting the standard test signal data of each first test signal data into a preset strong classifier, the strong classifier comprising: a plurality of convolutional threshold recurrent networks;

[0052] (3) performing convolution operations and signal data encoding conversion on the standard test signal data respectively through the convolution networks in the multiple convolutional threshold recurrent networks to obtain signal encoding conversion data of each standard test signal data;

[0053] (4) performing signal data modulation and analysis on the signal coding conversion data respectively through the gated recurrent networks in the plurality of convolutional gated recurrent networks, and obtaining the modulation and analysis signal data of each convolutional gated recurrent network;

[0054] (5) Obtain weight ratio data of multiple convolutional threshold recurrent networks, and perform signal data fusion on the modulation analysis signal data of each convolutional threshold recurrent network according to the weight ratio data to obtain second test signal data of each chip operating environment parameter.

[0055] Specifically, noise is removed from multiple first test signal data to obtain more accurate and clean standard test signal data. Through signal processing techniques such as waveform filtering, noise identification and removal algorithms, background noise and non-target signals in the test signal are removed to ensure the accuracy of subsequent analysis. For example, by using a digital filter, noise outside the frequency can be effectively removed to make the signal clearer. Each first test signal data is input into a preset strong classifier. The strong classifier is a complex network structure that combines a convolutional network and a threshold recurrent network, which is suitable for processing and analyzing signal data. The convolutional threshold recurrent network (ConvLSTM) can capture the spatial features (through the convolutional network part) and time series features (through the threshold recurrent network part) in the signal data. In the convolutional network part, the standard test signal data is converted into a form that is easier to analyze and process after convolution operation and signal data encoding conversion. The convolution operation extracts local features of the signal through multiple filters, and the signal data encoding conversion converts these features into a coded form. For example, the convolutional network can identify specific patterns or waveforms in the signal and encode the information into a vector form. In the gated recurrent network part, the signal data modulation and analysis are performed on the signal coding conversion data. The time series data is processed by the recurrent network, the signal is analyzed, and the pattern and characteristics of the signal changing over time are identified. The gated recurrent network can effectively process and memorize the information of the signal in the time dimension through its internal gating mechanism. Finally, by obtaining the weight ratio data of multiple convolutional gated recurrent networks and fusing the modulation and analysis signal data of each network according to the weight ratio data, the second test signal data of each chip working environment parameter is obtained. The fusion process takes into account the importance and contribution of different network outputs, and integrates this information by weighted averaging to obtain a comprehensive signal data representation. For example, if some networks capture signal characteristics more accurately under specific working conditions, then the outputs of these networks will obtain higher weights in the fusion process, thereby ensuring that the final signal data can accurately reflect the performance of the chip.

[0056] In a specific embodiment, the process of executing step 103 may specifically include the following steps:

[0057] (1) Performing Fourier transform on the second test signal data to obtain target chip test data for each chip operating environment parameter;

[0058] (2) inputting the target chip test data of each chip working environment parameter into a preset data set classification model, and calculating the cluster center of the target chip test data through the data set classification model to obtain a first transmission rate cluster center and a first timing delay cluster center;

[0059] (3) performing cluster center offset correction calculation on the target chip test data and the first transmission rate cluster center and the first timing delay cluster center respectively to obtain the second transmission rate cluster center and the second timing delay cluster center;

[0060] (4) Based on the second transmission rate cluster center and the second timing delay cluster center, the target chip test data is classified into data sets to obtain the transmission rate data and timing delay data of each chip working environment parameter.

[0061] Specifically, the second test signal data is Fourier transformed, the time domain signal is converted into a frequency domain signal, and the target chip test data under each chip working environment parameter is obtained. These data reflect the frequency domain response characteristics of the chip under different environments. For example, if a chip has a significant offset in its signal spectrum when working under high temperature conditions, this may indicate that temperature has a significant effect on the transmission characteristics of the chip. The frequency domain test data is input into the preset data set classification model for analysis. The clustering center of the transmission rate and timing delay is calculated by the clustering algorithm. The calculation of the clustering center is based on the pattern and trend in the data set, aiming to find the center point representing the characteristics of the entire data set. For example, through cluster analysis, it can be determined which transmission rate and timing delay values ​​best represent the average performance of the chip at a specific voltage and frequency. The clustering center obtained by the preliminary calculation may not be accurate enough due to the influence of sample deviation or outliers. Therefore, a clustering center offset correction calculation is performed, and the clustering center is adjusted by comparing the deviation between the original clustering center and the actual data point to obtain a more accurate second transmission rate clustering center and second timing delay clustering center. The correction process ensures the reliability and accuracy of the classification results. Based on the corrected second transmission rate cluster center and second timing delay cluster center, the target chip test data is finally classified into a data set to obtain specific transmission rate data and timing delay data for each chip working environment parameter. The test data is divided into different categories according to the characteristics of transmission rate and timing delay, so as to evaluate the performance of the chip under various environmental conditions in detail.

[0062] In a specific embodiment, the process of executing step 104 may specifically include the following steps:

[0063] (1) Construct the corresponding chip working condition transmission rate relationship curve based on the transmission rate data and various chip working condition environment parameters;

[0064] (2) Extracting the curve features of the chip working condition transmission rate relationship curve to obtain multiple initial working condition transmission rate features;

[0065] (3) performing probability distribution calculation on the transmission rate characteristics of the multiple initial working conditions to obtain the first characteristic probability distribution of the transmission rate characteristics of each initial working condition;

[0066] (4) calculating the first information entropy weight of the transmission rate feature of each initial working condition according to the first characteristic probability distribution;

[0067] (5) performing standard deviation and mean calculations on the transmission rate characteristics of multiple initial working conditions to obtain a first characteristic standard deviation and a first characteristic average value;

[0068] (6) calculating the first coefficient of variation weight of the transmission rate characteristic of each initial working condition according to the first characteristic standard deviation and the first characteristic average value;

[0069] (7) linearly combining the first information entropy weight and the first variation coefficient weight of each initial working condition transmission rate feature to obtain a first combination weight of each initial working condition transmission rate feature;

[0070] (8) weighting the transmission rate characteristics of the multiple initial working conditions according to the combined weight to obtain a first weight value of the transmission rate characteristic of each initial working condition;

[0071] (9) Linearly combine the multiple initial operating condition transmission rate characteristics and the corresponding first weights to obtain multiple chip operating condition transmission rate characteristics.

[0072] Specifically, the corresponding chip working condition transmission rate relationship curve is constructed according to the transmission rate data and various chip working condition environmental parameters. How the transmission rate changes with the change of different environmental parameters is intuitively shown in a graphical way. For example, the change curve of the chip transmission rate at different temperatures is drawn to intuitively show the impact of temperature increase on the transmission rate. The curve feature extraction of the chip working condition transmission rate relationship curve is performed, and the key information describing the law of transmission rate change is identified from the curve, including the slope of the curve, the extreme point, the area under the curve and other features. By calculating these features, the initial working condition transmission rate feature is obtained. The probability distribution operation is performed on the initial working condition transmission rate feature, the distribution characteristics of each feature are analyzed, and the change trend and stability of each feature under different conditions are evaluated. Based on the probability distribution, the information entropy weight of each feature is calculated, and the information entropy weight reflects the uncertainty and information amount of the feature when describing the change of the transmission rate. The higher the information entropy weight, the more important the feature is in evaluating chip performance. The standard deviation and mean of the features are calculated to evaluate their dispersion and central tendency. These statistics are the basis for calculating the coefficient of variation weights. The coefficient of variation weights reflect the degree of dispersion of the feature value relative to the mean and are an important indicator for measuring feature stability. Through this step, we can determine which features show greater stability under different working conditions. The information entropy weight and the coefficient of variation weight of each feature are linearly combined to obtain the comprehensive weight of each feature. The comprehensive weight takes into account the amount of information and stability of the feature, and can more accurately reflect the contribution of each feature to the chip performance evaluation. Based on the comprehensive weights, all initial working condition transmission rate features are weighted, and finally a set of key features that can fully reflect the transmission rate performance of the chip under different environmental conditions is obtained.

[0073] In a specific embodiment, the process of executing step 105 may specifically include the following steps:

[0074] (1) Construct the corresponding chip operating condition timing delay relationship curve based on the timing delay data and various chip operating condition environment parameters;

[0075] (2) Extracting the curve features of the chip operating condition timing delay relationship curve to obtain multiple initial operating condition timing delay features;

[0076] (3) performing probability distribution calculation on multiple initial operating condition timing delay characteristics to obtain a second characteristic probability distribution of each initial operating condition timing delay characteristic;

[0077] (4) calculating the second information entropy weight of each initial working condition timing delay feature according to the second feature probability distribution;

[0078] (5) Performing standard deviation and mean calculations on multiple initial operating condition timing delay characteristics to obtain a second characteristic standard deviation and a second characteristic average value;

[0079] (6) calculating the second coefficient of variation weight of each initial operating condition timing delay feature according to the second characteristic standard deviation and the second characteristic average value;

[0080] (7) Linearly combining the second information entropy weight and the second variation coefficient weight of each initial operating condition timing delay feature to obtain a second combination weight of each initial operating condition timing delay feature;

[0081] (8) weighting the multiple initial operating condition timing delay characteristics according to the combined weight to obtain a second weight value of each initial operating condition timing delay characteristic;

[0082] (9) Linearly combine the multiple initial operating condition timing delay characteristics and the corresponding second weights to obtain multiple chip operating condition timing delay characteristics.

[0083] Specifically, the corresponding chip working condition timing delay relationship curve is constructed according to the timing delay data and various chip working condition environmental parameters, and the relationship between the timing delay data and various chip working condition environmental parameters is visualized. For example, a curve is drawn to show how the timing delay changes as the temperature increases, so as to intuitively understand the impact of temperature change on chip timing delay. By extracting the curve features of the timing delay relationship curve, key information describing the law of timing delay change is obtained. The slope, extreme points, area under the curve, etc. of the curve are identified. Each feature reflects the changing characteristics of the timing delay under different working conditions. Through statistical analysis methods, such as calculating the frequency distribution or probability density function of the characteristic value, the probability distribution operation is performed on the initial working condition timing delay feature to evaluate the changing trend and stability of each feature under different conditions. Based on the probability distribution, the information entropy weight of each feature is calculated to evaluate the uncertainty and information content of the feature in describing the change of the timing delay. By calculating the standard deviation and mean of the features, their dispersion and central tendency are evaluated, and the coefficient of variation weight of each feature is calculated, which helps to identify features that show greater stability under different working conditions. The information entropy weight and the coefficient of variation weight of each feature are linearly combined to obtain the comprehensive weight of each feature. The comprehensive weight reflects the overall importance of the feature and takes into account the information content and stability of the feature. Based on the comprehensive weight, all initial working condition timing delay features are weighted, and then a set of key features that comprehensively reflect the timing delay performance of the chip under different environmental conditions are obtained through linear combination.

[0084] In a specific embodiment, the process of executing step 106 may specifically include the following steps:

[0085] (1) Inputting a plurality of chip operating condition transmission rate features and a plurality of chip operating condition timing delay features into a preset multi-task attention mechanism chip performance analysis model, wherein the multi-task attention mechanism chip performance analysis model comprises: an input layer, an attention mechanism layer, a first chip performance analysis task network, and a second chip performance analysis task network;

[0086] (2) Through the input layer, feature encoding is performed on the transmission rate features of multiple chip working conditions and the timing delay features of multiple chip working conditions respectively to obtain an input transmission rate feature vector and an input timing delay feature vector;

[0087] (3) Performing attention mechanism weighting on the input transmission rate feature vector and the input timing delay feature vector through the attention mechanism layer to obtain the first chip performance evaluation vector and the first chip performance evaluation vector;

[0088] (4) performing stacked chip signal simulation test performance analysis on the first chip performance evaluation vector through the first random forest network in the first chip performance analysis task network to obtain a signal transmission performance analysis result;

[0089] (5) Through the second random forest network in the second chip performance analysis task network, the stacked chip signal simulation test performance analysis is performed on the second chip performance evaluation vector to obtain the timing integrity analysis result.

[0090] Specifically, multiple chip working condition transmission rate features and multiple chip working condition timing delay features are input into a preset multi-task attention mechanism chip performance analysis model. The transmission rate features and timing delay features of the chip working conditions are effectively encoded through the input layer, and the high-dimensional features are converted into feature vector forms that the model can process. The role of the attention mechanism layer is to weight the feature vector so that the model can focus on the features that are more important to the current performance evaluation task. In this layer, each feature is assigned a weight, which is automatically learned through training data and reflects the importance of different features to the performance evaluation results. For example, if under certain specific working conditions, the change in transmission rate has a greater impact on chip performance than timing delay, then the transmission rate feature will obtain a higher weight under these working conditions. The signal transmission performance and timing integrity are evaluated through the first chip performance analysis task network and the second chip performance analysis task network. The two networks use the random forest algorithm, which is an integrated learning method that predicts and classifies by constructing multiple decision trees, can handle complex nonlinear relationships, and has good robustness and generalization capabilities. In the first task network, according to the transmission rate feature vector, the random forest model calculates the evaluation result of the signal transmission performance. Similarly, the second task network uses the timing delay feature vector to evaluate timing integrity. For example, suppose that under a certain operating condition, changes in temperature and voltage have different effects on the signal transmission performance and timing integrity of the stacked chip. The attention mechanism layer can recognize that when evaluating signal transmission performance, the impact of temperature changes may be more important than voltage. Therefore, when analyzing signal transmission performance, temperature-related features will receive higher weights. On the contrary, when evaluating timing integrity, changes in voltage may be a more critical factor. In this way, the multi-task attention mechanism chip performance analysis model can automatically adjust the focus for different performance evaluation tasks, thereby improving the accuracy and efficiency of the evaluation.

[0091] The signal simulation test method for stacked chips in the embodiment of the present application is described above. The signal simulation test device for stacked chips in the embodiment of the present application is described below. Figure 2 , an embodiment of the signal simulation test device for stacked chips in the embodiment of the present application includes:

[0092] The test module 201 is used to construct a signal simulation test platform and perform signal simulation tests on target stacked chips respectively through a variety of chip operating environment parameters to obtain a plurality of first test signal data for each chip operating environment parameter;

[0093] A fusion module 202 is used to perform signal fusion on a plurality of first test signal data through a preset strong classifier to obtain second test signal data of each chip operating environment parameter;

[0094] The transformation module 203 is used to perform Fourier transformation on the second test signal data to obtain target chip test data of each chip working environment parameter, and to classify the target chip test data into data sets to obtain transmission rate data and timing delay data of each chip working environment parameter;

[0095] The first extraction module 204 is used to construct a corresponding chip working condition transmission rate relationship curve according to the transmission rate data and multiple chip working condition environment parameters, and perform curve feature extraction and feature combination weighting on the chip working condition transmission rate relationship curve to obtain multiple chip working condition transmission rate features;

[0096] The second extraction module 205 is used to construct a corresponding chip operating condition timing delay relationship curve according to the timing delay data and multiple chip operating condition environment parameters, and perform curve feature extraction and feature combination weighting on the chip operating condition timing delay relationship curve to obtain multiple chip operating condition timing delay features;

[0097] The analysis module 206 is used to input the transmission rate characteristics of multiple chip working conditions and the timing delay characteristics of multiple chip working conditions into a preset multi-task attention mechanism chip performance analysis model to perform stacked chip signal simulation test performance analysis to obtain signal transmission performance analysis results and timing integrity analysis results.

[0098] Through the cooperation of the above components, the signal simulation test of the target stacked chip is realized through the signal simulation test platform by using the multi-layer structure information, wiring density information and various chip working environment parameters. This automated method can improve the test efficiency, reduce manual intervention and reduce the test cost. Through the comprehensive calculation of the signal simulation function, the environmental parameter influence function and the signal noise function, the signal characteristics under different chip working conditions can be accurately simulated, including signal amplitude, attenuation, frequency, phase, etc. This helps to obtain accurate test data. The first test signal data is fused by a strong classifier to reduce the impact of noise on the test results, improve the signal quality, and enhance the stability and reliability of the test data. Through Fourier transform and data set classification, the transmission rate data and timing delay data can be extracted from the second test signal data. These data can be used to analyze the signal transmission performance and timing integrity of the stacked chip. By constructing the chip working condition transmission rate relationship curve and the timing delay relationship curve, and performing feature extraction, combination and weighting, multiple chip working condition transmission rate features and timing delay features can be generated. These features can be input into the multi-task attention mechanism chip performance analysis model to perform stacked chip signal simulation test performance analysis, thereby improving the accuracy of the signal simulation test of the stacked chip.

[0099] The present application also provides a computer device, which includes a memory and a processor. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor executes the steps of the signal simulation test method for stacked chips in the above-mentioned embodiments.

[0100] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the signal simulation test method for stacked chips.

[0101] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0102] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0103] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A signal simulation test method for stacked chips, characterized in that: The signal simulation test method of the stacked chip comprises: A signal simulation test platform is constructed and a signal simulation test is performed on a target stacked chip respectively through a plurality of chip working environment parameters to obtain a plurality of first test signal data of each chip working environment parameter; specifically comprising: constructing a signal simulation test platform and obtaining multi-layer structure information and wiring density information of the target stacked chip; setting a plurality of chip working environment parameters based on the multi-layer structure information and the wiring density information, wherein each chip working environment parameter includes: temperature, voltage and frequency; performing a signal simulation test on the target stacked chip respectively according to the plurality of chip working environment parameters through the signal simulation test platform, and calculating the ideal output signal data of each chip working environment parameter through a signal simulation function, wherein the signal simulation function is: , Indicates the signal amplitude under the chip working environment parameter P, represents the attenuation coefficient, Indicates the signal frequency under the chip working environment parameter P, Indicates the signal phase under the chip operating environment parameter P, represents the interaction term between chip operating environment parameter P and time t, represents the ideal output signal data; the environmental parameter influence coefficient of each chip working condition environmental parameter is calculated by the environmental parameter influence function, wherein the environmental parameter influence function is: , represents the operating environment parameters of the i-th chip, represents the linear influence coefficient of the i-th chip operating environment parameter, represents the interaction coefficient between the i-th and j-th chip operating environment parameters, represents the environmental parameter influence coefficient; the noise signal data of each chip working condition environmental parameter is calculated by the signal noise function, wherein the signal noise function is: , It indicates the noise intensity under the chip working environment parameter P. represents the random noise function at time t, Indicates the noise frequency under the chip operating environment parameter P, Represents a noise phase constant; and generates a plurality of first test signal data of each chip operating environment parameter according to the ideal output signal data, the environmental parameter influence coefficient and the noise signal data through a total signal output function, wherein the total signal output function is: , represents first test signal data; The plurality of first test signal data are subjected to signal fusion by a preset strong classifier to obtain second test signal data for each chip operating environment parameter; specifically comprising: removing noise from the plurality of first test signal data respectively to obtain standard test signal data for each first test signal data; inputting the standard test signal data for each first test signal data into a preset strong classifier, the strong classifier comprising: a plurality of convolutional threshold recurrent networks; performing convolution operation and signal data encoding conversion on the standard test signal data respectively through the convolutional networks in the plurality of convolutional threshold recurrent networks to obtain signal encoding conversion data for each standard test signal data; performing signal data modulation and analysis on the signal encoding conversion data respectively through the threshold recurrent networks in the plurality of convolutional threshold recurrent networks to obtain modulation analysis signal data for each convolutional threshold recurrent network; obtaining weight ratio data for the plurality of convolutional threshold recurrent networks, and performing signal data fusion on the modulation analysis signal data of each convolutional threshold recurrent network according to the weight ratio data to obtain second test signal data for each chip operating environment parameter; Performing Fourier transform on the second test signal data to obtain target chip test data for each chip working environment parameter, and performing data set classification on the target chip test data to obtain transmission rate data and timing delay data for each chip working environment parameter; specifically comprising: performing Fourier transform on the second test signal data to obtain target chip test data for each chip working environment parameter; inputting the target chip test data for each chip working environment parameter into a preset data set classification model, and performing cluster center calculation on the target chip test data through the data set classification model to obtain a first transmission rate cluster center and a first timing delay cluster center; performing cluster center offset correction calculation on the target chip test data and the first transmission rate cluster center and the first timing delay cluster center respectively to obtain a second transmission rate cluster center and a second timing delay cluster center; performing data set classification on the target chip test data based on the second transmission rate cluster center and the second timing delay cluster center to obtain transmission rate data and timing delay data for each chip working environment parameter; A corresponding chip operating condition transmission rate relationship curve is constructed according to the transmission rate data and the various chip operating condition environment parameters, and curve feature extraction and feature combination weighting are performed on the chip operating condition transmission rate relationship curve to obtain multiple chip operating condition transmission rate features; specifically comprising: constructing a corresponding chip operating condition transmission rate relationship curve according to the transmission rate data and the various chip operating condition environment parameters; extracting curve features from the chip operating condition transmission rate relationship curve to obtain multiple initial operating condition transmission rate features; performing probability distribution calculation on the multiple initial operating condition transmission rate features to obtain a first feature probability distribution of each initial operating condition transmission rate feature; and calculating the probability distribution of each initial operating condition transmission rate feature according to the first feature probability distribution. The first information entropy weight; performing standard deviation and mean operations on the multiple initial working condition transmission rate characteristics to obtain a first characteristic standard deviation and a first characteristic average value; calculating a first coefficient of variation weight of each initial working condition transmission rate characteristic according to the first characteristic standard deviation and the first characteristic average value; linearly combining the first information entropy weight and the first coefficient of variation weight of each initial working condition transmission rate characteristic to obtain a first combined weight of each initial working condition transmission rate characteristic; weighting the multiple initial working condition transmission rate characteristics according to the combined weight to obtain a first weight value of each initial working condition transmission rate characteristic; linearly combining the multiple initial working condition transmission rate characteristics and the corresponding first weight values ​​to obtain multiple chip working condition transmission rate characteristics; According to the timing delay data and the various chip operating environment parameters, a corresponding chip operating condition timing delay relationship curve is constructed, and curve features are extracted and feature combination weights are assigned to the chip operating condition timing delay relationship curve to obtain multiple chip operating condition timing delay features; specifically including: according to the timing delay data and the various chip operating environment parameters, a corresponding chip operating condition timing delay relationship curve is constructed; curve features are extracted from the chip operating condition timing delay relationship curve to obtain multiple initial operating condition timing delay features; probability distribution calculation is performed on the multiple initial operating condition timing delay features to obtain a second feature probability distribution of each initial operating condition timing delay feature; and the probability distribution of each initial operating condition timing delay feature is calculated according to the second feature probability distribution. A second information entropy weight; performing standard deviation and mean operations on the multiple initial operating condition timing delay characteristics to obtain a second feature standard deviation and a second feature average value; calculating a second coefficient of variation weight of each initial operating condition timing delay characteristic based on the second feature standard deviation and the second feature average value; linearly combining the second information entropy weight and the second coefficient of variation weight of each initial operating condition timing delay characteristic to obtain a second combined weight of each initial operating condition timing delay characteristic; weighting the multiple initial operating condition timing delay characteristics according to the combined weight to obtain a second weight value of each initial operating condition timing delay characteristic; linearly combining the multiple initial operating condition timing delay characteristics and the corresponding second weight values ​​to obtain multiple chip operating condition timing delay characteristics; The transmission rate characteristics of the multiple chip operating conditions and the timing delay characteristics of the multiple chip operating conditions are input into a preset multi-task attention mechanism chip performance analysis model to perform stacked chip signal simulation test performance analysis to obtain signal transmission performance analysis results and timing integrity analysis results.

2. The signal simulation test method for stacked chips according to claim 1, characterized in that: The inputting the plurality of chip operating condition transmission rate characteristics and the plurality of chip operating condition timing delay characteristics into a preset multi-task attention mechanism chip performance analysis model to perform stacked chip signal simulation test performance analysis, and obtaining signal transmission performance analysis results and timing integrity analysis results, including: Input the plurality of chip operating condition transmission rate characteristics and the plurality of chip operating condition timing delay characteristics into a preset multi-task attention mechanism chip performance analysis model, wherein the multi-task attention mechanism chip performance analysis model comprises: an input layer, an attention mechanism layer, a first chip performance analysis task network, and a second chip performance analysis task network; Feature encoding is performed on the plurality of chip operating condition transmission rate features and the plurality of chip operating condition timing delay features respectively through the input layer to obtain an input transmission rate feature vector and an input timing delay feature vector; Performing attention mechanism weighting on the input transmission rate feature vector and the input timing delay feature vector respectively through the attention mechanism layer to obtain a first chip performance evaluation vector and a second chip performance evaluation vector; Performing stacked chip signal simulation test performance analysis on the first chip performance evaluation vector through the first random forest network in the first chip performance analysis task network to obtain a signal transmission performance analysis result; Through the second random forest network in the second chip performance analysis task network, the second chip performance evaluation vector is subjected to stacked chip signal simulation test performance analysis to obtain a timing integrity analysis result.

3. A signal simulation test device for stacked chips, characterized in that: The signal simulation test device for stacked chips comprises: The test module is used to construct a signal simulation test platform and perform signal simulation tests on target stacked chips respectively through multiple chip working environment parameters to obtain multiple first test signal data of each chip working environment parameter; specifically includes: constructing a signal simulation test platform, and obtaining multi-layer structure information and wiring density information of the target stacked chip; setting multiple chip working environment parameters based on the multi-layer structure information and the wiring density information, wherein each chip working environment parameter includes: temperature, voltage and frequency; through the signal simulation test platform, performing signal simulation tests on the target stacked chip respectively according to the multiple chip working environment parameters, and calculating the ideal output signal data of each chip working environment parameter through a signal simulation function, wherein the signal simulation function is: , Indicates the signal amplitude under the chip working environment parameter P, represents the attenuation coefficient, Indicates the signal frequency under the chip working environment parameter P, Indicates the signal phase under the chip operating environment parameter P, represents the interaction term between chip operating environment parameter P and time t, represents the ideal output signal data; the environmental parameter influence coefficient of each chip working condition environmental parameter is calculated by the environmental parameter influence function, wherein the environmental parameter influence function is: , represents the operating environment parameters of the i-th chip, represents the linear influence coefficient of the i-th chip operating environment parameter, represents the interaction coefficient between the i-th and j-th chip operating environment parameters, represents the environmental parameter influence coefficient; the noise signal data of each chip working condition environmental parameter is calculated by the signal noise function, wherein the signal noise function is: , It indicates the noise intensity under the chip working environment parameter P. represents the random noise function at time t, Indicates the noise frequency under the chip operating environment parameter P, Represents a noise phase constant; and generates a plurality of first test signal data of each chip operating environment parameter according to the ideal output signal data, the environmental parameter influence coefficient and the noise signal data through a total signal output function, wherein the total signal output function is: , represents first test signal data; A fusion module is used to perform signal fusion on the multiple first test signal data through a preset strong classifier to obtain second test signal data for each chip operating environment parameter; specifically comprising: removing noise from the multiple first test signal data respectively to obtain standard test signal data for each first test signal data; inputting the standard test signal data of each first test signal data into a preset strong classifier, the strong classifier comprising: multiple convolutional threshold recurrent networks; performing convolution operations and signal data encoding conversion on the standard test signal data respectively through the convolutional networks in the multiple convolutional threshold recurrent networks to obtain signal encoding conversion data for each standard test signal data; performing signal data modulation and analysis on the signal encoding conversion data respectively through the threshold recurrent networks in the multiple convolutional threshold recurrent networks to obtain modulation analysis signal data of each convolutional threshold recurrent network; obtaining weight ratio data of the multiple convolutional threshold recurrent networks, and performing signal data fusion on the modulation analysis signal data of each convolutional threshold recurrent network according to the weight ratio data to obtain second test signal data for each chip operating environment parameter; A transformation module is used to perform Fourier transformation on the second test signal data to obtain target chip test data for each chip working environment parameter, and to perform data set classification on the target chip test data to obtain transmission rate data and timing delay data for each chip working environment parameter; specifically comprising: performing Fourier transformation on the second test signal data to obtain target chip test data for each chip working environment parameter; inputting the target chip test data for each chip working environment parameter into a preset data set classification model, and performing cluster center calculation on the target chip test data through the data set classification model to obtain a first transmission rate cluster center and a first timing delay cluster center; performing cluster center offset correction calculation on the target chip test data and the first transmission rate cluster center and the first timing delay cluster center respectively to obtain a second transmission rate cluster center and a second timing delay cluster center; based on the second transmission rate cluster center and the second timing delay cluster center, performing data set classification on the target chip test data to obtain transmission rate data and timing delay data for each chip working environment parameter; The first extraction module is used to construct a corresponding chip operating condition transmission rate relationship curve according to the transmission rate data and the multiple chip operating condition environment parameters, and perform curve feature extraction and feature combination weighting on the chip operating condition transmission rate relationship curve to obtain multiple chip operating condition transmission rate characteristics; specifically comprising: constructing a corresponding chip operating condition transmission rate relationship curve according to the transmission rate data and the multiple chip operating condition environment parameters; performing curve feature extraction on the chip operating condition transmission rate relationship curve to obtain multiple initial operating condition transmission rate characteristics; performing probability distribution calculation on the multiple initial operating condition transmission rate characteristics to obtain a first feature probability distribution of each initial operating condition transmission rate characteristic; and calculating each initial operating condition transmission rate characteristic according to the first feature probability distribution. The first information entropy weight of the rate characteristic; perform standard deviation and mean operations on the multiple initial working condition transmission rate characteristics to obtain a first characteristic standard deviation and a first characteristic average value; calculate the first coefficient of variation weight of each initial working condition transmission rate characteristic according to the first characteristic standard deviation and the first characteristic average value; linearly combine the first information entropy weight and the first coefficient of variation weight of each initial working condition transmission rate characteristic to obtain a first combined weight of each initial working condition transmission rate characteristic; weight the multiple initial working condition transmission rate characteristics according to the combined weight to obtain a first weight value of each initial working condition transmission rate characteristic; linearly combine the multiple initial working condition transmission rate characteristics and the corresponding first weight values ​​to obtain multiple chip working condition transmission rate characteristics; The second extraction module is used to construct a corresponding chip operating condition timing delay relationship curve according to the timing delay data and the multiple chip operating condition environmental parameters, and perform curve feature extraction and feature combination weighting on the chip operating condition timing delay relationship curve to obtain multiple chip operating condition timing delay features; specifically includes: constructing a corresponding chip operating condition timing delay relationship curve according to the timing delay data and the multiple chip operating condition environmental parameters; performing curve feature extraction on the chip operating condition timing delay relationship curve to obtain multiple initial operating condition timing delay features; performing probability distribution calculation on the multiple initial operating condition timing delay features to obtain a second feature probability distribution of each initial operating condition timing delay feature; calculating each initial operating condition timing delay feature according to the second feature probability distribution. A second information entropy weight of the delay feature; performing standard deviation and mean operations on the multiple initial operating condition timing delay features to obtain a second feature standard deviation and a second feature average value; calculating a second coefficient of variation weight of each initial operating condition timing delay feature according to the second feature standard deviation and the second feature average value; linearly combining the second information entropy weight and the second coefficient of variation weight of each initial operating condition timing delay feature to obtain a second combined weight of each initial operating condition timing delay feature; weighting the multiple initial operating condition timing delay features according to the combined weight to obtain a second weight value of each initial operating condition timing delay feature; linearly combining the multiple initial operating condition timing delay features and the corresponding second weight values ​​to obtain multiple chip operating condition timing delay features; The analysis module is used to input the transmission rate characteristics of the multiple chip working conditions and the timing delay characteristics of the multiple chip working conditions into a preset multi-task attention mechanism chip performance analysis model to perform stacked chip signal simulation test performance analysis to obtain signal transmission performance analysis results and timing integrity analysis results.

4. A computer device, characterized in that: The computer device comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the computer device to execute the signal simulation test method for stacked chips as claimed in claim 1 or 2.

5. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the signal simulation test method for stacked chips as claimed in claim 1 or 2 is implemented.