Crosstalk delay prediction method, device, terminal device and readable storage medium

Abstract

The present application discloses a crosstalk delay prediction method, apparatus, terminal device and readable storage medium. The method uses a pre-trained crosstalk signal prediction model to quickly, accurately and stably predict the i-th crosstalk signal value of the i-th interference site of the predicted disturbed network based on the self-attribute parameter set of the predicted disturbed network and the site attribute parameter set of the i-th interference site. In addition, the pre-trained crosstalk delay prediction model can quickly, accurately and stably predict the crosstalk delay of the predicted disturbed network based on all crosstalk signal values ​​and the crosstalk key parameter set of the predicted disturbed network. The crosstalk delay prediction process of the present application is simpler and more efficient, which can greatly improve the signal integrity analysis efficiency in STA and effectively reduce the computing cost.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor technology, and in particular to a crosstalk delay prediction method, device, terminal equipment and readable storage medium. Background Art

[0002] Static Timing Analysis (STA) is an essential verification step throughout the integrated circuit design process. Stringent timing requirements for circuit paths force designers to frequently invoke STA tools to avoid timing conflicts. Currently, STA's efficiency is primarily limited by two factors: the complexity of the algorithm itself and the inevitable crosstalk effect.

[0003] While the algorithm is stable, integrated circuit technology is constantly being explored. The number of metal layers is increasing, the connections are narrowing, and there are a large number of interconnections between devices, which leads to a coupling effect in the circuit. When this coupling effect reaches a certain level, the glitches and crosstalk noise signals it produces will significantly change the waveform of the original signal, causing changes in signal delay and even affecting the logical function of the circuit.

[0004] The concept of signal integrity (SI) has been introduced in existing technologies to analyze these increasingly serious crosstalk issues. For example, to obtain relatively accurate results, the industry generally relies on current source models, simulating the waveforms of circuit signals and then, through multiple iterations, obtaining a converged result to determine the crosstalk delay caused by crosstalk. However, this simulation consumes a large amount of computing resources and is quite time-consuming. Summary of the Invention

[0005] In view of the above problems, the present application proposes a crosstalk delay prediction method, apparatus, terminal device and readable storage medium.

[0006] In a first aspect, an embodiment of the present application provides a crosstalk delay prediction method, the method comprising:

[0007] Obtain a set of attribute parameters of the network to be predicted, a set of crosstalk key parameters of the network to be predicted, and a set of site attribute parameters of the i-th interference site, where the i-th interference site is a point where the interferer corresponding to the network to be predicted interferes with the network to be predicted, 1≤i≤I, and I is the total number of interference sites that affect the network to be predicted;

[0008] Using a pre-trained crosstalk signal prediction model, based on the self attribute parameter set of the to-be-predicted disturbed network and the site attribute parameter set of the i-th interfering site, predict the i-th crosstalk signal value of the i-th interfering site of the to-be-predicted disturbed network;

[0009] The crosstalk delay of the to-be-predicted victim network is predicted using a pre-trained crosstalk delay prediction model based on I crosstalk signal values ​​and a set of crosstalk key parameters of the to-be-predicted victim network.

[0010] The crosstalk delay prediction method described in an embodiment of the present application, wherein the crosstalk delay prediction model is used to predict the crosstalk delay of the victim network to be predicted based on I crosstalk signal values ​​and a set of crosstalk key parameters of the victim network to be predicted, includes:

[0011] Arrange the I crosstalk signal values ​​in descending order to obtain a crosstalk signal sequence;

[0012] Determining whether I is greater than a predetermined number;

[0013] When I is greater than or equal to the predetermined number, a feature vector is determined based on a predetermined number of crosstalk signal values ​​that are ranked first in the crosstalk signal sequence and the crosstalk key parameter set; or when I is less than the predetermined number, a predetermined supplementary value is added after the crosstalk signal sequence so that the crosstalk signal sequence includes the predetermined number of crosstalk signal values, and a feature vector is determined based on the predetermined number of crosstalk signal values ​​and the crosstalk key parameter set;

[0014] The feature vector is input into a pre-trained crosstalk delay prediction model, so as to predict the crosstalk delay of the to-be-predicted disturbed network through the pre-trained crosstalk delay prediction model.

[0015] In the crosstalk delay prediction method described in the embodiment of the present application, the set of attribute parameters of the network to be predicted includes: at least one of the capacitance value of the accumulated capacitance of the network to be predicted, the transition time of the switching signal of the network to be predicted, and the capacitance value of the total coupling capacitance of the network to be predicted.

[0016] The crosstalk delay prediction method described in the embodiment of the present application obtains the site attribute parameter set of the i-th interference site, including:

[0017] Determine the first resistance value of the driving equivalent resistance of the i-th crosstalk model corresponding to the i-th interference site, the second resistance value of the upstream resistance of the i-th interference site, the third resistance value of the downstream resistance of the i-th interference site, the capacitance value of the coupling capacitance between the victim network to be predicted and the interferer in the i-th crosstalk model, the capacitance value of the load capacitance at the output of the victim network to be predicted in the i-th crosstalk model, the transition time of the switching signal of the interferer in the i-th crosstalk model, the capacitance value of the first connection capacitance in the i-th crosstalk model, and the capacitance value of the second connection capacitance in the i-th crosstalk model;

[0018] Determining a first time parameter of the i-th crosstalk model based on the first resistance, the second resistance, and the capacitance value of the coupling capacitor;

[0019] determining a second time parameter of the i-th crosstalk model based on the first resistance, the second resistance, the third resistance, the capacitance value of the coupling capacitor, the capacitance value of the load capacitor, the capacitance value of the first connection capacitor, and the capacitance value of the second connection capacitor;

[0020] Determine the site attribute parameter set of the i-th interference site based on the first resistance, the second resistance, the capacitance value of the coupling capacitor, the capacitance value of the load capacitor, the first time parameter, the second time parameter and the transition time of the interferer switching signal.

[0021] The crosstalk delay prediction method described in the embodiment of the present application uses the following formula to determine the first time parameter of the i-th crosstalk model:

[0022] t xi =(R di +R si )*C xi

[0023] t xi represents the first time parameter of the i-th crosstalk model, R di represents the first resistance value of the driving equivalent resistance of the i-th crosstalk model, R si The second resistance value of the upstream resistor of the i-th interference point, C xi represents the capacitance value of the coupling capacitance between the victim network to be predicted and the interferer in the i-th crosstalk model.

[0024] The crosstalk delay prediction method described in the embodiment of the present application uses the following formula to determine the second time parameter of the i-th crosstalk model, including:

[0025] t vi =(R di +R si )*(C xi +C 2i +C Li )+(R ei *C Li +R di *C 1i )

[0026] t vi represents the second time parameter of the i-th crosstalk model, R di represents the first resistance value of the driving equivalent resistance of the i-th crosstalk model, R siThe second resistance value of the upstream resistor of the i-th interference point, C xi represents the capacitance value of the coupling capacitance between the victim network to be predicted and the interferer in the i-th crosstalk model, C Li represents the capacitance value of the load capacitance at the output of the disturbed network to be predicted in the i-th crosstalk model, C 1i represents the capacitance value of the first connection capacitor in the i-th crosstalk model, C 2i represents the capacitance value of the second connection capacitor in the i-th crosstalk model.

[0027] The crosstalk delay prediction method described in an embodiment of the present application, the crosstalk key parameter set includes the transition time of the switching signal at the driving pin connected to the predicted disturbed network, the transition time of the switching signal at the input pin of the load unit connected to the output end of the predicted disturbed network, the transition time of the switching signal at the output pin of the load unit, the non-signal integrity mode delay at the driving pin, the non-signal integrity mode delay at the input pin, the non-signal integrity mode delay at the output pin and at least one of the voltage level of the load unit.

[0028] In a second aspect, the crosstalk delay prediction method described in an embodiment of the present application further includes: pre-acquiring a first training data set, and iteratively training the crosstalk signal prediction model using the first training data set until a loss function corresponding to the crosstalk signal prediction model converges;

[0029] The pre-acquisition of the first training data set includes:

[0030] Get the attribute parameter set of the jth disturbed network used for training, 1≤j≤J, J is the total number of disturbed networks used for training;

[0031] Obtaining a set of site attribute parameters associated with an interference site in the j-th disturbed network for training;

[0032] Calculating the product of the peak voltage of the signal waveform output by the crosstalk model corresponding to the one interference site and the time period corresponding to the signal waveform;

[0033] The jth signal training sample in the first training data set is determined according to the jth set of self-attribute parameters of the disturbed network used for training and the set of site attribute parameters corresponding to the jth disturbed network used for training, and the label corresponding to the jth signal training sample is the product.

[0034] In a third aspect, the crosstalk delay prediction method described in an embodiment of the present application further includes: pre-acquiring a second training data set, and iteratively training the crosstalk delay prediction model using the second training data set until the loss function corresponding to the crosstalk delay prediction model converges;

[0035] The pre-acquisition of the second training data set includes:

[0036] Obtaining a crosstalk key parameter set of the jth victim network for training and a signal integrity mode delay of the jth victim network for training;

[0037] Predicting each crosstalk signal value of the jth victim network for training using the pre-trained crosstalk signal prediction model based on the jth victim network's own attribute parameter set for training and the site attribute parameter set of each interfering site;

[0038] The jth delay training sample in the second training data set is determined according to the crosstalk key parameter set and all crosstalk signal values ​​of the jth victim network for training, and the label corresponding to the jth delay training sample is the signal integrity mode delay.

[0039] In a fourth aspect, an embodiment of the present application further provides a crosstalk delay prediction device, the device comprising:

[0040] A parameter acquisition module is used to obtain a set of attribute parameters of the network to be predicted, a set of crosstalk key parameters of the network to be predicted, and a set of site attribute parameters of the i-th interference site, where the i-th interference site is a point where the interferer corresponding to the network to be predicted interferes with the network to be predicted, 1≤i≤I, and I is the total number of interference sites that affect the network to be predicted;

[0041] A signal prediction module is used to predict the i-th crosstalk signal value of the i-th interference site of the to-be-predicted disturbed network based on the set of its own attribute parameters and the set of site attribute parameters of the i-th interference site using a pre-trained crosstalk signal prediction model;

[0042] The delay prediction module is used to predict the crosstalk delay of the to-be-predicted victim network based on I crosstalk signal values ​​and the crosstalk key parameter set of the to-be-predicted victim network using a pre-trained crosstalk delay prediction model.

[0043] In a fifth aspect, an embodiment of the present application further proposes a terminal device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program runs on the processor, the crosstalk delay prediction method described in the embodiment of the present application is executed.

[0044] In a sixth aspect, an embodiment of the present application further proposes a readable storage medium storing a computer program, which executes the crosstalk delay prediction method described in the embodiment of the present application when running on a processor.

[0045] The present application utilizes a pre-trained crosstalk signal prediction model to quickly, accurately, and stably predict the i-th crosstalk signal value of the i-th interference site of the to-be-predicted disturbed network based on the set of its own attribute parameters and the set of site attribute parameters of the i-th interference site. Furthermore, the pre-trained crosstalk delay prediction model can quickly, accurately, and stably predict the crosstalk delay of the to-be-predicted disturbed network based on all crosstalk signal values ​​and the set of crosstalk key parameters of the to-be-predicted disturbed network. The crosstalk delay prediction process of the present application is simpler and more efficient, which can greatly improve the efficiency of signal integrity analysis in STA and effectively reduce computing costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope of protection of the present invention. In each of the drawings, similar components are numbered similarly.

[0047] Figure 1 A schematic diagram showing a case where a single interference site exists between a single victim and a single interferer, as proposed in an embodiment of the present application;

[0048] Figure 2 A schematic diagram showing a case where a victim and a single interferer have multiple interference sites is provided in an embodiment of the present application;

[0049] Figure 3 A schematic diagram showing a case where a victim and multiple interferers have multiple interference sites, as proposed in an embodiment of the present application;

[0050] Figure 4 A schematic flow chart of a crosstalk delay prediction method proposed in an embodiment of the present application is shown;

[0051] Figure 5 A schematic structural diagram of a crosstalk model proposed in an embodiment of the present application is shown;

[0052] Figure 6 A schematic structural diagram of an analysis model proposed in an embodiment of the present application is shown;

[0053] Figure 7 A schematic diagram of a process for predicting crosstalk delay in a crosstalk delay prediction method proposed in an embodiment of the present application is shown;

[0054] Figure 8 A schematic flow chart of a method for pre-training a crosstalk signal prediction model proposed in an embodiment of the present application is shown;

[0055] Figure 9 A schematic diagram of a flow chart of a method for pre-training a crosstalk delay prediction model proposed in an embodiment of the present application is shown;

[0056] Figure 10 A schematic structural diagram of a crosstalk delay prediction device proposed in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0058] The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the figures is not intended to limit the scope of the claimed invention, but rather merely represents selected embodiments of the present invention. All other embodiments derived by those skilled in the art based on the embodiments of the present invention without inventive effort are intended to be within the scope of protection of the present invention.

[0059] Hereinafter, the terms "including", "having" and their cognates, which may be used in various embodiments of the present invention, are intended only to indicate specific features, numbers, steps, operations, elements, components or combinations of the foregoing items, and should not be understood as first excluding the existence of one or more other features, numbers, steps, operations, elements, components or combinations of the foregoing items or the possibility of adding one or more features, numbers, steps, operations, elements, components or combinations of the foregoing items.

[0060] Furthermore, the terms “first,” “second,” “third,” etc., are merely used for distinguishing descriptions and are not to be understood as indicating or implying relative importance.

[0061] Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by those skilled in the art to which the various embodiments of the present invention pertain. The terms (such as those defined in generally used dictionaries) will be interpreted as having the same meaning as in the context of the relevant technical field and will not be interpreted as having an idealized meaning or an overly formal meaning unless clearly defined in the various embodiments of the present invention.

[0062] For ease of analysis, in the present embodiment, the mutual coupling between a section of continuous traces is abstracted as an interference site and used as the source of the crosstalk noise signal. It is understandable that the interference site is not a real connection point, but only a site that can interfere with the signal.

[0063] In deep submicron circuit design, the routing design of interconnect circuits is very complex, and the following conditions are common:

[0064] The same disturbed individual and a single disturber have a single interference site;

[0065] The same disturbed individual has multiple interference sites with a single disturber;

[0066] The same victim has multiple interference sites with multiple interferers.

[0067] For example, see Figure 1 In an embodiment of the present application, a situation is given where the same victim and a single interferer have a single interference site, that is, if the connection between pin 1 of the first unit and pin 3 of the second unit is regarded as a victim, then the connection between pin 2 of the first unit and pin 2 of the second unit is an interferer, and a is regarded as an interference site. The victim, interferer and interference site in this example constitute a victim network.

[0068] For example, see Figure 2 In an embodiment of the present application, a situation is given where the same victim and a single interferer have multiple interference sites, that is, if the connection between pin 1 of the first unit and pin 3 of the second unit is regarded as a victim, then the connection between pin 2 of the first unit and pin 4 of the second unit is an interferer, and a and b are two interference sites. In this example, the victim, the interferer and the two interference sites constitute a victim network.

[0069] For example, see Figure 3 In an embodiment of the present application, a case is given in which the same victim and multiple interferers have multiple interference sites. That is, if the connection between pin 1 of the first unit and pin 3 of the second unit is taken as a victim, then the connection between pin 2 of the first unit and pin 2 of the second unit is the first interferer, and the connection between pin 4 of the first unit and pin 1 of the second unit is the second interferer. a is the first interference site generated by the first interferer on the victim network, and c is the second interference site generated by the second interferer on the victim. The victim, two interferers and two interference sites in this example constitute a victim network.

[0070] The technical solution of this application will be explained in detail with specific embodiments below.

[0071] Example 1

[0072] See Figure 4 In an embodiment of the present application, a crosstalk delay prediction method is proposed, which includes the following steps S100 to S300:

[0073] Step S100: Obtain a set of attribute parameters of the network to be predicted, a set of crosstalk key parameters of the network to be predicted, and a set of site attribute parameters of the i-th interference site, wherein the i-th interference site is a point where the interferer corresponding to the network to be predicted interferes with the network to be predicted, 1≤i≤I, and I is the total number of interference sites that affect the network to be predicted.

[0074] The set of attribute parameters of the victim network to be predicted and the set of site attribute parameters of the i-th interfering site are used to predict the i-th crosstalk signal value of the i-th interfering site of the victim network to be predicted. The set of crosstalk key parameters of the victim network to be predicted is used to predict the crosstalk delay of the victim network to be predicted.

[0075] It has been verified that the capacitance value C of the cumulative capacitance of the disturbed network to be predicted ty , the transition time t of the switching signal of the to-be-predicted disturbed network vic,ry and the capacitance value C of the total coupling capacitance of the disturbed network to be predicted c,ty All of them will affect the crosstalk delay of the disturbed network to be predicted. Therefore, the set of self-attribute parameters of the disturbed network to be predicted includes at least one of the capacitance value of the cumulative capacitance of the disturbed network to be predicted, the transition time of the switching signal of the disturbed network to be predicted, and the capacitance value of the total coupling capacitance of the disturbed network to be predicted. It can be understood that, preferably, when the set of self-attribute parameters of the disturbed network to be predicted includes the capacitance value of the cumulative capacitance of the disturbed network to be predicted, the transition time of the switching signal of the disturbed network to be predicted, and the capacitance value of the total coupling capacitance of the disturbed network to be predicted, the crosstalk signal prediction model has the best prediction effect on the crosstalk signal of the disturbed network to be predicted and the highest prediction accuracy.

[0076] For further information, see Figure 5 In this embodiment, in order to determine the site attribute parameter set of each interference site of the disturbed network to be predicted, a crosstalk model is introduced. Figure 5 The crosstalk model in is a crosstalk model that includes an interference site o. The interference site o divides the disturbed network into three parts: before coupling, during coupling, and after coupling, and reflects the relative position of the interference site o in the disturbed network through the distribution of RC values.

[0077] by Figure 5Taking the crosstalk model in as an example, if the interference site o is taken as the i-th interference site, the i-th crosstalk model corresponding to the i-th interference site of the predicted disturbed network can be determined, R di is the first resistance value of the driving equivalent resistance of the i-th crosstalk model, R si is the second resistance value of the upstream resistor of the i-th interference site, R ei is the third resistance value of the downstream resistor of the i-th interference site, C xi is the capacitance value of the coupling capacitance between the predicted disturbed network and the disturber, C Li is the capacitance value of the load capacitance at the output of the disturbed network to be predicted, t agg,ri is the transition time of the switching signal of a disturber in the predicted disturbed network, C 1i is the capacitance value of the first connection capacitor of the i-th crosstalk model and C 2i is the capacitance value of the second connection capacitor of the i-th crosstalk model.

[0078] Furthermore, based on R di 、R si and C xi The first time parameter t of the i-th crosstalk model can be determined xi , for example, the first time parameter t of the i-th crosstalk model xi =(R di +R si )*C xi .

[0079] Furthermore, based on the R in the i-th crosstalk model di 、R si 、R ei 、C xi 、C Li 、C 1i and C 2i The second time parameter of the i-th crosstalk model may be determined, for example, the second time parameter t vi =(R di +R si )*(C xi +C 2i +C Li )+(R ei *C Li +R di *C 1i ).

[0080] Furthermore, according to R in the i-th crosstalk model di 、R si 、C xi 、C Li , the first time parameter t xi , tvi and t agg,ri The site attribute parameter set of the i-th interference site can be determined. It can be understood that, preferably, the site attribute parameter set of the i-th interference site includes R di 、R si 、C xi 、C Li , the first time parameter t xi , t vi and t agg,ri When , the crosstalk signal prediction model has the best prediction effect on the crosstalk signal of the disturbed network and the highest prediction accuracy.

[0081] It should be noted that the crosstalk model proposed in this embodiment not only considers that the strength of the crosstalk noise signal is related to the equivalent coupling strength at the interference site (usually measured by parameters such as coupling capacitance and the transition time of the interferer's switching signal), but also considers that the crosstalk noise signal also varies with the location of the interference site. Therefore, the crosstalk model proposed in this embodiment can more accurately model the interconnect circuit.

[0082] For further information, see Figure 6 In this embodiment, to determine the parameters in the crosstalk key parameter set of the victim network to be predicted, an analysis model is introduced. The model includes a driver unit 100, a total crosstalk effect simulation unit 200, and a load unit 300. Node K1 serves as the driver pin connected to the victim network to be predicted, node K2 serves as the input pin of the load unit 300, and node K3 serves as the output pin of the load unit 300. Circuit signal activity is input from the driver pin of the driver unit 100, is affected by the crosstalk signal, and then reaches the load unit 300. This signal is also affected by the load unit 300 downstream of the circuit.

[0083] Therefore, the crosstalk key parameter set of the victim network to be predicted includes the transition time t of the switching signal at the driving pin connected to the victim network to be predicted. r1y , the transition time t of the switching signal at the input pin of the load unit 300 connected to the output end of the to-be-predicted disturbed network r2y , the transition time t of the switching signal at the output pin of the load unit 300 connected to the disturbed network to be predicted r3y , the non-signal integrity mode delay D at the driver pin of the victim network to be predicted 1y , the non-signal integrity mode delay D at the input pin of the load unit 300 connected to the victim network to be predicted 2y , the non-signal integrity mode delay D at the output pin of the load unit 300 connected to the victim network to be predicted 3y and the voltage level V of the load unit 300 thyIt can be understood that, preferably, the crosstalk key parameter set of the to-be-predicted disturbed network includes V thy , t r1y , t r2y , t r3y , D 1y , D 2y , D 3y When , the crosstalk delay prediction model has the best prediction effect on the crosstalk delay of the disturbed network and the highest prediction accuracy.

[0084] Step S200: using a pre-trained crosstalk signal prediction model to predict the i-th crosstalk signal value of the i-th interference site of the to-be-predicted disturbed network based on the own attribute parameter set of the to-be-predicted disturbed network and the site attribute parameter set of the i-th interference site.

[0085] For example, the ith feature vector x can be determined based on the attribute parameter set of the network to be predicted and the site attribute parameter set of the ith interference site: 1i =(R di , R si , C xi , C Li , t xi , t vi , C ty , t agg,ri , t vic,ry , C c,ty ) T , the crosstalk signal prediction model pre-trained with the feature vector can predict the disturbed network to be predicted corresponding to the feature vector, and obtain the i-th crosstalk signal value of the i-th interference site of the disturbed network to be predicted.

[0086] It can be understood that there may be multiple interference sites in the to-be-predicted disturbed network. Through step S200, the crosstalk signal values ​​corresponding to all the interference sites in the to-be-predicted disturbed network can be determined.

[0087] Step S300: using a pre-trained crosstalk delay prediction model to predict the crosstalk delay of the victim network to be predicted based on I crosstalk signal values ​​and a set of crosstalk key parameters of the victim network to be predicted.

[0088] It can be understood that the total number of interference sites corresponding to different disturbed networks is different. If you want to use a pre-trained crosstalk delay prediction model to predict the crosstalk delay of the disturbed network to be predicted based on I crosstalk signal values ​​and the crosstalk key parameter set of the disturbed network to be predicted, you need to ensure that the number of crosstalk signal values ​​input into the crosstalk delay prediction model is a predetermined number, that is, use a predetermined number of crosstalk signal values ​​and each parameter in the crosstalk key parameter set to form a feature vector of a predetermined size, and then input the feature vector into the pre-trained crosstalk delay prediction model to obtain the crosstalk delay of the disturbed network to be predicted.

[0089] In this embodiment, two embodiments of determining the feature vector are further considered in the two cases where I is greater than or equal to a predetermined number and less than a predetermined number. For example, see Figure 7 , the following steps S310 to S350 can be used to determine the feature vector based on the size of I:

[0090] Step S310: Arrange the I crosstalk signal values ​​in descending order to obtain a crosstalk signal sequence.

[0091] It can be understood that the larger the crosstalk signal value, the greater the impact on the predicted disturbed network. In this embodiment, the crosstalk signal sequence is obtained by arranging the I crosstalk signal values ​​in order from large to small, so as to determine that the crosstalk signal value that is ranked higher in the crosstalk signal sequence is more important for the prediction process of the crosstalk delay.

[0092] Step S320, determining whether I is greater than or equal to a predetermined number.

[0093] When I is greater than or equal to the predetermined number, steps S330 and S350 are executed; when I is less than the predetermined number, steps S340 and S350 are executed.

[0094] Among them, the predetermined number can be recorded as m. It should be noted that when using a pre-trained crosstalk delay prediction model to predict the crosstalk delay of the predicted disturbed network, the predetermined number is m. Similarly, when pre-training the crosstalk delay prediction model, the predetermined number must also be ensured to be m.

[0095] Step S330 : determining a feature vector according to a predetermined number of crosstalk signal values ​​ranked at the top in the crosstalk signal sequence and the crosstalk key parameter set.

[0096] It can be understood that when I is greater than or equal to the predetermined number, the sub-vector corresponding to the predetermined number of crosstalk signal values ​​can be expressed as B m ={b1, b2, ..., b m}, m≤I, b1 is the crosstalk signal value at the first position in the crosstalk signal sequence, b2 is the crosstalk signal value at the second position in the crosstalk signal sequence, b m is the crosstalk signal value at the mth position in the crosstalk signal sequence.

[0097] Step S340 : adding a predetermined supplementary value after the crosstalk signal sequence so that the crosstalk signal sequence includes the predetermined number of crosstalk signal values, and determining a feature vector according to the predetermined number of crosstalk signal values ​​and the crosstalk key parameter set.

[0098] It can be understood that when I is less than the predetermined number, the sub-vector corresponding to the predetermined number of crosstalk signal values ​​can be expressed as B m ={b1, b2, ..., b n ,0,……,0},b n is the crosstalk signal value at the nth position in the crosstalk signal sequence, where n is less than m.

[0099] Furthermore, in step S330 and step S340, based on the respective obtained sub-vectors B m With V thy , t r1y , t r2y , t r3y , D 1y , D 2y , D 3y A eigenvector can be determined. For example, the eigenvector can be expressed as x 2i =(V thy , t r1y , t r2y , t r3y , D 1y , D 2y , D 3y , B m ) T .

[0100] Step S350: input the feature vector into a pre-trained crosstalk delay prediction model, so as to predict the crosstalk delay of the victim network to be predicted by the pre-trained crosstalk delay prediction model.

[0101] In this embodiment, a pre-trained crosstalk signal prediction model is used to quickly, accurately, and stably predict the i-th crosstalk signal value of the i-th interference site of the to-be-predicted disturbed network based on the set of its own attribute parameters and the set of site attribute parameters of the i-th interference site. Moreover, a pre-trained crosstalk delay prediction model is used to quickly, accurately, and stably predict the crosstalk delay of the to-be-predicted disturbed network based on all crosstalk signal values ​​and the set of crosstalk key parameters of the to-be-predicted disturbed network. The crosstalk delay prediction process of this embodiment is simpler and more efficient, which can greatly improve the efficiency of signal integrity analysis in STA and effectively reduce computing costs.

[0102] Example 2

[0103] See Figure 8 In an embodiment of the present application, a method for pre-training a crosstalk signal prediction model is proposed. The method requires obtaining a first training data set in advance, and using the first training data set to iteratively train the crosstalk signal prediction model until the loss function corresponding to the crosstalk signal prediction model converges.

[0104] It can be understood that the crosstalk signal prediction model is a nonlinear model, which requires a large number of training samples to be trained iteratively multiple times to obtain the optimal values ​​corresponding to various parameters of the crosstalk signal prediction model.

[0105] Furthermore, pre-acquiring the first training data set includes the following steps:

[0106] Step S10: Obtain the j-th set of self-attribute parameters of the disturbed network used for training, 1≤j≤J, and J is the total number of disturbed networks used for training.

[0107] It can be understood that the number and type of parameters in the set of self-attribute parameters of the disturbed network used for training should be consistent with the set of self-attribute parameters of the disturbed network to be predicted. For example, the set of self-attribute parameters of the disturbed network used for training can be expressed as {C tj , t vic,rj , C c,tj}, where C tj is the capacitance value of the cumulative capacitance of the jth perturbed network used for training, t vic,rj is the transition time of the switching signal of the jth perturbed network used for training, C c,tj is the capacitance value of the total coupling capacitance of the jth disturbed network used for training.

[0108] Step S20: Obtain a set of site attribute parameters related to an interference site in the j-th disturbed network for training.

[0109] Determine the site attribute parameter set associated with an interference site in the jth disturbed network used for training. It is also necessary to determine the crosstalk model corresponding to the interference site. Moreover, the site attribute parameter set of the interference site should be consistent with the number and type of parameters in the site attribute parameter set of the disturbed network to be predicted.

[0110] For example, the set of site attribute parameters related to an interference site in the jth disturbed network for training can be expressed as {R dj , R sj , C xj , C Lj , t xj , t vj , t agg,rj}. Among them, R dj is the first resistance value associated with an interference site in the jth disturbed network used for training, R sj is the second resistance value associated with an interference site in the jth disturbed network used for training, C xj is the capacitance value of the coupling capacitor associated with an interference site in the jth disturbed network used for training, C Lj is the capacitance value of the load capacitance associated with a disturbance site in the jth disturbed network used for training, t xj is the first time parameter associated with an interference site in the jth disturbed network used for training, t vj is the second time parameter associated with an interference site in the jth disturbed network for training and t agg,rj is the transition time of the distractor switching signal associated with a distractor site in the jth distractor network used for training.

[0111] Step S30 , calculating the product of the peak voltage of the signal waveform output by the crosstalk model corresponding to the one interference site and the time period corresponding to the signal waveform.

[0112] This step is to determine the area by multiplying the time period and the peak voltage, and use the area as the label of a training sample.

[0113] Step S40: Determine the jth signal training sample in the first training data set based on the jth set of self-attribute parameters of the disturbed network used for training and the set of site attribute parameters corresponding to the jth disturbed network used for training, and the label corresponding to the jth signal training sample is the product.

[0114] For example, the jth signal training sample can be expressed as x 1j =(R dj , R sj , C xj , C Lj , txj , t vj , Ct j , t agg,rj , t vic,rj , C c,tj ) T .

[0115] It can be understood that by repeating steps S10 to S40 J times, J signal training samples can be obtained, and then the crosstalk signal prediction model can be iteratively trained using the J signal training samples to obtain a trained crosstalk signal prediction model. Based on the trained crosstalk signal prediction model, the victim signal value of any victim network can be quickly and accurately determined.

[0116] Example 3

[0117] See Figure 9 In an embodiment of the present application, a method for pre-training the crosstalk delay prediction model in the above-mentioned embodiment 1 is proposed, which includes pre-acquiring a second training data set and iteratively training the crosstalk delay prediction model using the second training data set until the loss function corresponding to the crosstalk delay prediction model converges.

[0118] Furthermore, pre-acquiring the second training data set includes the following steps:

[0119] Step S50: Obtain a crosstalk key parameter set of the jth victim network for training and a signal integrity mode delay of the jth victim network for training.

[0120] The crosstalk key parameter set of the jth victim network used for training should be consistent with the crosstalk key parameter set of the victim network to be predicted. For example, the crosstalk key parameter set of the jth victim network used for training can be expressed as {V thj , t r1j , t r2j , t r3j , D 1j , D 2j , D 3j}, where t r1j is the transition time of the switching signal at the driving pin connected to the jth perturbed network for training, t r2j is the transition time of the switching signal at the input pin of the load unit connected to the output terminal of the jth perturbed network for training, t r3j is the transition time of the switching signal at the output pin of the load unit connected to the jth perturbed network for training, D 1j is the non-signal integrity mode delay at the driver pin of the jth victim network used for training, D 2jis the non-signal integrity mode delay at the input pin of the load unit connected to the jth victim network for training, D 3j is the non-signal integrity mode delay and V at the output pin of the load cell connected to the jth victim network for training. thy is the voltage level of the load cell connected to the jth victim network for training.

[0121] Step S60: using the pre-trained crosstalk signal prediction model to predict each crosstalk signal value of the jth disturbed network for training based on the jth disturbed network's own attribute parameter set for training and the site attribute parameter set of each interfering site.

[0122] Step S70: determining the jth delay training sample in the second training data set according to the jth crosstalk key parameter set and all crosstalk signal values ​​of the victim network for training, wherein the label corresponding to the jth delay training sample is the signal integrity mode delay.

[0123] It can be understood that during training, all crosstalk signal values ​​still need to be sorted in descending order, and the sub-vectors corresponding to all crosstalk signal values ​​are determined according to different methods based on the total number of all crosstalk signal values. For example, if the total number of all crosstalk signal values ​​of the j-th victim network used for training is less than the predetermined number m, then the j-th sub-vector can be expressed as B mj ={b 1j , b 2j ,……,b nj ,0,……,0},b nj is the crosstalk signal value at the nth position among all the crosstalk signal values ​​corresponding to the jth victim network for training, n is less than m, and m is less than or equal to the total number of all crosstalk signal values ​​of the jth victim network for training; if the total number of all crosstalk signal values ​​of the jth victim network for training is greater than or equal to the predetermined number m, then the jth subvector can be expressed as B mj ={b 1j , b 2j ,……,b mj}, b mj is the crosstalk signal value at the mth position among all crosstalk signal values ​​corresponding to the jth disturbed network used for training.

[0124] For example, the jth delayed training sample can be expressed as x 2j =(V thj , t r1j , t r2j , t r3j , D 1j , D 2j , D 3j , Bmj ) T .

[0125] It can be understood that by repeating steps S50-S70 J times, J delay training samples can be obtained. Then, the crosstalk delay prediction model can be iteratively trained using the J delay training samples to obtain a trained crosstalk delay prediction model. Based on the trained crosstalk delay prediction model, the crosstalk delay of any victim network can be quickly and accurately determined.

[0126] Example 4

[0127] See Figure 10 In an embodiment of the present application, a crosstalk delay prediction device 10 is also proposed. The crosstalk delay prediction device 10 includes: a parameter acquisition module 11, a signal prediction module 12 and a delay prediction module 13.

[0128] A parameter acquisition module 11 is used to obtain a set of self-attribute parameters of the disturbed network to be predicted, a set of crosstalk key parameters of the disturbed network to be predicted, and a set of site attribute parameters of the i-th interference site, where the i-th interference site is a point where the corresponding interferer of the disturbed network to be predicted interferes with the disturbed network to be predicted, 1≤i≤I, and I is the total number of interference sites affecting the disturbed network to be predicted; a signal prediction module 12 is used to use a pre-trained crosstalk signal prediction model to predict the i-th crosstalk signal value of the i-th interference site of the disturbed network to be predicted based on the self-attribute parameter set of the disturbed network to be predicted and the set of site attribute parameters of the i-th interference site; a delay prediction module 13 is used to use a pre-trained crosstalk delay prediction model to predict the crosstalk delay of the disturbed network to be predicted based on I crosstalk signal values ​​and the set of crosstalk key parameters of the disturbed network to be predicted.

[0129] Furthermore, an embodiment of the present application also proposes a terminal device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program runs on the processor, it executes the crosstalk delay prediction method described in the above embodiment of the present application.

[0130] Furthermore, an embodiment of the present application also provides a readable storage medium storing a computer program. When the computer program is run on a processor, the crosstalk delay prediction method described in the above embodiment of the present application is executed.

[0131] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and structure diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in an alternative implementation, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes 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 structure diagram and / or flowchart, and the combination of boxes in the structure diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0132] In addition, the functional modules or units in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0133] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the 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 and includes several instructions for enabling a computer device (which can be a smart phone, a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned readable storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0134] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the scope of protection of the present invention.

Claims

1. A crosstalk delay prediction method, characterized in that: The method comprises: Obtain a set of attribute parameters of the network to be predicted, a set of crosstalk key parameters of the network to be predicted, and a set of site attribute parameters of the i-th interference site, where the i-th interference site is a point where the interferer corresponding to the network to be predicted interferes with the network to be predicted, 1≤i≤I, and I is the total number of interference sites that affect the network to be predicted; Using a pre-trained crosstalk signal prediction model, based on the self attribute parameter set of the to-be-predicted disturbed network and the site attribute parameter set of the i-th interfering site, predict the i-th crosstalk signal value of the i-th interfering site of the to-be-predicted disturbed network; The crosstalk delay of the to-be-predicted victim network is predicted using a pre-trained crosstalk delay prediction model based on I crosstalk signal values ​​and a set of crosstalk key parameters of the to-be-predicted victim network.

2. The crosstalk delay prediction method according to claim 1, wherein: The method of using a pre-trained crosstalk delay prediction model to predict the crosstalk delay of the victim network to be predicted based on I crosstalk signal values ​​and a crosstalk key parameter set of the victim network to be predicted includes: Arrange the I crosstalk signal values ​​in descending order to obtain a crosstalk signal sequence; Determining whether I is greater than or equal to a predetermined number; When I is greater than or equal to the predetermined number, a feature vector is determined based on a predetermined number of crosstalk signal values ​​that are ranked first in the crosstalk signal sequence and the crosstalk key parameter set; or when I is less than the predetermined number, a predetermined supplementary value is added after the crosstalk signal sequence so that the crosstalk signal sequence includes the predetermined number of crosstalk signal values, and a feature vector is determined based on the predetermined number of crosstalk signal values ​​and the crosstalk key parameter set; The feature vector is input into a pre-trained crosstalk delay prediction model, so as to predict the crosstalk delay of the to-be-predicted disturbed network through the pre-trained crosstalk delay prediction model.

3. The crosstalk delay prediction method according to claim 1, wherein: The attribute parameter set of the victim network to be predicted includes at least one of the capacitance value of the accumulated capacitance of the victim network to be predicted, the transition time of the switching signal of the victim network to be predicted, and the capacitance value of the total coupling capacitance of the victim network to be predicted.

4. The crosstalk delay prediction method according to claim 1, wherein: Obtaining a site attribute parameter set of the i-th interference site, including: Determine the first resistance value of the driving equivalent resistance of the i-th crosstalk model corresponding to the i-th interference site, the second resistance value of the upstream resistance of the i-th interference site, the third resistance value of the downstream resistance of the i-th interference site, the capacitance value of the coupling capacitance between the victim network to be predicted and the interferer in the i-th crosstalk model, the capacitance value of the load capacitance at the output of the victim network to be predicted in the i-th crosstalk model, the transition time of the switching signal of the interferer in the i-th crosstalk model, the capacitance value of the first connection capacitance in the i-th crosstalk model, and the capacitance value of the second connection capacitance in the i-th crosstalk model; Determining a first time parameter of the i-th crosstalk model based on the first resistance, the second resistance, and the capacitance value of the coupling capacitor; determining a second time parameter of the i-th crosstalk model based on the first resistance, the second resistance, the third resistance, the capacitance value of the coupling capacitor, the capacitance value of the load capacitor, the capacitance value of the first connection capacitor, and the capacitance value of the second connection capacitor; Determine the site attribute parameter set of the i-th interference site based on the first resistance, the second resistance, the capacitance value of the coupling capacitor, the capacitance value of the load capacitor, the first time parameter, the second time parameter and the transition time of the interferer switching signal.

5. The crosstalk delay prediction method according to claim 4, characterized in that: The first time parameter of the i-th crosstalk model is determined using the following formula: t xi =(R di +R si )*C xi t xi represents the first time parameter of the i-th crosstalk model, R di represents the first resistance value of the driving equivalent resistance of the i-th crosstalk model, R si The second resistance value of the upstream resistor of the i-th interference point, C xi represents the capacitance value of the coupling capacitance between the victim network to be predicted and the interferer in the i-th crosstalk model.

6. The crosstalk delay prediction method according to claim 4, characterized in that: The second time parameter of the i-th crosstalk model is determined using the following formula, including: t vi =(R di +R si )*(C xi +C 2i +C Li )+(R ei *C Li +R di *C 1i ) t vi represents the second time parameter of the i-th crosstalk model, R di represents the first resistance value of the driving equivalent resistance of the i-th crosstalk model, R si The second resistance value of the upstream resistor of the i-th interference point, C xi represents the capacitance value of the coupling capacitance between the victim network to be predicted and the interferer in the i-th crosstalk model, C Li represents the capacitance value of the load capacitance at the output of the disturbed network to be predicted in the i-th crosstalk model, C 1i represents the capacitance value of the first connection capacitor in the i-th crosstalk model, C 2i represents the capacitance value of the second connection capacitor in the i-th crosstalk model.

7. The crosstalk delay prediction method according to claim 1, wherein: The crosstalk key parameter set includes at least one of the transition time of the switching signal at the driving pin connected to the predicted disturbed network, the transition time of the switching signal at the input pin of the load unit connected to the output end of the predicted disturbed network, the transition time of the switching signal at the output pin of the load unit, the non-signal integrity mode delay at the driving pin, the non-signal integrity mode delay at the input pin, the non-signal integrity mode delay at the output pin and the voltage level of the load unit.

8. The crosstalk delay prediction method according to any one of claims 1 to 7, characterized in that: Also includes: Acquire a first training data set in advance, and iteratively train the crosstalk signal prediction model using the first training data set until a loss function corresponding to the crosstalk signal prediction model converges; The pre-acquisition of the first training data set includes: Get the attribute parameter set of the jth disturbed network used for training, 1≤j≤J, J is the total number of disturbed networks used for training; Obtaining a set of site attribute parameters associated with an interference site in the j-th disturbed network for training; Calculating the product of the peak voltage of the signal waveform output by the crosstalk model corresponding to the one interference site and the time period corresponding to the signal waveform; The jth signal training sample in the first training data set is determined according to the jth set of self-attribute parameters of the disturbed network used for training and the set of site attribute parameters corresponding to the jth disturbed network used for training, and the label corresponding to the jth signal training sample is the product.

9. The crosstalk delay prediction method according to any one of claims 1 to 7, characterized in that: Also includes: Pre-acquiring a second training data set, and iteratively training the crosstalk delay prediction model using the second training data set until a loss function corresponding to the crosstalk delay prediction model converges; The pre-acquisition of the second training data set includes: Obtaining a crosstalk key parameter set of the jth victim network for training and a signal integrity mode delay of the jth victim network for training; Predicting each crosstalk signal value of the jth victim network for training using the pre-trained crosstalk signal prediction model based on the jth victim network's own attribute parameter set for training and the site attribute parameter set of each interfering site; The jth delay training sample in the second training data set is determined according to the crosstalk key parameter set and all crosstalk signal values ​​of the jth victim network for training, and the label corresponding to the jth delay training sample is the signal integrity mode delay.

10. A crosstalk delay prediction device, characterized in that: The device comprises: A parameter acquisition module is used to obtain a set of attribute parameters of the network to be predicted, a set of crosstalk key parameters of the network to be predicted, and a set of site attribute parameters of the i-th interference site, where the i-th interference site is a point where the interferer corresponding to the network to be predicted interferes with the network to be predicted, 1≤i≤I, and I is the total number of interference sites that affect the network to be predicted; A signal prediction module is used to predict the i-th crosstalk signal value of the i-th interference site of the to-be-predicted disturbed network based on the set of its own attribute parameters and the set of site attribute parameters of the i-th interference site using a pre-trained crosstalk signal prediction model; The delay prediction module is used to predict the crosstalk delay of the to-be-predicted victim network based on I crosstalk signal values ​​and the crosstalk key parameter set of the to-be-predicted victim network using a pre-trained crosstalk delay prediction model.

11. A terminal device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is run on the processor, the method for predicting crosstalk delay according to any one of claims 1 to 9 is executed.

12. A readable storage medium, characterized in that: A computer program is stored therein, and when the computer program is run on a processor, the crosstalk delay prediction method according to any one of claims 1 to 9 is executed.

Images