An eye diagram search space optimization method, system and computer storage medium

Abstract

The application provides an eye diagram search space optimization method, system and computer storage medium, comprising: presetting an initial search space of a circuit parameter to be optimized; constructing an initial simulation case combination according to the initial search space; generating several simulation net tables and performing simulation to obtain corresponding eye diagram performance response values; establishing a second-order polynomial meta-model according to the eye diagram performance response values and the circuit parameter; calculating an eye diagram performance response maximum value and a corresponding circuit parameter value; presetting a scaling coefficient and updating the search space according to the eye diagram performance response maximum value and the corresponding circuit parameter value; updating the simulation case combination according to the updated search space to perform iterative calculation of the search space, and outputting an optimal search space. The scheme can automatically focus the search range on a potential optimal area, accurately locate a global optimal parameter combination that optimizes the eye height and eye width performance with the least simulation times, and is further beneficial to signal integrity judgment of a high-speed serial link.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of integrated circuit computer-aided design technology, in particular to an eye diagram search space optimization method, system and computer storage medium. BACKGROUND

[0002] In the integrated circuit computer-aided design, with the data rate rising to several Gbps or even hundreds of Gbps, the signal integrity (SI) design of high-speed serial link becomes crucial. The eye diagram quality (mainly measured by eye height and eye width) is the core indicator for evaluating the SI performance, which directly determines the system bit error rate.

[0003] However, in finding the optimal eye diagram, the prior art usually adopts manual trial and error, multiple simulations in the preset parameter space, etc., which has problems of serious dependence on engineers' experience, high simulation cost, low optimization efficiency, easy to fall into local optimum, etc. Therefore, there is an urgent need in the field for a method capable of intelligently and reliably finding the global optimal eye diagram with a limited number of simulations. SUMMARY

[0004] The purpose of the present application is to provide an eye diagram search space optimization method, system and computer storage medium, which intelligently combines experimental design, SPICE simulation, meta-modeling and parameter space dynamic contraction in one cycle, can automatically focus the search range on the potential optimal area, and gradually and accurately locate the global optimal parameter combination that optimizes the eye height and eye width performance with the least number of simulations, thereby facilitating the signal integrity judgment of high-speed serial link.

[0005] The technical solutions provided by the present application are as follows:

[0006] In a first aspect, the present application provides an eye diagram search space optimization method for signal integrity simulation of high-speed serial link, comprising the steps of:

[0007] S100, preset an initial search space of circuit parameters to be optimized, the initial search space comprising an initial minimum value, an initial maximum value and an initial intermediate value;

[0008] S200, constructing an initial simulation case combination according to the initial minimum value, the initial maximum value and the initial intermediate value;

[0009] S300, generating several simulation netlists according to the simulation case combination and performing SPICE simulation respectively to obtain corresponding eye diagram performance response values;

[0010] S400, establishing a second-order polynomial meta-model according to the eye diagram performance response values and the circuit parameters;

[0011] S500, obtaining an eye diagram performance response maximum value and a corresponding circuit parameter value according to the second-order polynomial meta-model;

[0012] S600, presetting a scaling coefficient, and updating a search space, a maximum value, a minimum value and a middle value of the search space according to the eye diagram performance response maximum value and the corresponding circuit parameter value;

[0013] S700, updating simulation case combinations according to the maximum value, the minimum value and the middle value of the updated search space, repeating steps S300 to S600, and performing iterative calculation of the search space until a preset iteration termination condition is reached, and outputting an optimal search space.

[0014] In some embodiments, after the second-order polynomial meta-model is established, the method further comprises:

[0015] When a difference between the eye diagram performance response maximum value obtained according to the second-order polynomial meta-model and an actual simulation value is within an error range, and an adjusted determination coefficient of the second-order polynomial meta-model is greater than a preset threshold, it is determined that the second-order polynomial meta-model is reliable, otherwise it is determined that the second-order polynomial meta-model is unreliable.

[0016] In some embodiments, the method further comprises:

[0017] When it is determined that the second-order polynomial meta-model is reliable, a willingness curve is adjusted, a series of willingness curves are obtained based on the eye diagram performance response maximum value by adjusting model coefficients, if a higher eye diagram performance response maximum value is found, the eye diagram performance response maximum value is updated, and if no higher eye diagram performance response maximum value is found, the eye diagram performance response maximum value is maintained unchanged.

[0018] In some embodiments, the method further comprises:

[0019] When it is determined that the second-order polynomial meta-model is unreliable, a better one of the actual simulation value in the current iteration and the actual simulation value in the last iteration is selected as the eye diagram performance response maximum value.

[0020] In some embodiments, the updating of the search space and the maximum value, the minimum value and the middle value of the search space comprises:

[0021] Let a0 be an initial minimum value, c0 be an initial maximum value; a be a minimum value of the search space in the last iteration, c be a maximum value of the search space in the last iteration, b be a point in the interval [a, c]; x be the eye diagram performance response maximum value in the current iteration, t = b-a, alpha be a preset scaling factor; aNew be the updated minimum value of the search space in the current iteration, cNew be the updated maximum value of the search space in the current iteration, bNew be the updated middle value of the search space in the current iteration;

[0022] When it is judged that x is in the interval [a, c], the current x value is reserved; when it is judged that x is not in the interval [a, c], x is set to the boundary value closer to the boundary, that is, x = max (a, min (x, c));

[0023] When it is judged that and are in the interval [a, c], then bNew = x, ;

[0024] When it is judged that x-t <= a0, and , aNew = a0, , ;

[0025] When it is judged that x-t <= a0, and , aNew = a0, bNew = x, ;

[0026] When it is judged that x-t > a0, aNew = x-t, bNew = x, cNew = x+t;

[0027] When it is judged that x+t >= c0, and , , cNew = c0;

[0028] When it is judged that x+t >= c0, and , bNew = x, cNew = c0;

[0029] When it is judged that and x+t < c0, aNew = x-t, bNew = x, cNew = x+t.

[0030] In some embodiments, the eye diagram performance response value includes eye height data or eye width data.

[0031] In some embodiments, the establishing the second-order polynomial metamodel specifically comprises:

[0032] In the first iteration, the second-order polynomial metamodel is constructed by using the least square method;

[0033] In the non-first iteration, the second-order polynomial metamodel is constructed by using the stepwise regression method.

[0034] In some embodiments, the preset iteration termination condition comprises a preset iteration number.

[0035] In a second aspect, the present application provides an eye diagram search space optimization system, comprising a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the eye diagram search space optimization method of the first aspect.

[0036] In a third aspect, the present application provides a computer storage medium having a computer program or instructions stored thereon, wherein the computer program or instructions are executed by a processor to implement the steps of the eye diagram search space optimization method of the first aspect.

[0037] The eye diagram search space optimization method, system and computer storage medium provided by the present application have at least the following technical effects:

[0038] 1) The present application intelligently combines experimental design, SPICE simulation, metamodeling and parameter space dynamic contraction in one cycle, can automatically focus the search range on the potential optimal region, and gradually and accurately locate the global optimal parameter combination that optimizes the eye height and eye width performance with the least number of simulations, thereby facilitating the signal integrity judgment of high-speed serial links.

[0039] 2) High efficiency: by iterative contraction, the simulation resources are concentrated in the most potential area, multiple results can be obtained by parallel simulation in one cycle, and the optimization can be quickly completed, thereby saving a lot of calculation time and cost compared with traditional methods.

[0040] 3) Global optimization: avoids the local optimal trap of manual method and single experimental design method, and can more reliably find the global optimal point or approximate global optimal point through multiple iterations and model guidance.

[0041] 4) Automation and intelligence: realizes full-process automation, reduces the dependence on engineers' experience, and can realize adaptive contraction of parameter space.

[0042] 5) Strong practicability: the method has strong universality and can be integrated into existing EDA design flow and is suitable for various optimization scenarios of high-speed serial links. BRIEF DESCRIPTION OF DRAWINGS

[0043] The above features, technical characteristics, advantages and implementation manners of the present application will be further described in a clear and understandable manner in combination with the preferred embodiments and the accompanying drawings.

[0044] Figure 1 is a schematic diagram of the overall flow of an embodiment of the present application. DETAILED DESCRIPTION

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the specific embodiments of the present application will be described below with reference to the drawings. Obviously, the drawings in the following description only represent some embodiments of the present application, and for those skilled in the art, other drawings can be obtained from these drawings without creative labor, and other embodiments can also be obtained.

[0046] In order to make the drawings simple, only the parts related to the present application are shown in each drawing, which do not represent the actual structure of the product. In addition, in order to make the drawings simple and easy to understand, in some drawings, only one of the components with the same structure or function is shown, or only one of them is marked. In this article, "one" not only means "only one", but also means "more than one".

[0047] In integrated circuit computer-aided design, as the data rate rises to several Gbps or even hundreds of Gbps, the signal integrity (SI) design of high-speed serial link becomes crucial. Eye diagram quality (mainly measured by eye height and eye width) is the core indicator for evaluating SI performance, which directly determines the bit error rate of the system. When searching for the optimal eye diagram performance, the existing technology usually adopts the following methods:

[0048] 1) Manual trial and error method: the design engineer relies on experience to manually adjust the key parameters such as transmitter equalization, receiver equalization, and terminal impedance. After each adjustment, a time-consuming SPICE simulation is run, and the next adjustment direction is determined according to the results. This method is highly dependent on the experience of engineers, has extremely low efficiency, and is easily trapped in a local optimal solution, and cannot guarantee to find the global optimal performance.

[0049] 2) Single experiment design (DOE) method: a one-time experiment design (such as full factorial design, partial factorial design) is used to perform multiple simulations in a pre-set parameter space, and a linear or second-order meta-model is constructed to find the optimal solution. However, one-time DOE has the following significant defects: when the initial parameter range is set wide, the global meta-model constructed has insufficient prediction accuracy near the optimal region; when there is complex nonlinear interaction between parameters, the single model may not accurately capture the true position of the optimal point, resulting in optimization failure.

[0050] 3) Optimization algorithm integration: Some advanced EDA tools integrate optimization algorithms, but the core of them is mostly genetic algorithm, gradient descent method, etc. These algorithms often need a large number of simulation times to converge, and for SPICE simulation that takes several minutes to several hours, the calculation cost is unacceptable.

[0051] Therefore, the prior art has problems of high simulation cost, low optimization efficiency, and easy to fall into local optimum. There is an urgent need in the art for a method that can intelligently and reliably find the globally optimal eye diagram performance with a limited number of simulations.

[0052] The present application is to overcome the above-mentioned defects, and provides an iterative and adaptive optimization method and system. The method intelligently combines experimental design, SPICE simulation, meta-modeling and dynamic parameter space contraction in one cycle, can automatically focus the search range on the potential optimal region, and gradually and accurately locate the globally optimal parameter combination that optimizes the eye height and eye width performance with the least number of simulations, thereby facilitating the signal integrity design of high-speed serial links. The present application will be described in detail below with reference to the accompanying drawings:

[0053] In one embodiment, with reference to the accompanying drawings Figure 1 The present application provides an eye diagram search space optimization method for signal integrity simulation of high-speed serial links, comprising the steps of:

[0054] S100, preset an initial search space of a circuit parameter to be optimized, the initial search space comprising an initial minimum value, an initial maximum value and an initial intermediate value.

[0055] At the initial stage, define a parameter variable composed of a minimum value, a maximum value and an intermediate value. For example, in one example, the circuit parameter is CTLE gain, the initial minimum value (a0) is 0 dB, the initial maximum value (c0) is 12 dB, and the initial intermediate value (b0) is 6 dB.

[0056] S200, constructing an initial simulation case combination according to the initial minimum value, the initial maximum value and the initial intermediate value.

[0057] According to the initial minimum value, the initial maximum value and the initial intermediate value of the initial search space, a plurality of initial simulation case combinations can be constructed, for example, 5 groups.

[0058] S300, generating a plurality of simulation netlists according to the simulation case combinations and performing SPICE simulation respectively to obtain corresponding eye diagram performance response values.

[0059] In the first simulation, different netlist files (for example, 5) can be generated by modifying the corresponding variables in the SPICE netlist through the simulation case combination. By running the SPICE simulation of the 5 netlists in turn and parsing the simulation result output file, the eye diagram performance response value after each simulation can be extracted. The eye diagram performance response value includes eye height data or eye width data, etc.

[0060] S400, a second-order polynomial meta-model is established according to the eye diagram performance response value and the circuit parameter.

[0061] In this scheme, all circuit parameter values (X) and eye diagram performance response values, such as eye height values (Y), are fitted into a second-order polynomial meta-model. The second-order polynomial meta-model is a kind of "proxy model" that uses a quadratic polynomial to approximate a complex input-output relationship. This scheme applies it to the fitting of eye diagram performance response values and circuit parameters, and does not make other changes to the model itself, so it will not be described again.

[0062] And since this scheme involves iterative calculation of the eye diagram search space, in order to ensure accuracy during iteration, in the first iteration, a least squares method is used to construct a second-order polynomial meta-model; in non-first iteration (in the subsequent iteration process), a stepwise regression method is used to construct a second-order polynomial meta-model. Stepwise regression is a kind of "try as you go" variable selection strategy: starting from an empty model or a full model, only one "most cost-effective" addition or deletion action is performed at each step, until the model is no longer significantly improved by continuing to add or delete.

[0063] S500, the maximum value of the eye diagram performance response and the corresponding circuit parameter value are calculated according to the second-order polynomial meta-model.

[0064] After establishing the second-order polynomial meta-model, the predicted maximum value of the eye diagram performance response, i.e., the predicted optimal point x_pred, and the circuit parameter value corresponding to the maximum value of the eye diagram performance response can be obtained by calculation.

[0065] S600, a preset scaling coefficient is set, and the search space and the maximum value, minimum value and intermediate value of the search space are updated according to the maximum value of the eye diagram performance response and the corresponding circuit parameter value.

[0066] After obtaining the maximum value of the eye diagram performance response and the corresponding circuit parameter value, the search space can be updated according to the preset scaling coefficient, and the maximum value, minimum value and intermediate value of the updated search space can be obtained.

[0067] S700, the simulation case combination is updated according to the maximum value, minimum value and intermediate value of the updated search space, and steps S300 to S600 are repeated to perform iterative calculation of the search space until a preset iteration termination condition is reached, and the optimal search space is output.

[0068] According to the maximum value, the minimum value and the intermediate value of the updated search space, the update of the simulation case combination is carried out, the SPICE simulation is re-performed, and the update of the second-order polynomial metamodel is carried out. The search space can be iteratively calculated to obtain the optimal search space and the maximum value, the minimum value and the intermediate value corresponding to the optimal search space. The eye diagram is obtained based on the optimal search space, and the signal integrity of the high-speed serial link is judged. The judgment accuracy is higher. Moreover, the preset iteration termination condition includes a preset iteration number (for example, 5 times). The search range can be automatically focused on the potential optimal region, and the global optimal parameter combination that optimizes the eye height and the eye width performance is accurately positioned with the least simulation number. Compared with the traditional method, a large amount of calculation time and cost can be saved, and the local optimal trap of the manual method and the single experiment design method is avoided.

[0069] According to the established model, the predicted optimal point and the corresponding parameter value are calculated. It is necessary to judge whether the model is reliable. In one specific implementation manner, after the second-order polynomial metamodel is established, the following is further included:

[0070] When the difference between the maximum value of the eye diagram performance response obtained according to the second-order polynomial metamodel and the actual simulation value is within the error range, and the adjusted coefficient of determination of the second-order polynomial metamodel is greater than the preset threshold, it is judged that the second-order polynomial metamodel is reliable, otherwise it is judged that the second-order polynomial metamodel is unreliable.

[0071] Specifically, if the following conditions are met simultaneously: the predicted value at the predicted optimal point x_pred is consistent (or within the error range) with the actual simulation value x_real, and the adjusted coefficient of determination Adj R² of the model is greater than a given threshold (for example, 0.96), then the model is determined to be reliable; otherwise, the model is determined to be unreliable.

[0072] When it is judged that the second-order polynomial metamodel is unreliable, the better one of the actual simulation value x_real in the current iteration and the actual simulation value in the last iteration is selected as the maximum value of the eye diagram performance response.

[0073] When it is judged that the second-order polynomial metamodel is reliable, the willingness curve is adjusted. Based on the current maximum value of the eye diagram performance response x_pred, a series of willingness curves are obtained by adjusting the model coefficients (such as increasing the weight). If a higher maximum value of the eye diagram performance response is found, the maximum value of the eye diagram performance response is updated, and if no higher maximum value of the eye diagram performance response is found, the maximum value of the eye diagram performance response is maintained. In addition, in order to reduce the calculation amount and ensure the accuracy, when a series of willingness curves are obtained by adjusting the model coefficients, the number of times is limited, for example, 10 times. The present scheme performs rapid mathematical optimization through the willingness curve adjustment, and does not perform simulation, which can further reduce the simulation number and reduce the consumption of calculation resources and time.

[0074] In one specific implementation, updating the search space and the maximum, minimum and intermediate values of the search space specifically includes:

[0075] Let a0 be the initial minimum value and c0 be the initial maximum value; a be the minimum value of the search space in the last iteration, c be the maximum value of the search space in the last iteration, b be a point in the interval [a, c]; x be the maximum eye diagram performance response in the current iteration, t = b - a, alpha be a preset scaling coefficient, for example, the preset scaling coefficient is 0.6; aNew be the minimum value of the updated search space in the current iteration, cNew be the maximum value of the updated search space in the current iteration, and bNew be the intermediate value of the updated search space in the current iteration.

[0076] When the adaptive search space contraction is performed, first, it is judged whether the maximum eye diagram performance response (predicted optimal value) obtained by the second-order polynomial meta-model calculation is in the search space [a, c] obtained in the last calculation. When it is judged that x is in the interval [a, c], the current x value is retained; when it is judged that x is not in the interval [a, c], x is set to the boundary value closer to the boundary, that is, x = max (a, min (x, c)).

[0077] Then, the new boundary of the theoretical search space is calculated, and when and are in the interval [a, c], then , bNew = x, . At the same time, the left boundary and the right boundary need to be judged respectively:

[0078] When , x - t <= a0, and , that is, the left boundary exceeds the initial range, aNew = a0, , .

[0079] When , x - t <= a0, and , that is, the left boundary touches the initial range, aNew = a0, bNew = x, .

[0080] When , x - t > a0, that is, the left boundary is contracted without exceeding the initial boundary, aNew = x - t, bNew = x, and cNew = x + t.

[0081] When , x + t >= c0, and , that is, the right boundary exceeds the initial range, , , cNew = c0.

[0082] When judging x+t>=c0, and the right boundary reaches the initial range, bNew=x, cNew=c0.

[0083] When judging and x+t<c0, that is, the right boundary is retracted to a super-initial boundary, aNew=x-t, bNew=x, cNew=x+t.

[0084] The eye diagram search space optimization method provided by the application has at least the following technical effects:

[0085] 1) The scheme can automatically focus the search range on the potential optimal area by intelligently combining experimental design, SPICE simulation, meta-modeling, and parameter space dynamic contraction in one cycle, gradually and accurately locate the global optimal parameter combination that optimizes the eye height and eye width performance with the least number of simulations, and then facilitate the signal integrity judgment of high-speed serial links.

[0086] 2) High efficiency: through iterative contraction, the simulation resources are concentrated in the most potential area, and multiple results can be obtained through one cycle of parallel simulation, thereby the optimization can be quickly completed, and a large amount of calculation time and cost can be saved compared with traditional methods.

[0087] 3) Global optimization: avoids the local optimal trap of manual method and single experimental design method, and can more reliably find the global optimal point or approximate global optimal point through multiple iterations and model guidance.

[0088] 4) Automation and intelligence: realizes full-process automation, reduces the dependence on engineers' experience, and can realize adaptive contraction of parameter space.

[0089] 5) Strong practicability: the method has strong universality and can be integrated into existing EDA design process, and is suitable for optimization scenarios of various high-speed serial links.

[0090] In one embodiment, on the basis of the foregoing embodiment, the application provides an eye diagram search space optimization system, which comprises a memory, a processor, and a computer program stored on the memory, and the processor executes the computer program to realize the steps of the eye diagram search space optimization method of the foregoing embodiment.

[0091] In one embodiment, the application provides a computer program product comprising a computer program or instructions, which, when executed by a processor, realizes the steps of the eye diagram search space optimization method of the foregoing embodiment.

[0092] The eye diagram search space optimization methods of the present application can be implemented in program code executable by a computing device, and thus can be stored in a storage device for execution by a computing device, or can be fabricated as individual integrated circuit modules, or a plurality of modules or steps thereof can be fabricated as a single integrated circuit module. As such, the present application is not limited to any particular hardware and software combination.

[0093] It should be noted that the above-mentioned embodiments can be freely combined according to requirements. The above merely describes the preferred embodiments of the present application, and it should be pointed out that, for those skilled in the art, several improvements and refinements can be made without departing from the principles of the present application, and these improvements and refinements should be considered as falling within the scope of the present application.

Claims

1. An eye diagram search space optimization method, comprising: The method comprises the steps of: S100, presetting an initial search space of a circuit parameter to be optimized, the initial search space comprising an initial minimum value, an initial maximum value and an initial intermediate value; S200, constructing an initial simulation case combination according to the initial minimum value, the initial maximum value and the initial intermediate value; S300, generating a plurality of simulation netlists according to the simulation case combination and respectively performing SPICE simulation to obtain corresponding eye diagram performance response values; S400, establishing a second-order polynomial meta-model according to the eye diagram performance response values and the circuit parameter; S500, calculating an eye diagram performance response maximum value and a corresponding circuit parameter value according to the second-order polynomial meta-model; S600, presetting a scaling coefficient and updating a search space and maximum, minimum and intermediate values of the search space according to the eye diagram performance response maximum value and the corresponding circuit parameter value; S700, updating a simulation case combination according to the maximum, minimum and intermediate values of the updated search space, repeating steps S300 to S600, performing iterative calculation of the search space, and outputting an optimal search space until a preset iteration termination condition is reached.

2. The eye diagram search space optimization method of claim 1, wherein, After the second-order polynomial meta-model is established, the method further comprises: When a difference between the eye diagram performance response maximum value calculated according to the second-order polynomial meta-model and an actual simulation value is within an error range and an adjusted determination coefficient of the second-order polynomial meta-model is greater than a preset threshold, it is determined that the second-order polynomial meta-model is reliable, otherwise it is determined that the second-order polynomial meta-model is unreliable.

3. The eye diagram search space optimization method of claim 2, wherein, The method further comprises: When it is determined that the second-order polynomial meta-model is reliable, a willingness curve is adjusted, a series of willingness curves are obtained by adjusting model coefficients based on the eye diagram performance response maximum value; If a higher eye diagram performance response maximum value is found, the eye diagram performance response maximum value is updated, otherwise the eye diagram performance response maximum value is maintained unchanged.

4. The eye diagram search space optimization method of claim 2, wherein, The method further comprises: When it is determined that the second-order polynomial meta-model is unreliable, a better one of an actual simulation value in the current iteration and an actual simulation value in the last iteration is selected as the eye diagram performance response maximum value.

5. The eye diagram search space optimization method of claim 1, wherein, The updating of the search space and the maximum, minimum and intermediate values of the search space comprises: a0 is the initial minimum value, c0 is the initial maximum value, a is the minimum value of the search space in the last iteration, c is the maximum value of the search space in the last iteration, b is a point in the interval [a, c], x is the eye diagram performance response maximum value in the current iteration, t = b-a, alpha is the preset scaling coefficient, aNew is the minimum value of the updated search space in the current iteration, cNew is the maximum value of the updated search space in the current iteration, and bNew is the intermediate value of the updated search space in the current iteration. When judging that x is in the interval [a, c], the current x value is reserved; when judging that x is not in the interval [a, c], x is set to the boundary value closer to the boundary, that is, x = max (a, min (x, c)); When judging and are both in the interval [a, c], then , bNew = x, ; When judging , x-t<=a0, and , aNew=a0, , ; When judging , x-t<=a0, and , aNew=a0, bNew=x, ; In the determination , when x-t > a0, aNew = x-t, bNew = x, cNew = x+t; When judging , x+t >= c0, and , , , cNew=c0; When judging , x+t>=c0, and , , bNew=x, cNew=c0; In the determination , and x + t < c0, aNew = x - t, bNew = x, cNew = x + t.

6. The eye diagram search space optimization method of claim 1, wherein, The eye diagram performance response value includes eye height data or eye width data.

7. The eye diagram search space optimization method of claim 1, wherein, The second-order polynomial meta-model is built, specifically including: In the first iteration, the least square method is used to build the second-order polynomial meta-model; In the non-first iteration, the stepwise regression method is used to build the second-order polynomial meta-model.

8. The eye diagram search space optimization method of claim 1, wherein, The preset iteration termination condition includes a preset iteration number.

9. An eye diagram search space optimization system, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the eye diagram search space optimization method in any one of claims 1-8.

10. A computer storage medium having stored thereon a computer program or instructions, characterized in that, The computer program or instructions are executed by the processor to implement the steps of the eye diagram search space optimization method in any one of claims 1-8.

Images