Standard cell pin accessibility analysis method, electronic device, and medium

Abstract

The present application relates to the technical field of integrated circuit design automation, and particularly relates to a standard cell pin accessibility analysis method, an electronic device and a medium, the method comprising: constructing one or more unique cell instances for a standard cell, generating an access mode for each unique cell instance; obtaining static feature information of each access mode and generating a static quantization value; listening to a wire disconnection and redistribution event of a routing tool, obtaining dynamic monitoring data and generating a dynamic quantization value; for each unique cell instance, generating a comprehensive quantization value of the unique cell instance based on the static quantization value and the dynamic quantization value of all access modes of the unique cell instance, and combining access mode set feature information; for each standard cell, generating a comprehensive quantization value of the standard cell based on the comprehensive quantization values of all unique cell instances of the standard cell; and outputting at least one level of quantization value for pin access optimization. The present application improves the accuracy and automation degree of pin accessibility evaluation.

Description

Technical Field

[0001] This invention relates to the field of integrated circuit design automation technology, and more particularly to a method for analyzing the accessibility of standard cell pins, electronic devices, and media. Background Technology

[0002] As integrated circuit manufacturing processes continue to shrink to advanced nodes, the layout density of standard cells is constantly increasing, design rules are becoming more complex, and routing resources are becoming increasingly scarce. Pin accessibility of standard cells—the ease with which routing tools can successfully connect the cell's input and output ports while meeting design constraints—has become a key factor restricting the successful convergence of chip physical implementation. In the detailed routing stage, the location of pin access points, their accessible directions, and their compatibility with neighboring cells directly affect whether the router can complete interconnections without design rule violations within limited routing resources. However, due to the lack of a systematic quantitative evaluation standard for pin accessibility, current physical design flows still primarily rely on engineers making qualitative judgments after placement and routing by analyzing Electronic Design Automation (EDA) reports.

[0003] This retrospective evaluation method has two drawbacks: First, the evaluation results lag behind design decisions, making it difficult to guide routers in making optimal decisions during pin access mode selection. Second, the evaluation process relies on human experience, with different engineers having different judgment standards, and it is difficult to automate and reuse the results across different designs, thus compromising the objectivity and accuracy of the evaluation. Existing research has improved pin accessibility by optimizing standard cell layouts or using machine learning methods to predict design rule violations. However, these methods are mostly based on static geometric features or global routing congestion information, failing to reflect the dynamic performance of access modes in real-world, detailed routing environments, such as teardown and rewiring events and neighborhood interference. Furthermore, they have not established a multi-level quantitative evaluation system from access modes to standard cells, resulting in insufficient accuracy and automation in pin accessibility evaluation. Therefore, improving the accuracy and automation of pin accessibility evaluation has become an urgent technical problem to be solved. Summary of the Invention

[0004] The purpose of this invention is to provide a method, electronic device, and medium for analyzing the accessibility of standard cell pins, thereby improving the accuracy and automation of pin accessibility assessment.

[0005] According to a first aspect of the present invention, a method for analyzing the accessibility of standard cell pins is provided, comprising:

[0006] Step S1: Construct one or more unique cell instances for a standard cell. The unique cell instance is determined by the standard cell corresponding to the unique cell instance, the placement direction of the standard cell in the layout, and the track offset. Generate one or more access modes that meet the design rule constraints for each unique cell instance.

[0007] Step S2: Obtain the static feature information of each access mode, and generate a static quantization value for each access mode based on the static feature information;

[0008] Step S3: During the circuit design and implementation process, monitor the wiring tool's disconnection and rewiring events, obtain dynamic monitoring data for each access mode, and generate dynamic quantization values ​​for each access mode based on the dynamic monitoring data.

[0009] Step S4: For each unique unit instance, based on the static and dynamic quantization values ​​of all access modes corresponding to the unique unit instance, and combined with the access mode set feature information of the unique unit instance, generate the comprehensive quantization value of the unique unit instance; for each standard unit, based on the comprehensive quantization values ​​of all unique unit instances corresponding to the standard unit, generate the comprehensive quantization value of the standard unit.

[0010] Step S5: Output at least one level of quantization value for pin access optimization.

[0011] According to a second aspect of the present invention, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being configured to perform the method described in the first aspect of the present invention.

[0012] According to a third aspect of the present invention, a computer-readable storage medium is provided, storing computer-executable instructions for performing the method described in the first aspect of the present invention.

[0013] Compared with existing technologies, this invention has significant advantages and beneficial effects. Through the above technical solution, the standard cell pin accessibility analysis method, electronic device, and medium provided by this invention achieve considerable technological advancement and practicality, and have broad industrial application value. It has at least the following beneficial effects:

[0014] The method described in this invention constructs unique cell instances for standard cells and generates access patterns. It combines the static and dynamic quantization values ​​of the access patterns to perform three-level comprehensive quantization and outputs at least one level of quantization value for pin access optimization. This transforms the traditional ex-post qualitative judgment that relies on human experience into automated pre-emptive quantization analysis, effectively improving the accuracy and automation of pin accessibility assessment. Furthermore, the quantization results can guide the selection of access patterns for routing tools, thereby reducing the number of detailed routing iterations, reducing bus length and via quantity, and accelerating design convergence. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 Flowchart of a standard cell pin accessibility analysis method provided for embodiments of the present invention;

[0017] Figure 2 This is an interactive diagram illustrating the standard unit pin connectivity evaluation and optimization framework and detailed routing process provided in the embodiments of the present invention.

[0018] Figure 3 This is a schematic diagram of the electronic device structure provided in an embodiment of the present invention;

[0019] Figure 4 This is a schematic diagram of a computer-readable storage medium structure provided for an embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] This invention provides a method for analyzing the accessibility of standard cell pins, such as... Figure 1 As shown, it includes the following steps:

[0022] Step S1: Construct one or more unique cell instances for a standard cell. The unique cell instance is determined by the standard cell corresponding to the unique cell instance, the placement direction of the standard cell in the layout, and the track offset. Generate one or more access modes that meet the design rule constraints for each unique cell instance.

[0023] It should be noted that unique cell instances are a standard cell instance classification method for efficiently analyzing pin accessibility. Two cell instances are considered to be the same unique cell instance if they have identical characteristics in the following three aspects: the standard cell corresponding to the cell instance, the placement orientation of the cell instance, and the track offset of the cell instance relative to the routing track. The placement orientation of the cell instance includes rotated and flipped states. Using this classification method, embodiments of the present invention only need to analyze each type of unique cell instance to cover all instances of the same type, thereby significantly reducing computational complexity. Access patterns refer to the combination schemes formed by selecting an access point for each pin of a unique cell instance. This scheme must ensure that all selected access points satisfy design rule constraints and are compatible with each other, allowing routing tools to connect them simultaneously. A unique cell instance may have multiple feasible access patterns.

[0024] As one embodiment, the placement direction refers to the rotation or flipping state of the cell instance, and the track offset is the positional offset of the cell instance relative to the routing track. It should be noted that the placement direction refers to the rotation or flipping state of the cell instance in the chip layout, such as 0 degrees, 90 degrees, 180 degrees, 270 degrees rotation, or horizontal flipping, vertical flipping, etc. Different placement directions will affect the absolute coordinate position of the pin access point in the layout. The track offset refers to the positional offset of the cell instance relative to the routing track, that is, the offset distance between the boundary of the cell instance and the standard routing track grid. These two features, together with the standard cell, constitute the three core features that distinguish different unique cell instances.

[0025] Step S2: Obtain the static feature information of each access mode, and generate a static quantization value for each access mode based on the static feature information.

[0026] It should be noted that static feature information refers to features directly extracted from the layout geometry data of standard units, independent of the specific cabling environment. These features include cabling track alignment, access freedom, distribution concentration, and cabling resource consumption. Static features are used to reflect the inherent quality of the access mode at the physical structure level.

[0027] Step S3: During the circuit design and implementation process, monitor the wiring tool's disconnection and rewiring events, obtain dynamic monitoring data for each access mode, and generate dynamic quantization values ​​for each access mode based on the dynamic monitoring data.

[0028] It should be noted that dynamic monitoring data refers to data collected in real time during the detailed cabling process by monitoring the teardown and rerouting events of the cabling tool. This data includes selection preference, direct conflict level, and neighborhood interference level. Teardown and rerouting is an iterative strategy in cabling algorithms used to resolve cabling resource conflicts. When a legal path cannot be found for a certain network, the cabling tool will temporarily remove some of the already laid signal lines that are causing the blockage, and then replan the path. The number of teardown and rerouting events triggered by a specific access mode directly reflects the connection difficulty in the actual cabling environment.

[0029] Step S4: For each unique unit instance, based on the static and dynamic quantization values ​​of all access modes corresponding to the unique unit instance, and combined with the access mode set feature information of the unique unit instance, generate the comprehensive quantization value of the unique unit instance; for each standard unit, based on the comprehensive quantization values ​​of all unique unit instances corresponding to the standard unit, generate the comprehensive quantization value of the standard unit.

[0030] It should be noted that the access mode set feature information refers to the features extracted from the overall perspective of the entire access mode set of a unique instance, including access mode sufficiency, access mode difference, and access point coverage, which are used to evaluate the completeness, diversity, and coverage of the unique instance in pin access schemes.

[0031] Step S5: Output at least one level of quantization value for pin access optimization.

[0032] It should be noted that at least one level of quantization value refers to the ability to selectively output one or more of the quantization values ​​at the access mode level, unique unit instance level, or standard unit level, depending on the needs of the actual application scenario, thereby providing a flexible application interface.

[0033] The method described in this invention constructs a classification and generation mechanism for unique unit instances and access modes, and establishes a complete analysis process from static feature extraction, dynamic monitoring, three-level comprehensive quantization to output feedback. This transforms the traditional ex-post qualitative judgment that relies on human experience into automated pre-quantitative analysis, effectively improving the accuracy and automation of pin accessibility assessment. At the same time, by outputting quantization values ​​at each level, it can provide data support for the access mode selection of wiring tools, the standard unit selection of EDA tools, and the optimization and development of standard unit libraries.

[0034] As one embodiment, the static feature information includes cabling track alignment, access freedom, distribution concentration, and cabling resource consumption. Step S2 includes:

[0035] Step S21: For each access point in the access mode, determine the alignment weight value of the access point based on the alignment degree of the physical coordinates of the access point relative to the corresponding wiring track of the metal layer and the upper metal layer, and normalize it. Take the arithmetic mean of the normalized alignment weight values ​​of each access point as the wiring track alignment degree of the access mode. The higher the wiring track alignment degree, the worse the overall alignment degree between the access mode and the wiring track.

[0036] It should be noted that cabling track alignment is used to assess whether the center of the access point is aligned with the cabling track on the metal layer. Specifically, for a given access point, based on its physical coordinates... Relative to the metal layer and its upper metal layer The alignment weight value of the access point is determined based on the alignment degree of the corresponding wiring track. There are four alignment types: fully aligned track, partially aligned track, shape center point, and other irregular positions. The alignment weight value for a fully aligned track is... The alignment weight value of the half-aligned track is The alignment weight of the shape center point is The alignment weight value for other irregular positions is... The alignment weights of the four elements increase sequentially, i.e. Since different access points may be located on different metal layers, their alignment weight values ​​will have different dimensions. Therefore, it is necessary to normalize the alignment weight values ​​of each access point to map the scores to a uniform scale. The purpose of this normalization is to eliminate the dimensional differences caused by the different metal layers on which the access points are located, so that the scores of each access point can be compared and averaged under the same standard, ensuring the fairness and rationality of the evaluation.

[0037] Alignment weight value of a single access point The calculation formula is:

[0038]

[0039] It is understandable that the alignment weight value of a single access point needs to be superimposed on the access point's position on its respective metal layer. and the upper metal layer The alignment state, where This is the upper-layer penalty amplification factor, used to emphasize the higher requirements of via access quality for upper-layer track alignment. The arithmetic mean of the normalized alignment weights of all access points corresponding to the access mode is used as the wiring track alignment degree for that access mode. .

[0040] Step S22: For each access point in the access mode, set a basic weight and normalize it. Subtract the corresponding incremental coefficient from the basic weight according to the number of incoming directions supported by the access point. Take the arithmetic mean of the scores of each access point after normalization as the access degree of freedom of the access mode. The higher the access degree of freedom, the fewer the access directions that the access mode can access.

[0041] It should be noted that the access degree of freedom is used to quantify the diversity of incoming line directions allowed by the access point, directly reflecting the spatial flexibility of the pin access path. Specifically, an initial basic weight is set. , representing the worst-case theoretical state of the access point when it is restricted in all directions. Let the set of legally accessible directions for this access point be . These correspond to the four planar directions of east, south, west, and north, as well as the direction of the overhead through-hole access. If the access point supports overhead through-hole access, the through-hole increment coefficient is subtracted from the basic weight. For each supported planar direction, the corresponding planar increment coefficient is subtracted. Since different access points may be located in different metal layers or have different geometric features, their basic weights and score dimensions differ, requiring normalization of the scores for each access point. The purpose of this normalization is the same as in step S21: to eliminate the dimensional differences between different access points, allowing the scores of each access point to be compared and averaged on a uniform scale. (Single access point) Access freedom score The calculation formula is:

[0042]

[0043] The arithmetic mean of the normalized scores of all access points corresponding to the access mode is taken as the access degree of freedom of that access mode. The more directions supported, the lower the score, indicating better access freedom.

[0044] Step S23: Calculate the centroid coordinates of all access points in the access mode and the mean square distance of each access point relative to the centroid coordinates. After mapping to a preset interval through a preset monotonically decreasing function, the distribution concentration of the access mode is obtained. The higher the distribution concentration of the access mode, the higher the degree of aggregation of its access points in the physical space.

[0045] Distribution concentration is used to evaluate the physical spatial dispersion of all access points within an access mode, i.e., the geometric distribution characteristics of the access point set. If the access points of each pin are clustered in the same small area, multiple leads will compete for cabling resources within the limited space, exacerbating local congestion. Let the set of access points for access mode t be... , This represents the number of access points in the set. To determine the coordinates of a single access point, first calculate the centroid coordinates of all access points in the access mode, which is the arithmetic mean of the coordinates of all access points. :

[0046]

[0047] Then calculate the mean square distance of each access point relative to the centroid coordinates. :

[0048]

[0049] A smaller mean square distance indicates that the access points are more concentrated near the centroid, resulting in higher spatial clustering and a greater potential risk of congestion. To transform the mean square distance into an easily quantifiable evaluation metric, a monotonically decreasing function is used. After mapping to a preset interval, the distribution concentration is obtained, where For adjustment coefficients, >0 indicates a higher degree of clustering.

[0050] Step S24: Count the total number of cabling tracks covered or blocked by the access mode, divide it by the total number of available cabling tracks within the standard unit boundary, and obtain the cabling resource consumption of the access mode. The higher the cabling resource consumption, the more cabling resources the access mode consumes.

[0051] It should be noted that the cabling resource consumption rate is used to count the number of cabling tracks occupied by an access mode, reflecting the cabling resources consumed. Specifically, it counts the total number of horizontal and vertical cabling tracks covered or blocked by the access mode. Divide by the total number of available wiring tracks within the boundary of the standard unit. Obtain cabling resource consumption The higher the value, the more cabling resources this access mode consumes.

[0052] Step S25: Calculate the static quantization value of access mode t using the following weighted summation formula. :

[0053]

[0054] in, For wiring track alignment, To access degrees of freedom, For distribution concentration, For cabling resource consumption, , , , They are respectively , , , The corresponding weighting coefficients, A higher value indicates poorer static pin accessibility for that access mode.

[0055] It should be noted that, based on the above four static characteristics, the static quantization value of access mode t is calculated using a linear weighted summation method. The weighting coefficients of each feature can be adjusted according to the actual application requirements.

[0056] Steps S21-S25 define four types of computable static quantification indicators to transform the physical geometric characteristics of the standard unit pin access mode into quantifiable values. This makes the evaluation of the access mode no longer dependent on human experience, but can be automatically calculated through deterministic calculation methods and formulas, providing an objective and repeatable static evaluation basis for subsequent comprehensive quantification.

[0057] As one embodiment, the dynamic monitoring data includes selection preference, direct conflict, and neighborhood interference, and step S3 includes:

[0058] Step S31: Obtain the number of times the access mode is selected as the selection preference. The higher the selection preference, the more times the cabling tool selects the access mode in the global access mode selection process.

[0059] It should be noted that the preference level is used to quantify the degree of preference of the access mode in the global access mode selection phase by the detailed cabling unit. In the global access mode selection phase, each standard cell instance is assigned an access mode. Let I be the set of all standard cell instances, and each instance i∈I is assigned an access mode t(i) in the global access mode selection phase. This is achieved by counting the total number of times each access mode is assigned. ,in This is an indicator function that takes the value when the condition within the parentheses is true. Otherwise This will give you the selection preference level. The higher the value, the more times the cabling tool selects that access mode in the global access mode selection process.

[0060] Step S32: Obtain the number of disconnection events caused by the standard unit instance corresponding to the access mode as the direct conflict degree. The higher the direct conflict degree, the more frequently the access mode will cause design rule violations in actual cabling.

[0061] It should be noted that direct conflict degree is used to quantify the severity of design rule violations directly caused by the access mode itself, leading to teardown and rerouting. During detailed routing, when a routing path of a net causes a violation that cannot be resolved through local patching, the router will tear down the net and its conflicting nets and re-search for paths. Let E be the set of all teardown events occurring during the detailed routing iteration. For event e ∈ E, if the event is caused by a design rule violation triggered by the pin access path of instance i, then event e is said to be attributed to instance i. Let the set of events attributed to instance i be denoted as . Let I be the set of all standard unit instances. Each instance i∈I is assigned an access mode t(i) during the global access mode selection phase. For indicator functions. Direct conflict degree of access mode t. Defined as the sum of the number of disconnection events triggered by all standard unit instances corresponding to this access mode, i.e. Attribution to instance i means that the teardown event is directly caused by a design rule violation triggered by the pin access path of that instance, rather than indirectly caused by neighboring cells or global routing congestion. By listening to teardown and rerouting events, violations directly related to the pins of standard cell instances are identified, and the number of teardown events for the corresponding access mode is accumulated to obtain the direct conflict degree. The higher the value, the more frequently the access mode causes design rule violations in actual routing.

[0062] Step S33: Obtain the number of disconnection events not caused by the standard unit instance itself within the neighborhood window of the standard unit instance corresponding to the access mode as the neighborhood interference degree. The higher the neighborhood interference degree, the more serious the interference of the access mode on the neighborhood cabling environment.

[0063] It should be noted that neighborhood interference is used to quantify the indirect impact of access patterns on the local cabling environment, reflecting the degree of spatial interference between the access pattern and neighboring units. Even if the access pattern itself does not directly cause a design rule violation, the spatial extension of its pin access path may occupy surrounding cabling tracks, forcing the access paths of adjacent unit instances to detour or causing congestion, thereby indirectly causing teardown and rewiring events within the neighborhood. For any standard unit instance i∈I, let its bounding box be... ,in , These are the width and height of the instance, respectively. Its neighborhood window is defined as... ,in, >0、 , Used to control the expansion range of the neighborhood window in the horizontal and vertical directions. For a line break event e∈E, let its violation center position be... ,like If event e is called the neighborhood teardown event of instance i, then event e is called the neighborhood teardown event of instance i.

[0064] To avoid duplicate calculations for direct conflict events, the neighborhood interference score only counts disconnection events not caused by the instance itself. For each instance using this access mode, the number of disconnection events not caused by the instance itself within its neighborhood window is counted. The statistical results of all instances are summed to obtain the neighborhood interference score for this access mode. The higher this value, the more significant the congestion interference caused by the pin access path of this access mode to the surrounding cabling environment, which may be the cause of local cabling deterioration.

[0065] Step S34: Calculate the dynamic raw quantization value of access mode t using the following formula. :

[0066]

[0067] in, To select preference, For direct conflict level, For neighborhood interference, , , They are respectively , , The corresponding weighting coefficients, A higher value indicates that the access mode is associated with more conflict events per unit of selection.

[0068] It should be noted that, based on the three dynamic characteristics mentioned above, the dynamic raw quantification value of the access mode is defined as the weighted sum of direct conflict degree and neighborhood interference degree divided by the selection preference degree. Under a unit of selection frequency, the more conflict events associated with the access mode, the higher its dynamic raw quantification value, indicating that the access mode performs worse in a real cabling environment. The selection preference degree appears in the denominator to eliminate the impact of differences in the number of selections on the evaluation.

[0069] Step S35: Calculate the dynamic quantization value of access mode t using the following normalization formula. :

[0070]

[0071] in, This is the slope adjustment coefficient. exp is an exponential function with base e. A higher value indicates that the access mode is associated with more conflict events per unit of selection.

[0072] It should be noted that, due to the large differences in the values ​​of the dynamic raw quantification values, the Sigmoid function is used to normalize them in order to eliminate the influence of the units and limit the scoring range. and Maintaining a positive correlation, meaning that the higher the original dynamic quantization value, the higher the normalized dynamic quantization value, ensures consistency in the evaluation direction. This normalization function maps the original dynamic quantization value to the [0,1] interval and maintains monotonicity, allowing dynamic quantization values ​​of different access modes to be compared on a unified scale. A higher value indicates that the access mode is associated with more conflict events per unit of selection.

[0073] Steps S31-S35, by real-time monitoring of disconnection and rewiring events during the detailed cabling process, transform the dynamic performance of the access mode in the real cabling environment into quantifiable values. This compensates for the shortcomings of static features in failing to perceive dynamic factors such as local congestion and neighborhood interference, making the evaluation results closer to the convergence of the actual design. Simultaneously, the introduction of selection preference in the denominator allows the evaluation results to eliminate bias caused by differences in the number of selections.

[0074] As one embodiment, in step S4, the access mode set feature information includes access mode sufficiency, access mode difference, and access point coverage. The comprehensive quantization value of the unique unit instance is calculated in the following manner:

[0075] Step S41: Determine the access mode sufficiency based on the total number of access modes of the unique unit instance and a preset threshold. The higher the access mode sufficiency, the less available access schemes are available for the unique unit instance.

[0076] It should be noted that the access mode sufficiency is used to measure the completeness of the number of pin access schemes for a unique instance. Specifically, let the set of legal access modes for a unique instance u be... Its size This determines the optimization space in the subsequent global access mode selection phase. If If the value is too small, the detailed cabling unit will lack sufficient alternatives when faced with conflicts between adjacent units or localized congestion. Access mode adequacy. ,in This is the threshold for the minimum desired number of access modes. When ≥ hour, =0 indicates that the number of access modes has reached the desired level, and no further penalty is imposed on the number of schemes; when When =0, =1 indicates a complete lack of legitimate access schemes. Therefore, the fewer the number of access modes, the higher the access mode abundance, and the less available access schemes.

[0077] Step S42: Determine the corresponding access mode difference degree based on the similarity of the access point sets between all access modes of the unique unit instance. The higher the access mode difference degree, the smaller the structural difference between the access modes of the unique unit instance.

[0078] It should be noted that the access mode difference is used to quantify the degree of structural overlap between access modes. If multiple access modes within a set highly overlap in their access point selection, their performance in actual cabling will tend to be homogeneous, failing to provide true routing flexibility for detailed cabling systems. Specifically, the average Jaccard similarity coefficient between sets is used to characterize the degree of homogeneity, and its calculation formula is: when <2 o'clock, =1; when When ≥2, = The higher the degree of difference in access modes, the more likely different access modes are to call the same combination of access points, and the smaller the structural differences between modes.

[0079] Step S43: Determine the corresponding access point coverage based on the proportion of the union of access points of all access modes of the unique unit instance to the total number of legal access points of the unique unit instance. The higher the access point coverage, the higher the proportion of legal access points of the unique unit instance that are idle.

[0080] It should be noted that access point coverage is used to evaluate the utilization of the current unique instance's set of access patterns across all potential legitimate access points. Specifically, let the complete set of all legitimate access points for a unique instance u in the current layout environment be denoted as . Access point coverage This refers to the percentage of legitimate access points not included in any access mode. This occurs when all legitimate access points are utilized by at least one access mode. The higher this value, the higher the proportion of legitimate access points that are idle, limiting the possibility of the cabling unit entering from more tracks or directions.

[0081] Step S44: Calculate the comprehensive quantization value S(u) of the unique cell instance u using the following formula:

[0082] ;

[0083] in, It is the set of the top K access modes with the best overall performance in a unique unit instance u. This is the static quantization value for access mode t. This is the dynamic quantization value for access mode t. To ensure sufficient access mode, To differentiate access modes, For access point coverage, , , They are respectively , , The corresponding weighting coefficient, K, is a preset positive integer.

[0084] It should be noted that at the unique instance level, not only the individual quality of its candidate access modes needs to be considered, but also the three set characteristics of the unique instance itself need to be superimposed. To avoid poorly performing access modes lowering the overall evaluation of the unique instance, only the top K access modes with the best overall performance are selected during aggregation. This means that the overall score... + The lower the value, the better the performance. The average comprehensive score of the top K performing access modes is calculated, and then a weighted sum of access mode sufficiency, access mode differentiation, and access point coverage is added to obtain the comprehensive quantitative value of the unique instance. Here, K is a preset positive integer, its purpose being to prevent poorly performing access modes from lowering the overall evaluation of the unique instance, ensuring that the selected top K optimal modes represent the best access capability of that unique instance.

[0085] Steps S41-S44 introduce unique instance-level evaluation metrics, expanding the evaluation dimension from a single access mode to a holistic perspective of the access mode set. This not only focuses on the quality of the optimal access mode but also on whether the available access solutions are sufficient, diverse, and whether they fully utilize legitimate access points, making the evaluation more comprehensive and systematic.

[0086] As one embodiment, in step S4, the comprehensive quantization value of the standard unit Calculate as follows:

[0087] ;

[0088] in, S(u) is the set of all unique cell instances corresponding to the standard cell c, and S(u) is the comprehensive quantization value of the unique cell instance u.

[0089] It should be noted that since the same standard cell can generate multiple unique instances under different placement orientations and track offsets, its overall pin accessibility should be reflected by the combined performance of all unique instances. The standard cell score can be used to guide EDA tools in cell selection during the design optimization phase, and can also be used to indicate whether to use standard cells with caution or prohibit their use. Furthermore, it can provide optimization feedback for the development of the standard cell library. By defining the comprehensive quantitative value of a standard cell as the arithmetic mean of the scores of all its unique instances, a complete closed loop of a three-tiered evaluation system—from access mode to unique cell instances to standard cells—is achieved. This allows the evaluation results to serve different decision-making scenarios at different levels: the access mode layer serves the access mode selection of cabling tools; the unique instance layer serves the overall solution evaluation of that instance; and the standard cell layer serves the cell selection of EDA tools and the optimized development of the standard cell library.

[0090] As one embodiment, the method further includes a pin access optimization step, specifically including:

[0091] Step S71: Obtain historical evaluation reports. The historical evaluation reports include at least the historical comprehensive quantitative values ​​of the access mode and the corresponding process library metadata. Verify whether the historical evaluation reports match the metadata of the current process library. If they match, the historical evaluation reports can be reused.

[0092] It should be noted that the pin access optimization algorithm takes the current routing database and historical evaluation reports as input, converts historical scores into prior penalty costs, optimizes the global access mode selection, and outputs an access mode allocation scheme and an update report. The historical evaluation reports are evaluation results accumulated from previous circuit designs. These reports contain process library metadata, and this verification is used to confirm compatibility, evaluation parameters, and scoring-related data. When verifying historical reports, it is necessary to check whether their process library metadata completely matches the current process library. Historical data is only allowed to be reused when the metadata completely matches, ensuring the validity of historical scores.

[0093] Step S72: For each access mode, if a matching historical integrated quantization value exists, the historical integrated quantization value and historical static quantization value of the access mode are directly reused; otherwise, the static quantization value of the access mode is calculated.

[0094] It should be noted that for each access mode, if a matching historical comprehensive quantitative value exists in the historical evaluation report, the historical comprehensive quantitative value and historical static quantitative value of that access mode can be directly reused without recalculation, thus saving computing resources; that is, both the historical static score and the comprehensive score can be reused. If no matching value exists, the static quantitative value of that access mode is calculated according to steps S21-S25.

[0095] Step S73: Calculate the prior cost of each access mode based on the historical comprehensive quantification value. The higher the prior cost, the worse the historical accessibility of the access mode.

[0096] It should be noted that the prior cost of each access mode is calculated based on historical comprehensive quantitative values. Where λ is the weighting coefficient, λ>0. Historical comprehensive quantitative value The larger the value, the greater the difficulty of establishing the connection in the historical design of this access mode, and the higher the corresponding prior cost.

[0097] Step S74: Take the prior cost as part of the optimization objective, and combine it with the constraint that no design rule violation occurs between the access modes of adjacent unit instances, input it into the global access mode selection stage of the cabling tool, so that the cabling tool selects the access mode according to the prior cost.

[0098] It should be noted that prior costs are included as part of the optimization objective, and the constraint that no design rule violations occur between the access modes of adjacent unit instances is also considered in the global access mode selection process of the routing tool. Let I be the set of all standard unit instances in the current design. For each instance i, let... This indicates the access mode assigned to it; the access mode allocation scheme for all instances constitutes this. Therefore, the global access mode selection problem is transformed into a minimization problem: ,in, This is a conflict penalty term used to constrain adjacent instance access patterns from violating design rules at physical boundaries. If a conflict occurs, a significant penalty is imposed. In other words, solving this minimization problem allows routing tools to proactively avoid historically problematic access patterns that are prone to design rule violations, while ensuring physical boundary compatibility between adjacent instances.

[0099] Step S75: After the wiring converges, update the dynamic quantization value of each access mode, perform multi-level scoring aggregation, and output a pin access optimization update report.

[0100] It should be noted that after detailed routing convergence, the system monitors local routing events that actually trigger design rule violations, updates the dynamic quantification values ​​of the relevant access modes in real time, and then performs multi-level scoring aggregation from access mode to unique cell instances and then to standard cells. That is, it aggregates scores level by level according to the path of access mode, unique cell instance, and standard cell, and finally outputs a pin access optimization update report, providing reusable historical data for subsequent designs. This update report includes process library metadata, evaluation parameters, and scoring-related data, which can serve as input for subsequent designs, achieving a data-driven closed loop.

[0101] It should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of these steps can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the steps can be rearranged. A process can be terminated when its operation is complete, but it may also have additional steps not included in the figures. A process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0102] like Figure 2 As shown, Figure 2 The left side of the image illustrates the standard detailed routing implementation flow. The detailed router first needs to model the physical environment and logical connections, a process that involves parsing technology library files, design files, and routing guide files. Next, the detailed router builds a physical design database. Based on this structured database, the detailed routing process enters the data preparation phase. The main task of this phase is to prepare the necessary data for subsequent pin access analysis, track assignment, and maze routing, including routing guide processing, via sorting, and DRC lookup table construction.

[0103] After data preparation is complete, the detailed routing process enters the pin access analysis phase. The core objective of this phase is to pre-plan the pin access schemes for each standard cell instance without relying on specific net connections, thereby avoiding design rule violations caused by pin access conflicts during subsequent routing. This phase follows a solution process of "access point enumeration—access mode generation—global access mode selection".

[0104] During the access point enumeration phase, the detailed router, based on the geometry of each pin, its metal layer, and its position relative to the routing path, enumerates as many legal access points as possible while satisfying process rules. Each access point records detailed physical coordinates and accessible direction information. For a given unique instance, the detailed router combines candidate access points for each pin to filter out a set of access patterns that satisfy intra-cell compatibility constraints. During generation, it is essential to ensure that the selected access points and their access paths satisfy intra-cell compatibility constraints, meaning that there are no design rule violations in pin access within the same cell. At this point, if... Figure 2 As shown, the Evaluation and Optimization Framework (PAE) will intervene for the first time, extracting the geometric features of each generated access mode and calculating the static quantization value.

[0105] Subsequently, in the global access mode selection phase, the detailed router must select the final access mode from the corresponding access mode set for each standard cell instance, using its access point as the endpoint for the corresponding path search. This phase needs to resolve pin access conflicts between cells caused by compact layout. In this phase, the PAE introduces its calculated evaluation cost as a penalty term. This cost integrates the static characteristics of the access mode with historical routing dynamics, serving as heuristic information to guide the router to prioritize access modes with better expected accessibility. Simultaneously, the PAE records the number of times each access mode is selected.

[0106] During the core detailed routing phase, the detailed router uses a maze search algorithm, with pin access points as endpoints, to sequentially search for interconnection paths for each net on the mesh graph. After each routing task, the detailed router performs a local design rule check. If a violation is found, it proceeds to the next iteration of teardown and rerouting; otherwise, the result is written back to the global database. Throughout the detailed routing process, PAE continuously monitors teardown and rerouting events, identifies pins that cause design rule violations, attributes them to the corresponding access modes, and records relevant data to calculate the dynamic quantization value of the access mode.

[0107] The method described in this embodiment of the invention can be implemented as a software system or a functional module integrated into existing EDA tools. The internal process of the PAE framework revolves around the calculation of static and dynamic evaluation of access modes, and its core modules and their interaction with the detailed cabling process are as follows: Figure 2 As shown on the right. PAE supports importing historical evaluation reports as an optional input to reuse previously accumulated evaluation data.

[0108] After the access mode is generated, the static feature extraction and quantization calculation module extracts four types of static features for each access mode t and calculates the static quantization value of the access mode. .

[0109] During the global access mode selection process, the PAE records the number of times each access mode is selected, and this data serves as one of the inputs for dynamic feature calculation. Simultaneously, during the detailed routing phase's teardown and rerouting iterations, the PAE continuously monitors teardown events, identifies pins causing design rule violations, and attributes them to the corresponding access mode. Based on this, the dynamic feature extraction and calculation module calculates the dynamic quantization value of the access mode. .

[0110] The access mode integrated quantization calculation module calculates static quantization values. With dynamic quantization value Summing is performed to obtain the comprehensive quantization value for each access mode. Subsequently, the unique instance feature extraction module extracts the access mode sufficiency, access mode difference, and access point coverage features for each unique instance. The unique instance and standard unit-level aggregation calculation module aggregates the access mode quantization values ​​hierarchically: first, it selects the top few access modes with the best overall performance within a unique instance, calculates the average of their quantization values, and then superimposes the three features of the unique instance to obtain the comprehensive quantization value S(u) of unique instance u; then, it takes the arithmetic mean of the quantization values ​​of all unique instances under the same standard unit to obtain the comprehensive quantization value S(c) of standard unit c. The above results are fed back to the pin access optimization module and output as an evaluation report.

[0111] After completing the aggregation calculation of multi-level quantization values, the pin access optimization module transforms the final quantization results of each level into heuristic penalty terms and feeds these penalty terms back to the global access mode selection stage of the detailed router. This allows the detailed router to prioritize access modes with better accessibility expectations when handling unique instances, thereby optimizing pin access decisions.

[0112] Through the above architecture design, PAE achieves bidirectional interaction with the detailed cabling process: it provides cost guidance for upstream access mode selection, monitors the downstream cabling execution process, and outputs the quantification results in the form of structured reports, which can support the optimization of standard unit design.

[0113] This invention constructs a three-tiered quantitative evaluation system, from access mode to unique cell instances to standard cells, transforming traditional qualitative judgments relying on human experience into automated pre-emptive quantitative analysis. This significantly improves the accuracy and automation of pin accessibility assessment. The method integrates the static geometric characteristics of access modes with dynamic routing performance, overcoming the limitations of existing methods in detecting local congestion and neighborhood interference, making the evaluation results closer to actual routing convergence. Simultaneously, by introducing a historical evaluation data reuse mechanism, the evaluation results accumulated in the early design phase are transformed into prior costs to guide subsequent access mode selection, achieving a positive evolution of continuous improvement without increasing algorithm complexity. This continuously improves routing convergence efficiency and result quality, provides objective quantitative feedback for standard cell library optimization, and offers quantifiable data support for collaborative optimization between design and manufacturing.

[0114] This invention also provides an electronic device, such as... Figure 3 As shown, it includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being configured to perform the method described in the embodiments of the present invention.

[0115] For example, the electronic device may be a mobile electronic device or a non-mobile electronic device. Further, a mobile electronic device may be any one of a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA); a non-mobile electronic device includes any one of a server, network attached storage (NAS), personal computer (PC), television (TV), ATM, or self-service machine.

[0116] This invention also provides a computer-readable storage medium, such as... Figure 4 As shown, computer-executable instructions are stored thereon, which are used to execute the methods described in the embodiments of the present invention.

[0117] Examples of computer-readable storage media include: computer read-only memory (ROM), random access memory (RAM), erasable programmable read-only memory (EPROM), compact disc read-only memory (CD-ROM), flash memory, magnetic disk, optical disk, portable computer disk, hard disk and / or solid-state drive, etc.

[0118] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for analyzing the accessibility of standard cell pins, characterized in that, Includes the following steps: Step S1: Construct one or more unique cell instances for a standard cell. The unique cell instance is determined by the standard cell corresponding to the unique cell instance, the placement direction of the standard cell in the layout, and the track offset. Generate one or more access modes that meet the design rule constraints for each unique cell instance. Step S2: Obtain the static feature information of each access mode, and generate a static quantization value for each access mode based on the static feature information; Step S3: During the circuit design and implementation process, monitor the wiring tool's disconnection and rewiring events, obtain dynamic monitoring data for each access mode, and generate dynamic quantization values ​​for each access mode based on the dynamic monitoring data. Step S4: For each unique unit instance, based on the static and dynamic quantization values ​​of all access modes corresponding to the unique unit instance, and combined with the access mode set feature information of the unique unit instance, generate the comprehensive quantization value of the unique unit instance; for each standard unit, based on the comprehensive quantization values ​​of all unique unit instances corresponding to the standard unit, generate the comprehensive quantization value of the standard unit. Step S5: Output at least one level of quantization value for pin access optimization.

2. The method according to claim 1, characterized in that, The static feature information includes cabling track alignment, access freedom, distribution concentration, and cabling resource consumption. Step S2 includes: Step S21: For each access point in the access mode, determine the alignment weight value of the access point based on the alignment degree of the physical coordinates of the access point relative to the corresponding wiring track of the metal layer and the upper metal layer, and normalize it. Take the arithmetic mean of the normalized alignment weight values ​​of each access point as the wiring track alignment degree of the access mode. The higher the wiring track alignment degree, the worse the overall alignment degree of the access mode with the wiring track. Step S22: For each access point in the access mode, set a basic weight and normalize it. Subtract the corresponding incremental coefficient from the basic weight according to the number of incoming directions supported by the access point. Take the arithmetic mean of the scores of each access point after normalization as the access degree of freedom of the access mode. The higher the access degree of freedom, the fewer the access directions that the access mode can access. Step S23: Calculate the centroid coordinates of all access points in the access mode and the mean square distance of each access point relative to the centroid coordinates. After mapping to a preset interval through a preset monotonically decreasing function, the distribution concentration of the access mode is obtained. The higher the distribution concentration of the access mode, the higher the degree of aggregation of its access points in the physical space. Step S24: Count the total number of cabling tracks covered or blocked by the access mode, divide it by the total number of available cabling tracks within the standard unit boundary, and obtain the cabling resource consumption of the access mode. The higher the cabling resource consumption, the more cabling resources the access mode consumes. Step S25: Calculate the static quantization value of access mode t using the following weighted summation formula. : ， in, For wiring track alignment, To access degrees of freedom, For distribution concentration, For cabling resource consumption, , , , They are respectively , , , The corresponding weighting coefficients, A higher value indicates poorer static pin accessibility for that access mode.

3. The method according to claim 1, characterized in that, The dynamic monitoring data includes selection preference, direct conflict, and neighborhood interference. Step S3 includes: Step S31: Obtain the number of times the access mode is selected as the selection preference. The higher the selection preference, the more times the access mode is selected by the cabling tool in the global access mode selection process. Step S32: Obtain the number of disconnection events caused by the standard unit instance corresponding to the access mode as the direct conflict degree. The higher the direct conflict degree, the more frequently the access mode will cause design rule violations in actual cabling. Step S33: Obtain the number of disconnection events not caused by the standard unit instance itself within the neighborhood window of the standard unit instance corresponding to the access mode as the neighborhood interference degree. The higher the neighborhood interference degree, the more serious the interference of the access mode on the neighborhood cabling environment. Step S34: Calculate the dynamic raw quantization value of access mode t using the following formula. : ， in, To select preference, For direct conflict level, For neighborhood interference, , , They are respectively , , The corresponding weighting coefficients, A higher value indicates that the access mode is associated with more conflict events per unit of selection. Step S35: Calculate the dynamic quantization value of access mode t using the following normalization formula. : ， in, This is the slope adjustment coefficient. exp is an exponential function with base e. A higher value indicates that the access mode is associated with more conflict events per unit of selection.

4. The method according to claim 1, characterized in that, In step S4, the access mode set feature information includes access mode sufficiency, access mode difference, and access point coverage. The comprehensive quantization value of the unique unit instance is calculated in the following manner: Step S41: Determine the access mode sufficiency based on the total number of access modes of the unique unit instance and a preset threshold. The higher the access mode sufficiency, the less available access schemes are available for the unique unit instance. Step S42: Determine the corresponding access mode difference degree based on the similarity of the access point sets between all access modes of the unique unit instance. The higher the access mode difference degree, the smaller the structural difference between the access modes of the unique unit instance. Step S43: Determine the corresponding access point coverage based on the proportion of the union of access points of all access modes of a unique unit instance to the total number of legal access points of that unique unit instance. The higher the access point coverage, the higher the proportion of legal access points of that unique unit instance that are idle. Step S44: Calculate the comprehensive quantization value S(u) of the unique cell instance u using the following formula: ， in, It is the set of the top K access modes with the best overall performance in a unique unit instance u. This is the static quantization value for access mode t. This is the dynamic quantization value for access mode t. To ensure sufficient access mode, To differentiate access modes, For access point coverage, , , They are respectively , , The corresponding weighting coefficient, K, is a preset positive integer.

5. The method according to claim 4, characterized in that, In step S41, the access mode sufficiency Calculate as follows: ， in, For unique unit instances The total number of access modes The preset threshold; In step S42, the access mode difference Calculate as follows: ， in, Let j be the set of access points for the j-th access mode. Let be the set of access points for the k-th access mode; In step S43, the access point coverage Calculate as follows: ， in, This is the set of all legal access points for a unique unit instance u.

6. The method according to claim 1, characterized in that, In step S4, the comprehensive quantization value of the standard unit is calculated in the following manner: ， in, S(u) is the set of all unique cell instances corresponding to the standard cell c, and S(u) is the comprehensive quantization value of the unique cell instance u.

7. The method according to claim 1, characterized in that, It also includes pin access optimization steps: Step S71: Obtain historical evaluation reports. The historical evaluation reports include at least the historical comprehensive quantitative values ​​of the access mode and the corresponding process library metadata. Verify whether the historical evaluation reports match the metadata of the current process library. If they match, the historical evaluation reports can be reused. Step S72: For each access mode, if a matching historical integrated quantization value exists, the historical integrated quantization value and historical static quantization value of the access mode are directly reused; otherwise, the static quantization value of the access mode is calculated. Step S73: Calculate the prior cost of each access mode based on the historical comprehensive quantification value. The higher the prior cost, the worse the historical accessibility of the access mode. Step S74: Take the prior cost as part of the optimization objective, and combine it with the constraint that no design rule violation occurs between the access modes of adjacent unit instances, input it into the global access mode selection stage of the cabling tool, so that the cabling tool selects the access mode according to the prior cost. Step S75: After the wiring convergence, update the dynamic quantization value of each access mode, perform multi-level scoring aggregation, and output a pin access optimization update report.

8. The method according to claim 1, characterized in that, The placement direction refers to the rotation or flipping state of the unit instance, and the track offset is the positional offset of the unit instance relative to the wiring track.

9. An electronic device, characterized in that, include: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions that are executed by the at least one processor, the instructions being configured to perform the method as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The device stores computer-executable instructions for performing the method as described in any one of claims 1-8.

Images