A negotiation-based advanced node transistor routing method, apparatus, device, storage medium, and program product

By employing a negotiated transistor routing method, the challenges of routing congestion and resource utilization in advanced nodes were resolved, enabling efficient and flexible routing path planning and improving routing quality and speed.

CN123117002APending Publication Date: 2026-11-17SOUTHEAST UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510580498.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-05-07
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

At advanced nodes, transistor wiring results are poor and wiring congestion is a prominent problem, making it difficult to effectively utilize metal resources in a limited design space and to meet strict design rules.

Method used

A negotiation-based transistor routing method is adopted, which achieves efficient routing path planning and optimization by means of techniques such as constructing routing mesh diagrams, merging pins, splitting nets, sorting routing order, congestion awareness, and cost function updates.

Benefits of technology

It improves the flexibility and quality of routing, significantly speeds up the solution process, adapts to the complex routing requirements of modern advanced process nodes, and reduces the bus length and running time of routing paths.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117002A_ABST
    Figure CN123117002A_ABST
Patent Text Reader

Abstract

This invention discloses a negotiation-based transistor routing method, apparatus, device, storage medium, and program product for advanced nodes. Based on the characteristics of the nets in the routing mesh, the routing order is arranged to obtain a sequence of net names representing the routing order. Following the net name sequence, two-pin pairs in each net are routed sequentially to obtain an initial routing result. The initial routing result is verified to ensure that the routing paths meet design rules. If all nets are uncongested, the final routing result is output. If any nets are congested, the congested paths in the initial routing result are removed based on the congestion information, and the cost function of the points on the paths is updated. This process is repeated until the final routing result is output. This invention effectively improves the flexibility and quality of routing while significantly accelerating the solution speed, adapting to the complex routing requirements of modern advanced process nodes.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to a negotiation-based advanced node transistor routing method, apparatus, device, storage medium, and program product, belonging to the field of circuit design automation technology. Background Technology

[0002] Standard cell layout design is a crucial part of the Very Large Scale Integration (VLSI) design flow, with transistor-level placement and routing forming its core steps. Among these, routing, due to its high complexity and resource consumption, is the most challenging task in layout design. The goal of routing is to interconnect transistor pins and extract the cell's input and output signals while adhering to strict design rules, providing the foundation for subsequent full-chip routing. The quality of routing directly affects the key performance indicators of standard cells, including latency and power consumption, thus profoundly impacting the overall chip design quality.

[0003] As semiconductor technology advances to 7nm and below nodes, advanced transistor architectures, such as FinFETs, have replaced traditional planar designs, enabling continuous performance improvements and energy efficiency enhancements. However, these technological advancements have also introduced greater design complexity, particularly during the routing phase of standard cell synthesis. The continuous shrinking of standard cell heights has led to a significant contraction in available routing resources, especially in the lower metal layers, where routing paths become more compact and pin access patterns become denser. In this resource-constrained environment, finer-grained routing path adjustments become crucial to avoid routing failures due to insufficient metal resources. Simultaneously, Design Rule Checks (DRCs) impose stringent constraints on routing, including minimum metal width, minimum spacing, and minimum area rules, which must be met to ensure correct circuit functionality and manufacturability.

[0004] To overcome these challenges, it is crucial to develop more granular and adaptable routing algorithms to enable effective exploration within constrained design spaces, provide greater flexibility in routing path selection, and promote more efficient use of limited metal resources. Summary of the Invention

[0005] Objective: To overcome the problems of poor transistor routing results and speed in the prior art, the present invention provides a negotiation-based transistor routing method, apparatus, device, storage medium and program product for advanced nodes, which not only helps to better comply with strict design rules, but also alleviates routing congestion problems in advanced process nodes.

[0006] Technical solution: To solve the above technical problems, the technical solution adopted by the present invention is as follows:

[0007] Firstly, a negotiation-based transistor routing method for advanced nodes specifically includes:

[0008] Step 101: Construct a wiring mesh diagram based on the circuit netlist and layout results.

[0009] Step 102: Based on the position of the multi-pin nets in the wiring mesh diagram, merge the pins directly connected on the M0 layer to obtain the mapping relationship between the net name and the pin coordinate set.

[0010] Step 103: In the mapping relationship between net name and pin coordinate set, nets with multiple pins are split into multiple two-pin pairs to obtain the mapping relationship between net name and two-pin pairs.

[0011] Step 104: Arrange the wiring order according to the properties of the nets in the wiring grid to obtain a sequence of net names representing the wiring order.

[0012] Step 105: Routing the two-pin pairs in each net in sequence according to the net name sequence to obtain the initial routing result.

[0013] Step 106: Verify whether the initial routing results meet the design rules. If all nets are uncongested, output the final routing results. If any nets are congested, proceed to step 107.

[0014] Step 107: Based on the congestion information, remove congested paths in the initial routing results and update the cost function for the points on the paths. Repeat steps 104 to 106 until the final routing result is output.

[0015] Optionally, the merging of pins directly connected on the M0 layer specifically includes:

[0016] Two active area pins belonging to the same wire network are adjacent to each other, and the two pins are directly merged.

[0017] Two gate pins belonging to the same wire mesh are merged if the distance between the two pins is 1 polysilicon pitch.

[0018] Two gate pins belonging to the same wire mesh are merged if the distance between the two pins is 2 or 3 polysilicon pitches and the area in between is a complete diffusion interruption.

[0019] Optionally, step 104 specifically includes:

[0020] Calculate the net ranking score based on the properties of the nets in the wiring mesh.

[0021] The nets are sorted from lowest to highest based on their net sorting scores, resulting in a sequence of net names representing the routing order.

[0022] Among them, the wire mesh sorting score The expression is as follows:

[0023]

[0024] in, Indicates online network The number of points on the wiring path, This indicates the number of congestion points encountered along the cabling path. Indicates wire mesh The length of the half-perimeter line, This represents the first weight parameter. This represents the second weighting parameter.

[0025] Optionally, step 105 specifically includes:

[0026] Step 501: Extract each net from the net name sequence and input each two-pin pair in the net into step 502.

[0027] Step 502, initialize the data structures: PossibleSet, SearchHeap, CostDict, and ParentDict. PossibleSet is a set of candidate points for building potential paths; SearchHeap is a heap for dynamically maintaining points sorted by cumulative cost; CostDict is a dictionary for storing the best known cumulative cost for each point; and ParentDict is a dictionary for storing the parent-child relationships required for backtracking paths.

[0028] Step 503: Initialize starts by inserting each grid point u in starts into SearchHeap with an initial cost of zero, and record it in CostDict. Here, starts refers to all grid points within one pin region or the other pin domain of the two pins.

[0029] Step 504: Take a grid point from starts and input it into step 505.

[0030] Step 505: Iterate through each adjacent point v of the input grid point u in the wiring mesh graph G. For each adjacent point v, calculate the cumulative cost from grid point u to grid point v. If grid point v has not been visited before, or if the cumulative cost to grid point v is found to be lower than the cumulative cost of grid point v in CostDict, then use the updated cumulative cost as the sorting index to insert grid point v into SearchHeap to obtain the updated SearchHeap. Update the cumulative cost of grid point v in CostDict to obtain the updated CostDict. Update the parent node of grid point v in ParentDict, and set grid point u as the parent node of grid point v to obtain the updated ParentDict.

[0031] Step 506: Retrieve the grid point u with the minimum cumulative cost from SearchHeap and check if grid point u belongs to the goals set. If not, input grid point u into step 505. Otherwise, call the path_backtrack(ParentDict, u) function to backtrack the wiring path, terminate the search, and return the wiring path R. Here, goals is the union of all points in the opposing region and their corresponding PossibleSet, and path_backtrack(ParentDict, u) represents the wiring path from grid point u backtracking to any grid point in starts according to the parent-child relationship in ParentDict.

[0032] Step 507: Repeat steps 505 to 506 until the routing path of the two-pin pair is successfully constructed, and obtain the routing result of the two-pin pair. Repeat steps 501 to 506 to obtain the routing result of all two-pin pairs corresponding to all nets as the initial routing result.

[0033] Optional, the cumulative cost of grid point v The expression is as follows:

[0034]

[0035] in: This represents the cumulative cost of parent node u. This represents the cost function based on congestion.

[0036] Optional, congestion-based cost function The expression is as follows:

[0037]

[0038] in: Represents the basic cost function. This represents the historical cost function value of network point v at the (k-1)th iteration. The variable k represents the current wiring iteration number. Indicates the degree of congestion.

[0039] Optionally, the update cost function The expression is as follows:

[0040]

[0041] in, It is a binary pointer function.

[0042] Secondly, a negotiation-based advanced node transistor wiring device specifically includes:

[0043] The first processing unit is used to construct a wiring mesh diagram based on the circuit netlist and layout results.

[0044] The second processing unit is used to merge the pins directly connected on the M0 layer according to the position of the multi-pin net in the wiring mesh diagram, and obtain the mapping relationship between the net name and the pin coordinate set.

[0045] The third processing unit is used to split the multi-pin nets in the mapping relationship between net names and pin coordinate sets into multiple two-pin pairs, and obtain the mapping relationship between net names and two-pin pairs.

[0046] The fourth processing unit is used to arrange the wiring order according to the attribute characteristics of the nets in the wiring mesh, and obtain a sequence of net names representing the wiring order.

[0047] The fifth processing unit is used to route the two-pin pairs in each net in sequence according to the net name sequence to obtain the initial routing result.

[0048] The sixth processing unit is used to verify whether the initial routing results meet the design rules. If all nets are uncongested, the final routing results are output. If any nets are congested, the seventh processing unit is used.

[0049] The seventh processing unit is used to tear down congested paths in the initial routing result based on congestion information, update the cost function for the points on the path, and repeat the fourth, fifth and sixth processing units until the final routing result is output.

[0050] Thirdly, a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a negotiation-based advanced node transistor routing method as described in any of the first aspects.

[0051] Fourthly, a computer device, which may include the aforementioned hardware verification device, specifically may include:

[0052] Memory is used to store instructions.

[0053] A processor for executing the instructions, causing the computer device to perform operations of a negotiation-based advanced node transistor routing method as described in any of the first aspects.

[0054] Fifthly, a program product, when executed by a processor, implements a negotiation-based advanced node transistor routing method as described in any of the first aspects.

[0055] Beneficial effects: The present invention provides a negotiation-based transistor routing method, apparatus, device, storage medium and program product for advanced nodes. The present invention can effectively improve the flexibility and quality of routing, while significantly accelerating the solution speed, and adapting to the complex routing needs under modern advanced process nodes. Attached Figure Description

[0056] Figure 1 This is a flowchart of a negotiation-based advanced node transistor routing method according to the present invention. Detailed Implementation

[0057] The technical solutions 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 protection scope of the present invention.

[0058] The present invention will be further described below with reference to specific embodiments.

[0059] Example 1:

[0060] This embodiment introduces a negotiation-based advanced node transistor routing method. First, a fine-grained routing path mesh is established based on the given circuit netlist and layout results using a dense mesh graph construction method to improve the flexibility of routing path selection. Next, a disjoint-set data structure-based pruning strategy is applied to merge pins that can be directly connected on the M0 layer, thereby effectively reducing the number of subsequent two-pin groups. Then, a minimum spanning tree (MST)-based network decomposition is performed, decomposing each multi-pin network into several two-pin groups. Next, a congestion-aware network prioritization strategy is adopted, prioritizing networks based on factors such as network congestion level and half-circuit length (HPWL). Subsequently, a congestion-driven 3D routing algorithm based on source region expansion is used to route each two-pin group. By leveraging the flexibility of 3D routing, fast and congestion-aware initial routing is achieved. In the routing exploration and path backtracking phases, this invention employs a boundary design rule checking technique with linear time complexity to efficiently detect congestion and verify whether the routing path meets the design rules. If a path conforms to the design rules, it is retained; otherwise, a negotiation-based path teardown and rerouting mechanism is activated, and the cost function is updated to guide subsequent path teardown and rerouting iterations until the design rules are met. Specific steps are as follows: Figure 1 As shown, the method includes the following steps:

[0061] Step 101: By considering the given circuit netlist and layout results, construct a refined routing mesh diagram to provide an accurate reference framework for routing and ensure that constraints can be effectively handled during the routing process.

[0062] Step 102: Based on the specific location of the multi-pin net in the routing grid in step 101, use a disjoint-set data structure-based pruning strategy to merge the pins directly connected on the M0 layer, and obtain the mapping relationship between the net name and the pin coordinate set.

[0063] Step 103: Using the minimum spanning tree algorithm, the nets with multiple pins in the mapping relationship between net names and pin coordinates obtained in Step 102 are split into multiple two-pin pairs, thus obtaining the mapping relationship between net names and two-pin pairs. This simplifies the routing problem, reduces computational complexity, and facilitates subsequent routing execution.

[0064] Step 104: Based on the characteristics of the nets in the wiring mesh (such as congestion level and half-circumference line length), arrange the wiring sequence reasonably to obtain the sequence of net names representing the wiring sequence;

[0065] Step 105: Following the net name sequence in Step 104, route each net sequentially using the congestion-driven 3D routing algorithm based on source region expansion for each two-pin pair obtained in Step 103, to obtain the initial routing result.

[0066] Step 106: Apply a boundary design rule checking technique with linear time complexity to the initial routing results from Step 105 to detect congestion and verify whether the routing paths meet the design rules. Obtain congestion information for each net. If all nets are free of congestion, output the final routing results; otherwise, proceed to Step 107.

[0067] Step 107: Based on the congestion information in step 106, remove the congested paths in the cabling results of 105, and update the cost function for the points on the path using a negotiation-based mechanism. Then repeat steps 104 to 106.

[0068] Furthermore, in one embodiment, in step 101, according to the ASAP7 PDK design rules, the standard cell height is 270nm, and the top and bottom rails are reserved for power signals (VDD and VSS). After excluding these reserved rails, the available effective wiring height is 216nm. The designed wiring grid has a side length of 9nm, and is expanded into a square with a side length of 18nm centered on the grid points for the actual wiring. Based on the input circuit netlist and layout results, data is read in sequence, and each transistor and its net are mapped one by one to the wiring grid according to coordinates to obtain the wiring network diagram.

[0069] Furthermore, in one embodiment, in step 102, under specific layout conditions, pins belonging to the same net can be directly connected on the M0 layer without additional M1 layer vias. To take advantage of this opportunity and reduce unnecessary routing complexity, a rapid pin merging procedure is performed before the routing process. This merging process efficiently identifies and groups connectable pins using a disjoint-set data structure.

[0070] Specifically, if two active region pins belonging to the same net are adjacent, they can be directly merged. For two gate pins belonging to the same net, if the distance between them is one polysilicon pitch, they can be merged. Alternatively, if the distance between them is two or three polysilicon pitches, and the area in between is entirely active region diffusion, they can also be merged.

[0071] It's important to note that no M0 connection is actually introduced at this stage. Since the final V0 hole location for signal extraction can only be determined after routing is complete, all merging in this stage is logical merging, not physical merging. This ultimately yields a mapping relationship between net names and the merged pin coordinate set.

[0072] Furthermore, in one embodiment, in step 103, for all nets, if the pins in the pin coordinate set mapped in step 102 are multiple pins, the minimum spanning tree algorithm will be used to decompose the complex multi-pin into several two-pin pairs, and obtain the mapping relationship between each net name and the two-pin pair.

[0073] Furthermore, in one embodiment, in step 104, all nets are sorted using formula (1), with nets having lower scores receiving higher priority for routing. Wherein, the net sorting score... The expression is as follows:

[0074] (1)

[0075] in, This indicates that in the previous iteration, the net... The number of points on the wiring path, This indicates the number of congestion points encountered along the cabling path. It is a wire mesh The half-perimeter length (HPWL). Weighting parameters. and Defined for users, satisfying Prioritize congestion information over geometric span.

[0076] Increased congestion along the cabling path can lead to Increase, thereby reducing the ratio This leads to a decrease in the corresponding score. Therefore, networks experiencing higher congestion will be prioritized for cabling to reduce potential resource conflicts. Meanwhile, networks with lower congestion... Larger networks will be prioritized because they offer greater flexibility in routing. By jointly considering congestion severity and network geometry, this scoring strategy improves the likelihood of achieving congestion-free cabling under limited resource conditions. This ultimately results in a sequence of network orders.

[0077] In a further embodiment, in step 105, each net is sequentially routed using a source region expansion-based congestion-driven 3D routing algorithm for each two-pin pair obtained in step 103, according to the net name sequence in step 104.

[0078] Unlike traditional routing methods, the congestion-driven 3D routing algorithm based on source region expansion in this invention selects a single point each from the source and sink regions as the start and end points. During the routing process, a source region expansion strategy is employed, directly using all legal grid points within the source and sink regions as the search start and end points for a bidirectional expansion search. Specifically, the source region dynamically expands along existing routing paths, allowing subsequent routing to start directly from any point on the existing path. This method avoids redundant searches, improves routing flexibility, and significantly accelerates the overall routing speed.

[0079] Table 1 shows the proposed congestion-driven 3D routing algorithm:

[0080] Table 1 shows the congestion-driven 3D wiring algorithm based on source region expansion.

[0081]

[0082] This algorithm employs a bidirectional exploration strategy to accelerate convergence. The algorithm expands simultaneously from both the source and sink regions. Depending on the chosen starting region, the input `starts` includes all points within either the source or sink region, while `goals` is defined as the union of all points in the opposing region and the corresponding `PossibleSet`, specifically including:

[0083] Step 501: Extract each net from the net name sequence and input each two-pin pair in the net into step 502.

[0084] Step 502, initialize the data structures: PossibleSet, SearchHeap, CostDict, and ParentDict. PossibleSet is a set of candidate points for building potential paths; SearchHeap is a heap for dynamically maintaining points sorted by cumulative cost; CostDict is a dictionary for storing the best known cumulative cost for each point; and ParentDict is a dictionary for storing the parent-child relationships required for backtracking paths.

[0085] Step 503: Initialize starts by inserting each grid point u in starts into SearchHeap with an initial cost of zero, and record it in CostDict. Here, starts refers to all grid points within one pin region or the other pin domain of the two pins.

[0086] Step 504: Take a grid point from starts and input it into step 505.

[0087] Step 505: Iterate through each adjacent point v of the input grid point u in the wiring mesh graph G. For each adjacent point v, calculate the cumulative cost from grid point u to grid point v. If grid point v has not been visited before, or if the cumulative cost to grid point v is found to be lower than the cumulative cost of grid point v in CostDict, then use the updated cumulative cost as the sorting index to insert grid point v into SearchHeap to obtain the updated SearchHeap. Update the cumulative cost of grid point v in CostDict to obtain the updated CostDict. Update the parent node of grid point v in ParentDict, and set grid point u as the parent node of grid point v to obtain the updated ParentDict.

[0088] Step 506: Retrieve the grid point u with the minimum cumulative cost from SearchHeap and check if grid point u belongs to the goals set. If not, input grid point u into step 505. Otherwise, call the path_backtrack(ParentDict, u) function to backtrack the wiring path, terminate the search, and return the wiring path R. Here, goals is the union of all points in the opposing region and their corresponding PossibleSet, and path_backtrack(ParentDict, u) represents the wiring path from grid point u backtracking to any grid point in starts according to the parent-child relationship in ParentDict.

[0089] Step 507: Repeat steps 505 to 506 until the routing path of the two-pin pair is successfully constructed, and obtain the routing result of the two-pin pair. Repeat steps 501 to 506 to obtain the routing result of all two-pin pairs corresponding to all nets as the initial routing result.

[0090] In line 12, CostDict[u] and calculate ,in: This represents the cumulative cost from start to point v. The formal definition is:

[0091] (2)

[0092] in: This represents the set of consecutive pairs of points on the path from starts to point v. This represents the cost of traversing between consecutive points. This represents the i-th u-th point. This represents the (i+1)th u-point.

[0093] During the cabling exploration phase, the cumulative cost It can be updated incrementally in the following ways:

[0094] (3)

[0095] Where u is the parent node of point v.

[0096] Cost function The cost of traversing from the current grid point u to its adjacent point v is defined as follows:

[0097] (4)

[0098] in: Represents the basic cost function. This represents the cumulative historical cost of v at the (k-1)th iteration. The variable k represents the current wiring iteration number. Defined as:

[0099] (5)

[0100] in, = 1 indicates that congestion was detected between u and v. = 0 indicates no congestion was detected. This formula dynamically increases the cost of traversing congested regions based on the number of iterations k, thereby encouraging the algorithm to gradually prioritize wiring paths with less congestion. (Related information...) More details of the evaluation and boundary design rule checking process are in step 106.

[0101] Basic cost function The traversal cost from the current grid point u to the adjacent grid point v is defined as follows:

[0102] (6)

[0103] in: This represents the cost associated with the change of direction. This represents the cost associated with layer switching. (Cost of direction change) Defined as:

[0104]

[0105] in: and The experimental settings are 1 and 5.

[0106] Layer switching cost Defined as:

[0107]

[0108] in: Indicates the cost of a through hole. express The wiring layer to which it belongs, = 10, respectively It belongs to the wiring layer representing M0, M1 and M2.

[0109] Furthermore, in one embodiment, in step 106, for the initial wiring result obtained in step 105, it is necessary to verify whether the wiring path conforms to the design rule constraints. For a certain point v on the path, determining the congestion of point v not only requires checking whether point v has been occupied by the wiring of other networks, but also requires checking the usage of other points in the surrounding area of ​​point v.

[0110] For a minimum side-to-side (S2S) spacing of 18nm, this corresponds to a coordinate difference of 3 or more between two points in the mesh diagram. For a minimum side-to-tip (S2T) spacing of 25nm, this corresponds to a coordinate difference of 4 or more between two points in the mesh diagram. For a minimum tip-to-tip (T2T) spacing of 27nm, this corresponds to a coordinate difference of 4 or more between two points in the mesh diagram. Therefore, it is only necessary to ensure two points meet the wiring path load design rule constraints: first, for points on the side, points with coordinate differences of 3 or more must belong to the same mesh or be unoccupied; second, for points on the tip, points with coordinate differences of 4 or more must belong to the same mesh or be unoccupied. A common approach is to search and determine the attributes of points within a coordinate difference range of x one by one.

[0111] Furthermore, in one embodiment, in step 107, due to limited cabling resources, congestion-driven 3D cabling algorithms typically optimize based on already cabled paths, failing to anticipate unrouted networks, which limits their ability to proactively avoid congestion. To address this issue, a negotiation-based approach is employed, dynamically updating the cost function through teardown and rerouting mechanisms.

[0112] In step 106, during the boundary design rule check for point v, if congestion is detected at a neighboring point p, both points v and p are added to the congestion set for the current iteration. Historical cost function. The purpose is to capture the cumulative congestion penalty up to the k-th iteration, defined as follows:

[0113] (7)

[0114] in, It is a binary pointer function, defined as:

[0115]

[0116] According to formula (4), the historical cost function The weighted congestion penalty is accumulated in each iteration, where each congestion in the j-th iteration will generate a penalty proportional to j. Therefore, the ratio in formula (4) This quantifies the average congestion severity experienced by v before the (k-1)th iteration. It serves as a dynamic congestion memory. Adaptively prioritizing congestion occurring in recent iterations mitigates the disproportionate impact of early-stage congestion on routing decisions. Furthermore, k in Equation (5) promotes negotiation flexibility in the early stages of routing. As k increases, the cumulative penalty grows, guiding the solution toward convergence and reducing path oscillations.

[0117] Example 2:

[0118] This embodiment presents a comparative simulation experiment of the routing method, advanced router, and manual layout of the present invention against the ASAP7 standard cell library. The bus length is measured in micrometers, and the runtime is measured in seconds. Note that manual layout cannot provide time statistics. Normalization is based on the results of the present invention. Experimental results show that compared to manual layout, the bus length is reduced by 28%, and compared to the advanced router, the bus length is reduced by 13%, while the runtime is increased by 62.7 times. This demonstrates that the routing method proposed in this invention can achieve high-quality routing in a very short time, as detailed in Table 2.

[0119] Table 2 Comparison of simulation experiment line length and running time

[0120]

[0121] Example 3:

[0122] This embodiment introduces a negotiation-based advanced node transistor routing apparatus. This hardware verification apparatus is a virtual device that can be loaded and executed by a computer device. The computer device may include the aforementioned hardware verification apparatus. The hardware verification apparatus may include a first processing unit to a seventh processing unit, which, when executing the aforementioned hardware verification method, can:

[0123] The first processing unit is used to construct a wiring mesh diagram based on the circuit netlist and layout results.

[0124] The second processing unit is used to merge the pins directly connected on the M0 layer according to the position of the multi-pin net in the wiring mesh diagram, and obtain the mapping relationship between the net name and the pin coordinate set.

[0125] The third processing unit is used to split the multi-pin nets in the mapping relationship between net names and pin coordinate sets into multiple two-pin pairs, and obtain the mapping relationship between net names and two-pin pairs.

[0126] The fourth processing unit is used to arrange the wiring order according to the attribute characteristics of the nets in the wiring mesh, and obtain a sequence of net names representing the wiring order.

[0127] The fifth processing unit is used to route the two-pin pairs in each net in sequence according to the net name sequence to obtain the initial routing result.

[0128] The sixth processing unit is used to verify whether the initial routing results meet the design rules. If all nets are uncongested, the final routing results are output. If any nets are congested, the seventh processing unit is used.

[0129] The seventh processing unit is used to tear down congested paths in the initial routing result based on congestion information, update the cost function for the points on the path, and repeat the fourth, fifth and sixth processing units until the final routing result is output.

[0130] Example 4:

[0131] This embodiment describes a computer-readable storage medium storing a computer program that, when executed by a processor, implements a negotiation-based advanced node transistor routing method as described in any of Embodiment 1.

[0132] Example 5:

[0133] This embodiment describes a computer device, which may include the aforementioned hardware verification device, specifically including:

[0134] Memory is used to store instructions.

[0135] A processor is configured to execute the instructions, causing the computer device to perform operations of a negotiation-based advanced node transistor routing method as described in any of Embodiment 1.

[0136] Example 6:

[0137] This embodiment describes a program product that, when executed by a processor, implements a negotiation-based advanced node transistor routing method as described in any of Embodiment 1.

[0138] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0139] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0140] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0141] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0142] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A negotiation-based advanced node-under-transistor wiring method, characterized by: Specifically, it includes: Step 101: Based on the circuit netlist and layout results, construct the wiring mesh diagram; Step 102: Based on the position of the multi-pin nets in the wiring mesh diagram, merge the pins directly connected on the M0 layer to obtain the mapping relationship between the net name and the pin coordinate set; Step 103: In the mapping relationship between net name and pin coordinate set, nets with multiple pins are split into multiple two-pin pairs to obtain the mapping relationship between net name and two-pin pairs; Step 104: Arrange the wiring order according to the properties of the nets in the wiring mesh to obtain a sequence of net names representing the wiring order; Step 105: Routing the two-pin pairs in each net in sequence according to the net name sequence to obtain the initial routing result; Step 106: Verify whether the initial routing results meet the design rules. If all nets are uncongested, output the final routing results. If any nets are congested, proceed to step 107. Step 107: Based on the congestion information, remove congested paths in the initial routing results and update the cost function for the points on the paths. Repeat steps 104 to 106 until the final routing result is output.

2. The method of claim 1, wherein: The merging of pins directly connected on the M0 layer specifically includes: Two active area pins belonging to the same wire network are adjacent to each other, and the two pins are directly merged; If two gate pins belonging to the same wire mesh are spaced one polysilicon pitch apart, then the two pins are merged. Two gate pins belonging to the same wire mesh are merged if the distance between the two pins is 2 or 3 polysilicon pitches and the area in between is a complete diffusion interruption.

3. The method of claim 1, wherein: Step 104 specifically includes: Calculate the net ranking score based on the properties of the nets in the wiring mesh; The nets are sorted from low to high according to their net sorting scores to obtain a sequence of net names representing the routing order. wherein the line net ordering score is expressed as follows: ; wherein, represents the number of points on the routing path of the online network represents the number of points on the routing path of the online network represents the number of congested points encountered along the routing path, represents the half-perimeter wire length of the online network represents the half-perimeter wire length of the online network represents a first weight parameter, represents a second weight parameter.

4. The negotiation-based transistor routing method for advanced nodes according to claim 1, characterized in that: Step 105 specifically includes: Step 501: Extract each net from the net name sequence one by one, and input each two-pin pair in the net into step 502; Step 502, initialize the data structures: PossibleSet, SearchHeap, CostDict, and ParentDict; where PossibleSet is a set of candidate points for storing potential path construction; SearchHeap is a heap for dynamically maintaining points sorted by cumulative cost; CostDict is a dictionary for storing the best known cumulative cost for each point; and ParentDict is a dictionary for storing the parent-child relationships required for backtracking paths. Step 503: Initialize starts, insert each grid point u in starts into SearchHeap with an initial cost of zero, and record it in CostDict; where starts is all grid points in one pin region of the two pins or the other pin domain of the two pins; Step 504: Take a grid point from starts and input it into step 505; Step 505: Traverse each adjacent point v of the input grid point u in the wiring mesh graph G. For each adjacent point v, calculate the cumulative cost from grid point u to grid point v. If grid point v has not been visited before, or if the cumulative cost to grid point v is found to be lower than the cumulative cost of grid point v in CostDict, then use the updated cumulative cost as the sorting index to insert grid point v into SearchHeap to obtain the updated SearchHeap. Update the cumulative cost of grid point v in CostDict to obtain the updated CostDict. Update the parent node of grid point v in ParentDict, and use grid point u as the parent node of grid point v to obtain the updated ParentDict. Step 506: Extract the grid point u with the minimum cumulative cost from SearchHeap and check if grid point u belongs to the goals set. If not, input grid point u into step 505; otherwise, call the path_backtrack(ParentDict, u) function to backtrack the wiring path, terminate the search, and return the wiring path R. Here, goals is the union of all points in the opposing region and their corresponding PossibleSet, and path_backtrack(ParentDict, u) represents the wiring path from grid point u backtracking to any grid point in starts according to the parent-child relationship in ParentDict. Step 507: Repeat steps 505 to 506 until the routing path of the two-pin pair is successfully constructed, and obtain the routing result of the two-pin pair. Repeat steps 501 to 506 to obtain the routing result of all two-pin pairs corresponding to all nets as the initial routing result.

5. The negotiation-based transistor routing method for advanced nodes according to claim 4, characterized in that: Cumulative cost of grid point v The expression is as follows: ; in: This represents the cumulative cost of parent node u. This represents the cost function based on congestion. Cost function based on congestion The expression is as follows: ; in: Represents the basic cost function. This represents the historical cost function value of network point v at the (k-1)th iteration; variable k represents the current wiring iteration number. Indicates the degree of congestion.

6. The negotiation-based transistor routing method for advanced nodes according to claim 5, characterized in that: The update cost function The expression is as follows: ; in, It is a binary pointer function.

7. A negotiation-based transistor wiring apparatus for advanced nodes, characterized in that: Specifically, it includes: The first processing unit is used to construct a wiring mesh diagram based on the circuit netlist and layout results; The second processing unit is used to merge the pins directly connected on the M0 layer according to the position of the multi-pin net in the wiring mesh diagram, and obtain the mapping relationship between the net name and the pin coordinate set. The third processing unit is used to split the nets with multiple pins in the mapping relationship between net names and pin coordinate sets into multiple two-pin pairs, and obtain the mapping relationship between net names and two-pin pairs. The fourth processing unit is used to arrange the routing order according to the attribute characteristics of the nets in the routing mesh, and obtain a sequence of net names representing the routing order; The fifth processing unit is used to route the two-pin pairs in each net in sequence according to the net name sequence to obtain the initial routing result; The sixth processing unit is used to verify whether the routing path meets the design rules for the initial routing result. If all nets are free of congestion, the final routing result is output. If any net is congested, the seventh processing unit is used. The seventh processing unit is used to tear down congested paths in the initial routing result based on congestion information, update the cost function for the points on the path, and repeat the fourth, fifth and sixth processing units until the final routing result is output.

8. A computer-readable storage medium, characterized in that: It stores a computer program that, when executed by a processor, implements a negotiation-based advanced node transistor routing method as described in any one of claims 1 to 6.

9. A computer device, characterized in that: Specifically, this may include: Memory, used to store instructions; A processor for executing the instructions, causing the computer device to perform operations of a negotiation-based advanced node transistor routing method as described in any one of claims 1 to 6.

10. A program product, characterized in that: When the program product is executed by the processor, it implements a negotiation-based advanced node transistor routing method as described in any one of claims 1 to 6.