CN-121981065-A - Wiring method, electronic device, and computer-readable storage medium

CN121981065ACN 121981065 ACN121981065 ACN 121981065ACN-121981065-A

Abstract

The invention relates to the technical field of wiring, and discloses a wiring method, electronic equipment and a computer readable storage medium, wherein the method comprises the steps of converting an initial layout of a circuit board into a grid diagram; the method comprises the steps of determining a coarse wiring result of an initial layout according to a grid diagram, wherein the coarse wiring result is an initial wiring path which meets wiring tasks and achieves a wiring crossing minimization target, generating a wiring geometric figure according to a safe distance and the initial wiring path, generating a device geometric figure according to the safe distance and component information in the initial layout, and optimizing the position of the wiring geometric figure under the limitation of the device geometric figure by taking the meeting wiring constraint as the target to obtain a fine wiring result. The invention divides the wiring process into a coarse wiring stage meeting the goals of connectivity and minimum wiring crossing number and a fine wiring stage meeting wiring constraint, which can reduce the solving complexity, thereby obtaining the wiring scheme with minimum wiring crossing number in a shorter time.

Inventors

CAI XINYUE
XU JIAO
CHEN YAOCONG
LUAN JINGYE
MO XIAOZHEN

Assignees

广州视源电子科技股份有限公司

Dates

Publication Date: 20260505
Application Date: 20251128

Claims (12)

1. A wiring method, characterized in that the wiring method comprises: Converting an initial layout of a circuit board into a grid diagram, wherein the grid diagram comprises a plurality of nodes and grids corresponding to the nodes one by one, the nodes are used for representing one pin in the initial layout, and the grids are used for representing routable areas of the corresponding nodes; Determining a thick wiring result of the initial layout according to the grid diagram, wherein the thick wiring result is an initial wiring path which meets wiring tasks and achieves a wiring cross minimization target; generating a wiring geometric figure according to the safety distance and the initial wiring path, and generating a device geometric figure according to the safety distance and the component information in the initial layout; And under the limitation of the device geometric figure, optimizing the position of the wiring geometric figure with the aim of meeting wiring constraint, and obtaining a thin wiring result.
2. The routing method according to claim 1, wherein said determining a coarse routing result of said initial layout from said mesh map comprises: and when the initial layout is of a target type, determining a coarse wiring result according to a target reinforcement learning model and the grid diagram, wherein the target type is a layout with a corresponding wiring result and a storage capacity larger than a preset capacity.
3. The routing method of claim 2, wherein the target reinforcement learning model comprises an agent and a policy network employing a multi-layer graph sampling aggregation layer, wherein determining coarse routing results from the target reinforcement learning model and the grid graph comprises: Determining node characteristics and edge characteristics corresponding to the nodes where the intelligent agent is located according to the node information of the grid graph; Inputting the node characteristics and the edge characteristics into the strategy network, and determining an action set of the intelligent agent according to the output of the strategy network, wherein the action set comprises at least one optional action and a probability corresponding to each optional action; determining a next action of the agent according to the probability of the optional action and whether the optional action causes a path crossing, so as to obtain the coarse wire-bonding result.
4. The wiring method according to claim 3, wherein the number of the agents is plural, the mesh map includes a plurality of local mesh maps, the plurality of agents are in one-to-one correspondence with the plurality of local mesh maps, the plurality of agents adopt the same policy network, the determining node characteristics and edge characteristics corresponding to the selected nodes of the agents according to node information of the mesh maps includes: and determining node characteristics and edge characteristics of the intelligent agent corresponding to the local grid graph according to the node information of the local grid graph.
5. The wiring method as in claim 4, wherein after the next step of action of the agent is taken, the wiring method further comprises: updating node information of the local mesh map; And updating the node information of the grid graph and the node information of the grid graph containing prior information according to the updated node information of the local grid graphs.
6. The wiring method according to claim 4 or 5, wherein the node information includes information indicating whether a node is occupied or not and distance information, and when a first node is occupied by the agent but a distance corresponding to the first node does not reach a maximum value, the first node is marked as unoccupied, and the first node is a node other than an initial node and a target node in a partial mesh map corresponding to the agent.
7. The wiring method according to any one of claims 1 to 5, wherein the mesh pattern is a voronoi diagram.
8. The routing method according to claim 1, wherein said determining a coarse routing result of said initial layout from said mesh map comprises: And when the initial layout is not of the target type, determining a coarse wiring result according to the constraint relaxation flow model.
9. The routing method of claim 8, wherein determining coarse routing results from a constraint relaxation flow model comprises: Under the condition that the wiring task is met, solving the constraint relaxation flow model by taking the maximized loop communication quantity and the minimized loop line length as optimization targets to obtain an initial coarse wiring result; When the initial rough wiring result has a conflict edge, adding the conflict edge as a constraint condition to the constraint relaxation flow model and re-solving the constraint relaxation flow model until the conflict edge does not exist or the solving frequency reaches the maximum limiting frequency; And determining an initial coarse wiring result without a conflict edge as the coarse wiring result, or determining the initial coarse wiring result after the solving frequency reaches the maximum limiting frequency as the coarse wiring result.
10. The routing method of any of claims 1 to 5, wherein the number of routing geometries is a plurality, the initial routing path comprises a plurality of loops, and the optimizing the location of the routing geometries, with the goal of satisfying routing constraints, under the constraints of the device geometries, results in a fine routing result comprises: Traversing the wiring geometric figures, and determining a second wiring geometric figure which is spatially overlapped with a first wiring geometric figure to form a conflict set, wherein the first wiring geometric figure is any one of a plurality of wiring geometric figures; Determining at least one pushing loop according to the conflict set, wherein the pushing loop is a loop with conflict points in the loops; performing pushing treatment on the wiring geometric figure in at least one pushing loop to obtain a pushing result; And updating the positions of the plurality of wiring geometric figures based on the pushing result, and returning to the step of forming a conflict set until the plurality of wiring geometric figures meet the wiring constraint or the updating times reach the preset updating times.
11. An electronic device, comprising: A memory and a processor communicatively coupled to each other, the memory having stored therein computer instructions that, upon execution, perform the wiring method of any of claims 1-10.
12. A computer-readable storage medium having stored thereon computer instructions for causing a computer to execute the wiring method according to any one of claims 1 to 10.

Description

Wiring method, electronic device, and computer-readable storage medium Technical Field The present invention relates to the field of wiring technology, and in particular, to a wiring method, an electronic device, and a computer-readable storage medium. Background In integrated circuit designs and printed circuit board (Printed Circuit Board, PCB) designs, routing is an important step, requiring the connection of a large number of circuit components (multiple pins) without violating printed circuit or integrated circuit rules. At present, automatic wiring is performed by means of a heuristic algorithm of electronic design automation (Electronic Design Automation, EDA) software, but when the minimum number of wiring crossings is taken as an optimization target on the premise of meeting connectivity, a possible wiring scheme grows exponentially with the increase of a loop scale, and an optimal solution with the minimum wiring crossings is difficult to find in a short time. Disclosure of Invention The invention provides a wiring method, an electronic device and a computer readable storage medium, which are used for solving the problem that an optimal solution with minimum wiring crossing is difficult to find in a short time. The first aspect of the invention provides a wiring method, which comprises the steps of converting an initial layout of a circuit board into a grid diagram, wherein the grid diagram comprises a plurality of nodes and grids corresponding to the nodes one by one, the nodes are used for representing one pin in the initial layout, the grids are used for representing a routable area of the corresponding nodes, a coarse wiring result of the initial layout is determined according to the grid diagram, the coarse wiring result is an initial wiring path meeting a wiring task and achieving a wiring crossover minimization target, a wiring geometric figure is generated according to a safety distance and the initial wiring path, a device geometric figure is generated according to the safety distance and component information in the initial layout, and under the limitation of the device geometric figure, the position of the wiring geometric figure is optimized with the aim of meeting wiring constraint, so that a fine wiring result is obtained. According to the wiring method provided by the embodiment, the initial layout is firstly converted into the grid diagram, then the coarse wiring result is determined according to the grid diagram, and finally the coarse wiring result is converted into the fine wiring result meeting wiring constraint. The wiring problems meeting the requirements of connectivity, wiring constraint and wiring crossing quantity minimization are decoupled and split into two-stage solving, so that the solving complexity can be reduced, the solving speed is greatly improved, the wiring scheme with the minimum wiring crossing quantity is realized in a short time, the design efficiency is improved, and the design cost is reduced. Moreover, by combining action reduction of a geometric extrusion strategy and path optimization design, the coarse wiring result is converted into the fine wiring result, so that the problem of low conflict resolution efficiency in the traditional conversion process is solved, and the wiring quality is ensured through geometric optimization. In an alternative embodiment, determining the coarse wiring result of the initial layout according to the grid diagram comprises determining the coarse wiring result according to the target reinforcement learning model and the grid diagram when the initial layout is of a target type, wherein the target type is a layout with a corresponding wiring result and a storage capacity larger than a preset capacity. In an alternative embodiment, the target reinforcement learning model comprises an agent and a strategy network adopting a multi-layer graph sampling aggregation layer, a coarse wiring result is determined according to the target reinforcement learning model and a grid graph, the method comprises the steps of determining node characteristics and edge characteristics corresponding to nodes where the agent is located according to node information of the grid graph, inputting the node characteristics and the edge characteristics into the strategy network, determining an action set of the agent according to output of the strategy network, wherein the action set comprises at least one optional action and probability corresponding to each optional action, and determining the next action of the agent according to the probability of the optional action and whether the optional action can cause path crossing so as to obtain the coarse wiring result. In this embodiment, through the k-layer graph sampling aggregation layer, the sensing range of the agent can be extended to k-order neighbor nodes, and the next action of the agent can be determined according to the probability of the optional action and whether the optional action can cau