Time sequence analysis method and related device
By extracting and utilizing line time delay sequence features and process angle features in the pre-wiring stage, combining RC information and wiring congestion information, the accuracy of timing analysis in the pre-wiring stage is solved, and the accuracy of timing analysis of integrated circuit design is improved and Design optimization effect.
Patent Information
- Application Number
- CN202411087230.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-12-03
AI Technical Summary
In integrated circuit design, the accuracy of timing analysis in the pre-wiring stage is poor, which mainly leads to the lack of RC information due to the lack of actual wiring information, which in turn affects the accuracy of line delay analysis and leads to difficulty in design optimization.
By extracting the look-ahead tree, wiring tree and layout input information based on the target information of the netlist file, the line-time delay sequence characteristics are obtained, including the line network's own characteristics and neighboring line-network characteristics, and the time delay analysis is carried out in combination with the process angle characteristics, and RC is introduced Information and wiring congestion and coupling information to correct ideal wiring results in the wiring tree.
It improves the accuracy of timing analysis, reduces the error of timing analysis in the pre-wiring stage, enhances the accurate prediction of line delay, and guides more effective circuit design optimization.
Smart Images

Figure CN119067065A_ABST
Abstract
Description
Technical Field
[0001] The present specification relates to the field of integrated circuit design technology, specifically, to the timing analysis technology in the field of integrated circuit design technology, and more specifically, to a timing analysis method and related devices. Background Art
[0002] Very Large-Scale Integration (VLSI) design is a very difficult problem, spanning multiple directions such as high-performance computing, numerical optimization, and computer systems. Modern chip design has a huge number of transistors, up to tens of billions, and is very complex. Therefore, Electronic Design Automation (EDA) consists of multiple tools. Starting from the logic netlist, the chip design can be completed through processes such as layout planning, component placement, clock tree synthesis, and wiring. In these processes, timing convergence is a critical and laborious task.
[0003] Accurate timing analysis is essential for optimizing design, ensuring timing closure and operating performance. The earlier the timing analysis is involved, the greater the room for timing optimization. Currently, the accuracy of timing analysis before routing is poor, and its guiding significance for optimizing integrated circuit design is weak. Summary of the invention
[0004] The embodiments of this specification provide a timing analysis method and related devices to achieve the purpose of improving the accuracy of timing analysis in the pre-routing stage.
[0005] To achieve the above technical objectives, the embodiments of this specification provide the following technical solutions:
[0006] In a first aspect, an embodiment of the present specification provides a timing analysis method, comprising:
[0007] Based on the target information of the target circuit in the netlist file, the line delay timing characteristics of the target circuit are obtained, the target information includes: a look-ahead tree, a routing tree and layout input information; the routing tree is used to describe the pre-wiring style of the circuit in the netlist file; the look-ahead tree is obtained based on the process corner characteristics and the routing tree, the look-ahead tree is used to describe the routing information between the two-end connection relationship in the routing tree, the routing information is used to describe the RC information of the routing between the two-end connection relationship, and the two-end connection relationship is used to characterize the connection relationship between two nodes in the circuit; the layout input information is used to describe the position information of the node, and the node is used to characterize the input and output ports or pins; the line delay timing characteristics include: the line net itself characteristics and the neighbor line net characteristics; the line net itself characteristics are obtained based on the look-ahead tree, the routing tree and the layout input information, the neighbor line net characteristics are obtained based on the layout input information, and the neighbor line net characteristics are used to describe the routing congestion and coupling information related to the target circuit;
[0008] Based on the line delay timing characteristics and the process corner characteristics, a delay analysis is performed on the target circuit to obtain the arrival time of each of the nodes in the target circuit, and based on the arrival time of each of the nodes in the target circuit and the required arrival time corresponding to each of the nodes, a timing analysis result of each path in the target circuit is obtained.
[0009] In a second aspect, an embodiment of the present specification provides a timing analysis device, the timing analysis device comprising:
[0010] A feature extraction module is used to obtain the line delay timing characteristics of the target circuit based on the target information of the target circuit in the netlist file, the target information includes: a look-ahead tree, a wiring tree and layout input information; the wiring tree is used to describe the pre-wiring style of the circuit in the netlist file; the look-ahead tree is obtained based on the process corner feature and the wiring tree, the look-ahead tree is used to describe the wiring information between the two-end connection relationship in the wiring tree, the wiring information is used to describe the RC information of the wiring between the two-end connection relationship, and the two-end connection relationship is used to characterize the connection relationship between two nodes in the circuit; the layout input information is used to describe the position information of the node, and the node is used to characterize the input and output port or pin; the line delay timing characteristics include: the line net itself characteristics and the neighbor line net characteristics; the line net itself characteristics are obtained based on the look-ahead tree, the wiring tree and the layout input information, the neighbor line net characteristics are obtained based on the layout input information, and the neighbor line net characteristics are used to describe the wiring congestion and coupling information related to the target circuit;
[0011] The delay analysis module is used to perform delay analysis on the target circuit based on the line delay timing characteristics and the process corner characteristics to obtain the arrival time of each node in the target circuit, and obtain the timing analysis results of each path in the target circuit based on the arrival time of each node in the target circuit and the required arrival time corresponding to each node.
[0012] In a third aspect, an embodiment of the present specification further provides a computing device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the timing analysis method as described above when executing the computer program.
[0013] In a fourth aspect, an embodiment of the present specification further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the timing analysis method as described above is implemented.
[0014] In a fifth aspect, an embodiment of the present specification provides a computer program product or a computer program, wherein the computer program product includes a computer program, and the computer program is stored in a computer-readable storage medium; the processor of the computer device reads the computer program from the computer-readable storage medium, and the processor implements the steps of the above-mentioned timing analysis method when executing the computer program.
[0015] It can be seen from the above technical solution that an embodiment of the present specification provides a timing analysis method. When the timing analysis method performs timing analysis in the pre-wiring stage, based on the target information of the target circuit, the line delay timing characteristics of the target circuit are obtained, wherein the target information includes: a look-ahead tree, a wiring tree and layout input information, the netlist file includes multiple wire nets, the wire net includes multiple nodes and multiple edges, the nodes are used to characterize the input and output ports or pins of the driver or receiver in the wire net, and the edges are used to characterize the timing arcs between two nodes; the wiring tree is used to describe the pre-wiring style of the circuit in the netlist file; the look-ahead tree is obtained based on the process corner characteristics and the wiring tree, the look-ahead tree is used to describe the wiring information between the connection relationships at both ends in the wiring tree, and the wiring information is used to describe the wiring between the connection relationships at both ends. RC information, the connection relationship between the two ends is used to characterize the connection relationship between the two nodes in the circuit; the layout input information is used to describe the location information of the node; the line delay timing characteristics include: the line network's own characteristics and the neighboring line network characteristics; the line network's own characteristics are obtained based on the look-ahead tree, the wiring tree and the layout input information, the neighboring line network characteristics are obtained based on the layout input information, and the neighboring line network characteristics are used to describe the wiring congestion and coupling information related to the target circuit; then, based on the line delay timing characteristics and the process corner characteristics, the target circuit is subjected to a delay analysis to obtain the arrival time of each of the nodes in the target circuit, and based on the arrival time of each of the nodes in the target circuit and the required arrival time corresponding to each of the nodes, the timing analysis results of each path in the target circuit are obtained. In this way, the line network's own characteristics obtained during the timing analysis process not only take into account the layout input information and the relevant information in the wiring tree, but also take into account the RC information of the wiring between the two end connection relationships in the look-ahead tree, thereby solving the problem of inaccurate timing analysis (especially line delay analysis) in the pre-wiring stage due to the lack of actual wiring information in the pre-wiring stage. In addition, the line delay timing characteristics also include neighbor line network characteristics, which enables the wiring congestion and coupling information related to the target circuit described based on the neighbor line network characteristics to assist in correcting the ideal wiring results in the wiring tree, comprehensively considering the delay impact of the neighbor line network on each path in the target circuit, and is conducive to improving the accuracy of timing analysis. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0017] Figure 1A schematic diagram of a timing path;
[0018] Figure 2 Based on Figure 1 Schematic diagram of the obtained DAG graph;
[0019] Figure 3 A flowchart of a timing analysis method provided in an embodiment of this specification;
[0020] Figure 4 An overview schematic diagram of a timing analysis method provided in an embodiment of this specification;
[0021] Figure 5 A schematic diagram of determining a neighbor network provided in an implementation manner of this specification;
[0022] Figure 6 A schematic diagram of the importance ranking of line delay timing characteristics provided in an embodiment of this specification;
[0023] Figure 7 A schematic diagram comparing the delay analysis results of the timing analysis method provided in the embodiments of this specification, the delay analysis results obtained by the timing analysis method in the related art, and the delay analysis results obtained in the Sign-Off stage;
[0024] Figure 8 A schematic diagram of the structure of a computing device provided for an embodiment of this specification. DETAILED DESCRIPTION
[0025] Unless otherwise defined, the technical terms or scientific terms used in the embodiments of this specification shall have the common meanings understood by persons with ordinary skills in the field to which this specification belongs. The words "first", "second" and similar words used in the embodiments of this specification do not indicate any order, quantity or importance, but are only used to avoid confusion of constituent elements.
[0026] Unless the context requires otherwise, throughout the specification, "plurality" means "at least two", and "including" is interpreted as an open, inclusive meaning, that is, "including, but not limited to". In the description of the specification, the terms "one embodiment", "some embodiments", "exemplary embodiments", "examples", "specific examples" or "some examples" are intended to indicate that a particular feature, structure, material or characteristic associated with the embodiment or example is included in at least one embodiment or example of the specification. The schematic representation of the above terms does not necessarily refer to the same embodiment or example.
[0027] The technical solutions in the embodiments of this specification will be described below in conjunction with the drawings in the embodiments of this specification. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this specification.
[0028] Overview
[0029] In the process of integrated circuit design, accurate timing analysis is of great significance for ensuring the correct function of circuits, optimizing performance and predicting power consumption. Designers can verify and optimize circuits based on the results of timing analysis to meet design requirements. Therefore, fast and accurate timing analysis methods are of great significance for guiding circuit timing optimization and shortening circuit design cycles.
[0030] Static Timing Analysis (STA) is a method to verify the timing performance of a design by checking all possible paths for timing violations. STA decomposes the design into timing paths, calculates the signal propagation delay along each path, and checks whether the design internal and input / output interfaces violate timing constraints. Each path can include elements such as startpoint, combinatorial logic circuit, and terminal node, where the start node is the starting point of the timing path, where data is initiated by the clock edge, or data can be available at a specific time. Each start node can be an input port or the clock pin of a register. Combinational logic circuits are elements without memory or internal state. Combinational logic circuits can include elements such as AND gates, OR gates, XOR gates, inverters, but do not contain flip-flops, latches, registers, and random access registers. The terminal node is the end of the timing path, where data is captured by the clock edge, or data must be available at a specific time, and the terminal node can be an output port or the data input pin of a register.
[0031] refer to Figure 1 Each timing path starts from the data transmission point (starting node A), passes through a series of D flip-flops and combinational logic circuits, and ends at the data capture point (terminal node Z). Figure 1 CLK in the figure represents the transmitting node of the clock signal. The combinational logic circuit between register 1 and register 2 may include AND gate A, AND gate B and NOR gate C. Figure 11 to 7 in the figure represent the input and output ports of AND gate A, AND gate B and NOR gate C respectively. The starting part of the timing path Clk to Q is assumed to be fixed, and the remaining calculation includes starting from the output Q of register 1, reaching inverters A and B respectively through net 1, then reaching XOR gate C through net 2 and net 3, and finally reaching the input D of register 2 through net 4. Figure 1 The circuit shown can be modeled as Figure 2 In the directed acyclic graph (DAG) shown in FIG. 1 , the primary input & output port (PIO) or pins of the driver and receiver in the circuit are represented as nodes in the DAG graph. The timing arcs inside the wire net connection or the logic gate correspond to the edges in the DAG graph, and the weights of the edges correspond to the delays of the timing arcs. It can be understood that the timing arcs representing the wire net connection can be called net arcs, and the timing arcs representing the connection from the input pin inside the logic gate to the output pin of the logic gate can be called cell arcs. In the timing path Q→D, the total delay is completed by alternating the sum of the gate delay (Gate Delay) and the wire delay (Wire Delay). The gate delay refers to the delay amount of the logic gate from the input end to the output end in the path (that is, the delay amount of the cell arc), which is generally calculated based on the standard cell library provided in the logic library. For example, Figure 1 1→3 in inverter A. Line delay is the delay from the output of a logic gate to the input of the next logic gate in the timing path (i.e., the delay of the line arc). This line delay is caused by complex factors such as the resistance and capacitance (RC) interconnected between the two logic gates and the driving strength of the driving network. For example, Figure 1 Q in the inner network 1 is 1. In addition, factors such as different equivalent models and propagation methods also have a great influence on the calculation of line delay accuracy.
[0032] In the DAG graph, all nodes can be topologically sorted and traversed by layer. The arrival time (AT) of the node can be calculated based on the sum of the gate delay and the line delay. In addition, the required arrival time (RAT) of each node is calculated in reverse based on the required time of the terminal node. Finally, based on the AT and RAT of the node, the difference between the two is calculated to obtain the time margin (Slack) of the node. When Slack<0, there is a timing violation in the circuit, and optimization operations need to be performed in the subsequent design. Otherwise, it means that this timing path meets the timing constraints.
[0033] The inventors have found through research that in the pre-wiring stage, due to the lack of specific RC information, the prediction of line delay is more complicated than the prediction of gate delay, because the former is highly dependent on wiring information and multiple process corners (Process, Voltage and Temperature, PVT) conditions. However, in actual design, early timing analysis usually only focuses on a few PVT corners to save running time, which may cause many real violations to be hidden in unobserved PVT corners. Line delay is extremely sensitive to different wire lengths, densities and wiring distributions. When considering changes across multiple PVT corners, it becomes more difficult to achieve accurate and effective timing estimation. For example, taking the 28nm technology node as an example, there may be a significant 150ps timing gap between the timing analysis in the pre-wiring stage and the timing analysis in the sign-off stage using Elmore delay estimation, which is mainly due to the lack of wiring information and over-simplification in the Elmore model. Furthermore, under the same process and temperature conditions, the path timing difference between different voltages (e.g., 0.99V and 0.81V) can be as high as 80ps, while the impact of different PVT variations manifests itself as complex and non-monotonic characteristics.
[0034] In actual design, circuit design relies on worst-case assumptions combined with overly pessimistic timing margins to meet performance specifications under the most extreme PVT variations. Industry experience shows that the gap between the performance predicted by EDA tools and the actual silicon performance can be as high as 30%. Therefore, after routing, design optimization may encounter significant differences with wire delay information, resulting in multiple design iterations and poor trade-offs in performance, power consumption, and area when sign-off is completed.
[0035] In order to solve this problem, the inventors have found through research that the wiring tree of the target circuit can be obtained based on the target circuit represented by a DAG graph or the like, the target circuit can be divided into multiple two-end connection relationships, and the RC information of the wiring can be added to the two-end connection relationships based on the multi-process corner features to obtain a look-ahead tree; thus, based on the wiring tree describing the pre-wiring style of the circuit in the netlist file, a look-ahead tree describing the RC information of the wiring between multiple two-end connection relationships is obtained, thus, based on the look-ahead tree, RC information can be introduced in the timing analysis process of the pre-wiring stage, thus avoiding the problem of large timing analysis errors caused by the lack of RC information. Specifically, the line network's own characteristics can be obtained based on the look-ahead tree, the wiring tree and the layout input information, and the line network's own characteristics can be used as part of the line delay timing characteristics for timing analysis, thus, the line network's own characteristics not only consider the layout input information and the related information in the wiring tree, but also consider the RC information of the wiring between the two-end connection relationships in the look-ahead tree, thus solving the problem of the lack of RC information in the pre-wiring stage due to the lack of actual wiring information, which in turn leads to inaccurate timing analysis (especially line delay analysis) in the pre-wiring stage.
[0036] In addition to RC information, the inventor further analyzed the features that affect the accuracy of timing analysis and found that in addition to the characteristics of the wire net itself, the neighbor wire nets are physically close to the positions of the target wire nets in the target circuit, and the layout and other related features of the neighbor wire nets will also affect the timing of the target wire nets in the target circuit, that is, the neighbor wire net characteristics also have a significant impact on the accuracy of timing analysis. Therefore, based on the wiring congestion and coupling information related to the target circuit described by the neighbor wire net characteristics, the ideal wiring results in the wiring tree can be assisted to correct, and the delay impact of the neighbor wire net on each path in the target circuit can be comprehensively considered, which is conducive to improving the accuracy of timing analysis.
[0037] Based on the above concept, an embodiment of the present specification provides a timing analysis method. The timing analysis method provided in the embodiment of the present specification will be exemplarily described below in conjunction with the accompanying drawings.
[0038] Exemplary Methods
[0039] The embodiments of this specification provide a timing analysis method, such as Figure 3 As shown, including:
[0040] S301: Based on the target information of the target circuit in the netlist file, the line delay timing characteristics of the target circuit are obtained, the target information includes: a look-ahead tree, a routing tree and layout input information; the routing tree is used to describe the pre-wiring style of the circuit in the netlist file; the look-ahead tree is obtained based on the process corner characteristics and the routing tree, the look-ahead tree is used to describe the routing information between the two ends of the connection relationship in the routing tree, the routing information is used to describe the RC information of the routing between the two ends of the connection relationship, and the two ends of the connection relationship are used to characterize the connection relationship between two nodes in the circuit; the layout input information is used to describe the location information of the node, and the node is used to characterize the input and output ports or pins; the line delay timing characteristics include: the line net's own characteristics and the neighboring line net characteristics; the line net's own characteristics are obtained based on the look-ahead tree, the routing tree and the layout input information, the neighboring line net characteristics are obtained based on the layout input information, and the neighboring line net characteristics are used to describe the routing congestion and coupling information related to the target circuit;
[0041] The netlist file may include multiple circuits, including various types of circuits such as adder circuits, multiplier circuits, and comparator circuits. The target circuit may refer to a circuit in the netlist file that is to be subjected to timing analysis.
[0042] In the circuit, the driver may include circuit elements for enhancing or controlling the transmission of signals on the wire network; the receiver may include circuit elements for receiving or processing signals transmitted through the wire network, and the target receiver may include the predetermined end point element of the signal transmission.
[0043] In the target information, the wiring tree can be obtained by processing the netlist file using algorithms such as the Minimum Spanning Tree (MST). The pre-wiring style of each circuit is described in the wiring tree. The pre-wiring style can include the connection method between each input and output port or pin in the circuit to achieve the purpose of pre-wiring. The wiring tree obtained using the minimum Steiner tree can also be called a Steiner tree. The Steiner tree can obtain the shortest connection method for each input and output port or pin in the circuit, which is conducive to reducing the wiring length.
[0044] However, the wiring tree describes the pre-wiring style of the circuit, but does not include the RC information in the circuit. Based on the pre-set process corner information and the wiring tree, a look-ahead tree that describes the wiring information between the two end connection relationships in the wiring tree can be obtained to implement the look-ahead wiring of the circuit based on the look-ahead tree. During the timing analysis process, relevant features containing RC information can be obtained based on the look-ahead tree to implement accurate timing analysis.
[0045] The layout input information may be used to describe the position information of each node in the circuit, and specifically may describe the position of the node in the circuit and the positional relationship between the node and other adjacent nodes of other circuits in the form of coordinates or the like.
[0046] Based on the look-ahead tree, the routing tree and the layout input information, the characteristics of the line network itself can be obtained, so that the characteristics of the line network itself can obtain characteristics related to the circuit RC information from the look-ahead tree. In the subsequent timing analysis process, the impact of the circuit RC information on the delay can be comprehensively considered, which is conducive to improving the accuracy of the timing analysis.
[0047] Based on the layout input information, the neighbor net characteristics can be obtained, and the neighbor net characteristics are used to describe the routing congestion and coupling information related to the target circuit. In this way, the routing congestion and coupling information related to the target circuit described by the neighbor net characteristics can be used to assist in correcting the ideal routing results in the routing tree, and comprehensively consider the impact of the neighbor net on the delay of each path in the target circuit, which is conducive to improving the accuracy of timing analysis.
[0048] S302: Based on the line delay timing characteristics and the process corner characteristics, perform delay analysis on the target circuit to obtain the arrival time of each of the nodes in the target circuit, and based on the arrival time of each of the nodes in the target circuit and the required arrival time corresponding to each of the nodes, obtain the timing analysis results of each path in the target circuit.
[0049] As mentioned above, the arrival time of a node can refer to the sum of the line delay and the gate delay in the process of a signal starting from the node and reaching the target receiver through timing analysis, that is, the theoretical time required for the signal to start from the node and reach the target receiver. Figure 1 and Figure 2 For example, the arrival time of the signal from node 3 to node D may include: the line delay of network 2, the gate delay of nodes 5→7, and the line delay of network 4. The required arrival time of the node may refer to the design time required for the signal to reach the target receiver from the node during circuit design, that is, the signal needs to reach the target receiver from the node before the required arrival time. Assuming that the required arrival time from node 3 to D is 100ps, and through timing analysis, it is found that the arrival time from node 3 to D is 110ps, then the time margin of node 3 = 100ps-110ps = -10ps < 0, which means that there is a timing violation in the path from node 3 to D, and the signal cannot reach D from node 3 before the required arrival time, and the circuit needs to be optimized. Assuming that through timing analysis, it is found that the arrival time from node 3 to D is 90ps, then the time margin of node 3 = 100ps-90ps = 10ps > 0, which indicates that the path meets the timing constraint.
[0050] In this embodiment, when the timing analysis method performs timing analysis in the pre-wiring stage, based on the target information of the target circuit, the line delay timing characteristics of the target circuit are obtained, wherein the target information includes: a look-ahead tree, a wiring tree and layout input information, the netlist file includes multiple wire nets, the wire net includes multiple nodes and multiple edges, the nodes are used to represent the input and output ports or pins of the driver or receiver in the wire net, and the edges are used to represent the timing arcs between two nodes; the wiring tree is used to describe the pre-wiring style of the circuit in the netlist file; the look-ahead tree is obtained based on the process corner characteristics and the wiring tree, the look-ahead tree is used to describe the wiring information between the connection relationship at both ends in the wiring tree, the wiring information is used to describe the RC information of the wiring between the connection relationship at both ends, and the two ends are connected The relationship is used to characterize the connection relationship between two nodes in the circuit; the layout input information is used to describe the location information of the node; the line delay timing characteristics include: the line network's own characteristics and the neighboring line network characteristics; the line network's own characteristics are obtained based on the look-ahead tree, the routing tree and the layout input information, the neighboring line network characteristics are obtained based on the layout input information, and the neighboring line network characteristics are used to describe the routing congestion and coupling information related to the target circuit; then, based on the line delay timing characteristics and the process corner characteristics, the target circuit is subjected to a delay analysis to obtain the arrival time of each of the nodes in the target circuit, and based on the arrival time of each of the nodes in the target circuit and the required arrival time corresponding to each of the nodes, the timing analysis results of each path in the target circuit are obtained. In this way, the line network's own characteristics obtained during the timing analysis process not only take into account the layout input information and the relevant information in the wiring tree, but also take into account the RC information of the wiring between the two end connection relationships in the look-ahead tree, thereby solving the problem of inaccurate timing analysis (especially line delay analysis) in the pre-wiring stage due to the lack of actual wiring information in the pre-wiring stage. In addition, the line delay timing characteristics also include neighbor line network characteristics, which enables the wiring congestion and coupling information related to the target circuit described based on the neighbor line network characteristics to assist in correcting the ideal wiring results in the wiring tree, comprehensively considering the delay impact of the neighbor line network on each path in the target circuit, and is conducive to improving the accuracy of timing analysis.
[0051] In an optional implementation, in order to improve the generalization capability of performing timing analysis on a target circuit based on the line delay timing characteristics and the process corner characteristics during the timing analysis process, and to ensure that timing analysis results of each path can be obtained when the line delay timing characteristics and the process corner characteristics are different, performing delay analysis on the target circuit based on the line delay timing characteristics and the process corner characteristics to obtain the arrival time of each node in the target circuit, and obtaining the timing analysis results of each path in the target circuit based on the arrival time of each node of the target circuit and the required arrival time corresponding to each node includes:
[0052] Inputting the line delay timing characteristics and the process corner characteristics into a pre-trained target model, so as to use the target model to perform delay analysis on the target circuit, obtain the arrival time of each of the nodes in the target circuit, and obtain the timing analysis result of each path in the target circuit based on the arrival time of each of the nodes in the target circuit and the required arrival time corresponding to each of the nodes;
[0053] The target model is obtained by training with target training samples, and the target training samples include target training information and annotation information corresponding to the target training information, the target training information includes the line delay timing characteristics of the training circuit, and the annotation information includes the timing analysis results of each path in the training circuit obtained based on the timing report generated in the sign-off stage.
[0054] refer to Figure 4 , Figure 4 shows an overview diagram of the timing analysis method based on the target model. Figure 4 The figure shows a feasible process diagram of the target model training and prediction process. In order to distinguish the training process from the prediction process, the training circuit and the target circuit are used to indicate the circuits with different uses in different stages. The training circuit can refer to the circuit used for model training. In some embodiments, the training circuit can be obtained from a public circuit data set. In the training process, the input information may include the technology library, the netlist file, the layout input information ( Figure 4The process library may include information such as the electrical characteristics and physical dimensions of basic components, and information such as gate delays in the circuit may be obtained based on the process library. The netlist file includes related information such as identification, type, parameters, and electrical connection information of components in multiple circuits. Based on the netlist file, the wiring tree in the target information may be obtained, the neighboring wire net features may be obtained based on the layout input information, and the wire delay timing features may be obtained based on the target information. In the feature extraction step, the wire delay timing features of the target circuit may be obtained based on the target information. The wire delay timing features may include the wire net's own features and the neighboring wire net features. The wire delay timing features and the process corner features ( Figure 4 (represented by PVT corner features in the figure). After feature extraction is completed, model training or prediction is performed based on the extracted features. In one embodiment, during the training process, unlike the single process corner model obtained by training based on a typical process corner (referred to as the typical corner), the target model can be obtained based on multi-process corner feature training, that is, the target model is a multi-process corner model, which can support the output of timing analysis results based on multi-process corner feature input to achieve comprehensive consideration of multi-process corner situations. The output timing analysis results may include information such as the arrival time, required arrival time, time margin, and identified critical paths that violate time constraints corresponding to each node in the target circuit. In this way, the timing analysis method can support timing analysis based on multi-process corner features, avoiding the problem of large errors in timing analysis of paths due to different process corner features.
[0055] In a feasible implementation, in order to improve the generalization ability of the target model obtained through training and ensure that the target model has the ability to output accurate timing analysis results when encountering features that are not encountered during training, during the training process of the target model, the target training samples are divided into a training set and a validation set, the target model is trained based on the training set, and the target model is validated based on the validation set;
[0056] The line delay timing characteristics of the training circuit included in the target training information in the training set are at least partially different from the line delay timing characteristics of the training circuit included in the target training information in the validation set;
[0057] and / or
[0058] The process corner features of the training circuit included in the target training information in the training set are at least partially different from the process corner features of the training circuit included in the target training information in the validation set.
[0059] In this way, by setting at least some different features in the training set and the validation set, it is verified whether the generalization ability of the target model obtained through training meets the requirements, and by means of hyperparameter adjustment in the verification stage, a target model that meets the generalization ability requirements is obtained.
[0060] In an optional embodiment, the target model may include a LightGBM model (LightGradient Boosting Machine, lightweight gradient boosting decision tree model).
[0061] Through the testing of RF (Random Forest), XGBoost (Extreme Gradient Boosting) and LightGBM models, it was found that under the same feature set, the time series prediction correlation between the three models changed the least, and the difference was less than 0.01. However, the training time of the LightGBM model can be shortened to less than 50 seconds, which is much shorter than other models. The training process of RF and XGBoost models is more than three times the training time of the LightGBM model. This is because the LightGBM model uses a histogram algorithm to discretize features, thereby reducing the computational cost of building a decision tree. In addition, the LightGBM model also uses a leaf-based growth strategy, which results in a shallower generated tree and more efficient training time. Therefore, in this embodiment, using the LightGBM model as the target model is conducive to shortening the training time.
[0062] Regarding the neighboring network features, an optional implementation of this specification provides a feasible determination method, and the process of determining the neighboring network features includes:
[0063] Based on the layout input information, determining the bounding box of each target net in the target circuit and the bounding box of other nets; the bounding box is used to define the range of the net; the other nets are located in other circuits outside the target circuit;
[0064] Determine neighboring nets of each of the target nets based on the bounding box of the target net and bounding boxes of other nets, wherein the bounding box of the neighboring net at least partially overlaps with the bounding box of the target net;
[0065] Based on each of the target nets and the neighbor nets, the neighbor net features are obtained.
[0066] refer to Figure 5 , Figure 5A schematic diagram of a bounding box of a target network and bounding boxes of other networks is shown. The bounding box of the target network can be determined based on the location of each node in the target network. Similarly, the bounding boxes of other networks can be determined based on the location of each node in other networks. Figure 5 In the embodiment, the target circuit may include multiple target nets, and one target net may include a driver and all fan-outs of the driver. For example, 1→3 may be one target net, and Q→1 and Q→2 may be another target net. In some embodiments, other nets may refer to nets that have no direct connection relationship with the target net in the target circuit. Figure 5 The topological structure of other nets is not shown in the figure, only the positional relationship between the bounding box of other nets and the bounding box of the target net is shown. Figure 5 In the example, since the bounding box of other net 1 overlaps with the bounding box of the target net, other net 1 can be called a neighbor net of the target net, while other net 2 does not overlap with the target net, so other net 2 is not a neighbor net of the target net.
[0067] Since the physical layout position of the neighboring net is close to or overlaps with the physical layout position of the target net, the signal transmission and node layout in the neighboring net may affect the routing congestion and coupling information of the target net. Determining the neighboring nets of each target net in the above manner is simple and fast.
[0068] In one embodiment, specific feasible feature types of neighboring network features and network characteristics are provided. Specifically, the neighboring network features include:
[0069] at least one of the number of neighboring wire nets, the number of long-range neighboring wire nets, the area of the largest overlapping region, the average area of the overlapping regions, the distance between the driver and the geometric center of the largest overlapping region, the average distance between the driver and the geometric centers of all overlapping regions, the product of the rectangular uniform wire density of the target wire net, and the rectangular uniform wire density of the target wire net taking into account the long-range wire lengths of the neighboring wire nets;
[0070] The long-distance neighbor network includes neighbor networks whose physical distance exceeds a preset distance threshold among neighbor networks;
[0071] The overlapping area includes an overlapping area of a bounding box of the target net and a bounding box of the neighbor net;
[0072] The driver includes circuit elements for enhancing or controlling the transmission of signals on the line network;
[0073] The long-range wire length includes the first 30% of the wire length in the wire network.
[0074] The line network characteristics include characteristics that affect the path delay in the target line.
[0075] The process of determining the characteristics of the wire mesh itself includes:
[0076] Obtaining a first type of feature based on the forward-looking tree;
[0077] Obtaining a second type of feature based on the wiring tree;
[0078] Obtaining a third type of feature based on the layout input information;
[0079] The wire mesh characteristics include at least one of the first type of characteristics, the second type of characteristics and the third type of characteristics;
[0080] The first type of features includes: at least one of the speed of the rising edge or falling edge of the signal at the input pin of the target receiver, the speed of the rising edge or falling edge of the signal at the input pin of the driver output pin, the capacitance value of the input pin of the target receiver, the load capacitance of the driver, the Elmore delay between the driver and the target receiver, the sum of the Elmore delays between the driver and other receivers, and the second-order moment of the impulse response;
[0081] The second type of features includes: at least one of a driving strength and a number of receiving pins of each target line net in the target circuit;
[0082] The third category of features includes at least one of: a pre-routing distance between a driver output pin location and a target receiver input pin location, a sum of routing requirements in a path traversed by pre-routing of a target net, and a maximum distance from a driver output pin to all receiver input pins;
[0083] The receiver includes a circuit element for receiving or processing a signal transmitted through a wire network, and the target receiver includes a predetermined destination element for signal transmission.
[0084] Refer to Table 1 below, which shows the types of features that may be included in the line delay timing features and process corner features:
[0085] Table 1 Line delay timing characteristics and process corner characteristics
[0086]
[0087]
[0088] Among them, in the neighbor network features:
[0089] (1) Neigh-num: The number of neighboring networks.
[0090] (2) Long-neigh-num: The number of long-distance neighbor lines.
[0091] (3)Neigh-overlap-max / mean: the area of the maximum overlapping region / the average area of the overlapping regions.
[0092] (4) Neigh-dist-max / mean: the distance between the driver and the geometric center of the maximum overlapping area / the average distance between the driver and the geometric centers of all overlapping areas.
[0093] (5)RUDY: The product of the rectangular uniform wire density of the target net, used as an early routing congestion estimate.
[0094] (6) Long-RUDY: The RUDY value of the target network only considers the long-range line lengths of its adjacent networks, that is, the first 30% of the line lengths.
[0095] Among the online network’s own characteristics:
[0096] (1) Sink-slew: The speed (Slew) of the rising or falling edge of the signal at the target receiver input pin can be calculated using the Elmore delay model based on its associated pulse and load capacitance.
[0097] (2) Driver-slew: The speed of the rising or falling edge of the signal at the input pin of the driver output pin.
[0098] (3) Sink-cap: The capacitance value of the target receiver input pin.
[0099] (4) Driver-cap: The load capacitance of the driver.
[0100] (5) Elmore: Elmore delay between the driver and the target receiver. This feature can calculate the Elmore delay between the driver and the target receiver by obtaining the load capacitance and resistance from the look-ahead tree.
[0101] (6) Context-Elmore: In addition to the target sink, other sinks are called Context sinks. Context-Elmore represents the sum of the Elmore delays between the driver and the Context sinks.
[0102] (7)D2M: second-order moment of the impulse response.
[0103] (8) Driver-strength: Driver strength, which can be extracted from the cell type, determines the output impedance of the driver. Generally, the higher the driver strength, the smaller the delay.
[0104] (9) Fanout-num: the number of receiving pins of each target net in the target circuit.
[0105] (10) Distance: The pre-routing distance between the driver output pin location and the target receiver input pin location.
[0106] (11)Steiner-demand: is the sum of the routing requirements in the paths that the pre-routing of the target network passes through.
[0107] (12) Max-dist: The maximum distance from the driver output pin to all receiver input pins.
[0108] For the process corner feature:
[0109] (1) Temperature (e.g. -40°, 0°, 25° and 125°).
[0110] (2) Voltage (for example: 0.81V, 0.9V and 0.99V).
[0111] (3) RC (for example: RC best, RC worst, C best, C worst, etc.)
[0112] Through further research on the above features, it is found that different features have different degrees of influence on delay. In order to reflect the importance of different features in the timing analysis process (the features with higher degree of influence on delay have higher importance in the timing analysis process), in one embodiment of the present specification, the delay analysis is performed on the target circuit based on the line delay timing feature and the process corner feature to obtain the arrival time of each node in the target circuit, which includes:
[0113] Based on the line delay timing characteristics, the weights corresponding to the line delay timing characteristics, and the process corner characteristics, a delay analysis is performed on the target circuit to obtain the arrival time of each of the nodes in the target circuit.
[0114] When the target model is used to perform delay analysis on the target circuit based on the line delay timing characteristics and the process corner characteristics, different weights can be set for neurons that receive different input characteristics during the training process of the target model to achieve the purpose of setting different weights corresponding to different line delay timing characteristics. In some embodiments, the process corner characteristics can also correspond to different weights to reflect the different degrees of influence of different process corner characteristics on delay analysis.
[0115] In this embodiment, by considering the weights corresponding to the line delay timing characteristics during the delay analysis process, the different degrees of influence of different characteristics on the delay can be considered during the timing analysis process, which is conducive to improving the accuracy of the timing analysis.
[0116] Specifically, in a feasible implementation, the weights of some line delay timing features are provided. For example, for the following features: the speed of the rising edge or falling edge of the signal of the target receiver input pin, the speed of the rising edge or falling edge of the signal of the input pin of the driver output pin, the average area of the overlapping area, the distance between the driver and the geometric center of the maximum overlapping area, the sum of the wiring requirements in the path passed by the pre-wiring of the target line network, the capacitance value of the target receiver input pin, the maximum distance from the driver output pin to all receiver input pins, the product of the rectangular uniform line density of the target line network, the number of long-range neighboring line networks and the average distance between the driver and the geometric center of all overlapping areas, the weights of these features include:
[0117] The weights corresponding to the speed of the rising or falling edge of the signal of the target receiver input pin, the speed of the rising or falling edge of the signal of the input pin of the driver output pin, the average area of the overlapping area, the distance between the driver and the geometric center of the maximum overlapping area, the sum of the wiring requirements in the path passed by the pre-wiring of the target line network, the capacitance value of the target receiver input pin, the maximum distance from the driver output pin to all receiver input pins, the product of the rectangular uniform line density of the target line network, the number of long-range neighboring line networks and the average distance between the driver and the geometric center of all overlapping areas decrease in sequence.
[0118] The inventors found through constructing a target model and conducting experimental evaluation that the above ten line delay timing characteristics have a great influence on the delay (especially line delay) in the timing evaluation process, and evaluated the influence of the above ten line delay timing characteristics on the delay.
[0119] Specifically, the timing analysis method is implemented based on C++17 programming language, the training environment of the target model uses Python3.8 environment, and is evaluated on a server equipped with Intel Xeon Core (@3.20GHz) and 256GB DDR4 memory. The experimental introduction, result analysis description and detailed discussion are as follows:
[0120] The experiments were conducted on the ITC'99 (International Test Conference 1999) benchmark circuit, CV32E40P, and CVA6, using a 28nm industrial process. Each design was first synthesized by Synopsys design tools, then physically designed (P&R) by Cadence Innovus, and then the circuit was imported into Synopsys StarRC for parasitic extraction. Finally, Synopsys Prime Time (a static timing analysis tool) was called in SI (Signal Integrity) mode to generate a timing report as the Ground Truth. The circuit information is summarized in Table 2. In order to obtain enough wire net samples for training, we adjusted several tool parameters and generated 4 different ITC'99 circuit design implementations and 3 different CV32E40P circuit design implementations.
[0121] Table 2 Circuit information summary
[0122]
[0123]
[0124] In Table 2, the number of cells refers to the number of basic logic gates or functional modules in the standard cell library in the circuit, and the number of FFs (Flip-Flop Count) refers to the number of flip-flops. The number of nets refers to the number of target nets in the circuit, and the number of samples refers to the number of training samples obtained from the circuit.
[0125] To evaluate the performance, we first predict the timing of typical process corners. Considering the diversity of the dataset, we selected the CVA6 circuit as the training set, while other circuits were used for testing. During the experiment, the LightGBM model was used to evaluate the importance of the line delay timing features mentioned above.
[0126] Table 3 Typical process corner end-of-line delay prediction results
[0127]
[0128] In Table 3, the analysis results obtained by the timing analysis method provided in the embodiments of the present specification are compared with the corresponding results obtained by other methods (i.e., "DAC'20" and "MLCAD'22", both of which use the XGBoost model). "DAC'20" may refer to the timing analysis method proposed in the Design Automation Conference held at the Design Automation Conference (DAC) in 2020, and "MLCAD'22" may refer to the timing analysis method based on the XGBoost model proposed in Machine Learning in Computer-Aided Design (MLCAD) in 2022. It should be noted that in "DAC'20" and "MLCAD'22", the proposed timing features are all extracted in the routing stage. In order to fairly compare the methods "DAC'20" and "MLCAD'22", all results are generated by using the features proposed by them.
[0129] In Table 3, "Ours" represents the timing analysis method provided in the embodiment of this specification, and "Mean", "Max" and "Cor" are the average error, maximum error and correlation in the timing analysis report in SI mode, respectively. For a total of 4,814K training data from circuit CVA6 as shown in Table 2, the training time of "Ours" is about 50s. It can be seen from Table 3 that the correlation of "Cor" of "Ours" is improved by about 6% on average compared with "DAC'20" and "MLCAD'22". In addition, the "Mean" and "Max" errors predicted by the timing analysis method provided in the embodiment of this specification are also small.
[0130] Compared to previous work in the routing phase, timing prediction in the layout phase lacks accurate routing information. Timing features that are only effective in the routing phase, namely wire length, must be replaced by inaccurate features such as Steiner tree line length, which only reflects the ideal routing results. In feature extraction, other routing factors such as routing congestion and coupling effects must be explicitly considered. In the timing analysis method provided in the embodiments of this specification, features related to neighboring wire nets are added to provide more routing congestion and coupling information.
[0131] The importance of a feature can be evaluated from different perspectives. A commonly used metric is called "Gain", which quantifies the degree of accuracy improvement caused by a specific feature. Through experimental analysis, it is found that during the pre-routing stage, the top ten features that have the greatest impact on the accuracy of timing analysis are: Figure 6As shown, and this is a feature sorted by importance, Figure 6 In the figure, the horizontal axis is "Gain". The larger the value, the greater the influence of the corresponding feature on the line delay. The vertical axis is the line delay timing feature (Features of Wire Delay). Figure 6 It can be found that in Figure 6 Among the ten line delay timing features shown, Sink-slew: "the speed of the rising or falling edge of the signal at the target receiver input pin" has the greatest impact on the line delay timing. In addition, some neighbor line network features also have a relatively important impact on the analysis of line delay, such as: Neigh-overlap-mean: the average area of the overlapping area, Neigh-dist-max: the distance between the driver and the geometric center of the maximum overlapping area, RUDY: the product of the rectangular uniform line density of the target line network, Long-neigh-num: the number of long-range neighbor line networks, Neigh-dist-mean: the average distance between the driver and the geometric center of all overlapping areas, etc.
[0132] In the multi-process corner timing analysis experiment, we used the CVA6 circuit and evaluated the generated samples of six PVT corners under different PVT conditions. These PVT corners include ml-rcworst, wcl-rcworst, bc-rcworst, tc-rcworst, wc-rcworst, and wcz-rcworst, with a temperature range of -40℃ to 125℃ and a voltage range of 0.81V to 0.99V.
[0133] The results of the multi-process corner line delay analysis are shown in Table 4:
[0134]
[0135] Table 4 shows the line delay prediction results of CVA6 under multiple process corners. The training data set includes two process corners: ml-rcworst (Maximal Leakage Resistance-
[0136] Capacitance worst) and wcl-rcworst (Worst Case Cold Resistance-Capacitance worst), while other process corners are used as test data sets; among them, "ml" can refer to the maximum leakage process corner, which means that the leakage current of the device is the largest and the device speed is the fastest under the highest temperature, highest voltage and fastest process conditions; "rcworst" means the worst case of parasitic effects (Resistance-Capacitance, RC for short), that is, the combination of the highest resistance and the largest capacitance is considered, which will lead to increased signal transmission delay; "wcl" can refer to the Worst Case Cold process corner, which means that the device performance is the worst under the lowest temperature, lowest voltage and slowest process conditions. The timing characteristics of all methods, namely, "DAC'20", "MLCAD'22" and "Ours", include process corner characteristics. For the same circuit, different process corners show strong correlation. Experiments show that when a limited number of process corners are used for training, the correlation of the prediction results of the three methods under other process corners is above 0.93. Due to the effective feature extraction of the "Ours" method, the "Cor" of all test corners is always over 0.98. Compared with the "DAC'20" and "MLCAD'22" methods, the timing analysis method provided by the embodiment of this specification has an improvement of more than 4% in "Cor", and the "Max" and "Mean" errors are reduced by 67% and 43% on average, respectively.
[0137] refer to Figure 7 , Figure 7 The comparison result of the line delay with the test angle wc-rcworst is shown. The timing analysis method ("Ours") provided in the embodiment of this specification is closer to the Sign-Off result (i.e., the timing analysis result performed in the Sign-Off stage, Figure 7 In particular, for long-line networks with a delay of more than 100 ps, the delay prediction error distribution of the timing analysis method provided in the embodiments of this specification is more concentrated around 0. However, the error distribution range of "DAC'20" and "MLCAD'22" for long-line networks is larger than that for short-line networks. Figure 7 In the figure, the horizontal axis is the delay (Sign-Off Delay) in the Sign-Off result, in ps, and the vertical axis is the predicted delay (Predict Delay) obtained by timing analysis using different methods (DAC'20, MLCAD'22 and Ours), in ps.
[0138] Table 5 Path arrival time analysis under multiple process angles
[0139]
[0140] Referring to Table 5, we integrate the line delay prediction results into OpenTimer for path timing analysis. By traversing the path level by level in topological order, the endpoint arrival time can be obtained. Table 5 shows the arrival time performance results. Compared with "DAC'20" and "MLCAD'22", the timing analysis method provided in the embodiment of this specification has the highest "Cor" and the "Mean" error is 12ps smaller on average.
[0141] Exemplary Devices
[0142] In an exemplary embodiment of the present specification, a timing analysis device is further provided, the timing analysis device comprising:
[0143] A feature extraction module is used to obtain the line delay timing characteristics of the target circuit based on the target information of the target circuit in the netlist file, the target information includes: a look-ahead tree, a wiring tree and layout input information; the wiring tree is used to describe the pre-wiring style of the circuit in the netlist file; the look-ahead tree is obtained based on the process corner feature and the wiring tree, the look-ahead tree is used to describe the wiring information between the two-end connection relationship in the wiring tree, the wiring information is used to describe the RC information of the wiring between the two-end connection relationship, and the two-end connection relationship is used to characterize the connection relationship between two nodes in the circuit; the layout input information is used to describe the position information of the node, and the node is used to characterize the input and output port or pin; the line delay timing characteristics include: the line net itself characteristics and the neighbor line net characteristics; the line net itself characteristics are obtained based on the look-ahead tree, the wiring tree and the layout input information, the neighbor line net characteristics are obtained based on the layout input information, and the neighbor line net characteristics are used to describe the wiring congestion and coupling information related to the target circuit;
[0144] The delay analysis module is used to perform delay analysis on the target circuit based on the line delay timing characteristics and the process corner characteristics to obtain the arrival time of each node in the target circuit, and obtain the timing analysis results of each path in the target circuit based on the arrival time of each node in the target circuit and the required arrival time corresponding to each node.
[0145] For the specific definition of the timing analysis device, please refer to the definition of the timing analysis method above, which will not be repeated here. Each module in the above timing analysis device can be implemented in whole or in part by software, hardware and a combination thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0146] Exemplary Computing Devices
[0147] Another embodiment of this specification also provides a computing device, see Figure 8 As shown, an exemplary embodiment of the present specification also provides a computing device, including: a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the timing analysis method according to various embodiments of the present specification described in the above embodiments of the present specification are executed.
[0148] The internal structure of the computing device can be as follows Figure 8 As shown, the computing device includes a processor, a memory, a network interface and an input device connected through a system bus. Among them, the processor of the computing device is used to provide computing and control capabilities. The memory of the computing device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computing device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the steps of the timing analysis method according to various embodiments of the present specification described in the above embodiments of the present specification are performed.
[0149] The processor may include a main processor and may also include a baseband chip, a modem, etc.
[0150] The memory stores a program for executing the technical solution of the present invention, and may also store an operating system and other key services. Specifically, the program may include a program code, and the program code includes computer operation instructions. More specifically, the memory may include a read-only memory (ROM), other types of static storage devices that can store static information and instructions, a random access memory (RAM), other types of dynamic storage devices that can store information and instructions, a disk storage, a flash, and the like.
[0151] The processor may be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the scheme of the present invention. It may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component.
[0152] The input device may include a device for receiving data and information input by a user, such as a keyboard, a mouse, a camera, a scanner, a light pen, a voice input device, a touch screen, a pedometer, or a gravity sensor.
[0153] Output devices may include means that allow information to be output to a user, such as display screens, printers, speakers, etc.
[0154] The communication interface may include using any transceiver or the like to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.
[0155] The processor executes the program stored in the memory and calls other devices, which can be used to implement each step of any timing analysis method provided in the above embodiments of this specification.
[0156] The computing device may also include a display component and a voice component. The display component may be a liquid crystal display or an electronic ink display. The input device of the computing device may be a touch layer covered on the display component, or a button, trackball or touchpad provided on the housing of the computing device, or an external keyboard, touchpad or mouse.
[0157] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of this specification, and does not constitute a limitation on the computing device to which the scheme of this specification is applied. The specific computing device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.
[0158] Exemplary computer program products and storage media
[0159] In addition to the above-mentioned methods and devices, the timing analysis method provided in the embodiments of the present specification may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the timing analysis method according to various embodiments of the present specification described in the above “Exemplary Method” section of the present specification.
[0160] The computer program product may be implemented in hardware, software or a combination thereof. In one optional embodiment, the computer program product is implemented as a computer storage medium. In another optional embodiment, the computer program product is implemented as a software product, such as a software development kit (SDK).
[0161] The computer program product may be written in any combination of one or more programming languages to write program code for performing the operations of the embodiments of the present specification, including object-oriented programming languages such as Java, C++, etc., and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.
[0162] In addition, an embodiment of the present specification also provides a computer-readable storage medium on which a computer program is stored, and the computer program is executed by a processor to execute the steps of the timing analysis method according to various embodiments of the present specification described in the above "Exemplary Method" section of the present specification.
[0163] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this specification can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0164] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0165] The above-mentioned embodiments only express several implementation methods of this specification, and the descriptions are relatively specific and detailed, but they cannot be understood as limiting the scope of the solutions provided by the embodiments of this specification. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of this specification, which all belong to the protection scope of this specification. Therefore, the protection scope of the patent of this specification shall be based on the attached claims.
Claims
1. A timing analysis method, characterized in that: include: Based on the target information of the target circuit in the netlist file, the line delay timing characteristics of the target circuit are obtained, the target information includes: a look-ahead tree, a routing tree and layout input information; the routing tree is used to describe the pre-wiring style of the circuit in the netlist file; the look-ahead tree is obtained based on the process corner characteristics and the routing tree, the look-ahead tree is used to describe the routing information between the two-end connection relationship in the routing tree, the routing information is used to describe the RC information of the routing between the two-end connection relationship, and the two-end connection relationship is used to characterize the connection relationship between two nodes in the circuit; the layout input information is used to describe the position information of the node, and the node is used to characterize the input and output ports or pins; the line delay timing characteristics include: the line net itself characteristics and the neighbor line net characteristics; the line net itself characteristics are obtained based on the look-ahead tree, the routing tree and the layout input information, the neighbor line net characteristics are obtained based on the layout input information, and the neighbor line net characteristics are used to describe the routing congestion and coupling information related to the target circuit; Based on the line delay timing characteristics and the process corner characteristics, a delay analysis is performed on the target circuit to obtain the arrival time of each of the nodes in the target circuit, and based on the arrival time of each of the nodes in the target circuit and the required arrival time corresponding to each of the nodes, a timing analysis result of each path in the target circuit is obtained.
2. The method according to claim 1, characterized in that The process of determining the neighbor network characteristics includes: Based on the layout input information, determining the bounding box of each target net in the target circuit and the bounding box of other nets; the bounding box is used to define the range of the net; the other nets are located in other circuits outside the target circuit; Determine neighboring nets of each of the target nets based on the bounding box of the target net and bounding boxes of other nets, wherein the bounding box of the neighboring net at least partially overlaps with the bounding box of the target net; Based on each of the target nets and the neighbor nets, the neighbor net features are obtained.
3. The method according to claim 2, characterized in that The neighbor network features include: at least one of the number of neighboring wire nets, the number of long-range neighboring wire nets, the area of the largest overlapping region, the average area of the overlapping regions, the distance between the driver and the geometric center of the largest overlapping region, the average distance between the driver and the geometric centers of all overlapping regions, the product of the rectangular uniform wire density of the target wire net, and the rectangular uniform wire density of the target wire net taking into account the long-range wire lengths of the neighboring wire nets; The long-distance neighbor network includes neighbor networks whose physical distance exceeds a preset distance threshold among neighbor networks; The overlapping area includes an overlapping area of a bounding box of the target net and a bounding box of the neighbor net; The driver includes circuit elements for enhancing or controlling the transmission of signals on the line network; The long-range wire length includes the first 30% of the wire length in the wire network.
4. The method according to claim 1, characterized in that The line network characteristics include characteristics that affect the path delay in the target line.
5. The method according to claim 4, characterized in that The process of determining the characteristics of the wire mesh itself includes: Obtaining a first type of feature based on the forward-looking tree; Obtaining a second type of feature based on the wiring tree; Obtaining a third type of feature based on the layout input information; The wire mesh characteristics include at least one of the first type of characteristics, the second type of characteristics and the third type of characteristics; The first type of features includes: at least one of the speed of the rising edge or falling edge of the signal at the input pin of the target receiver, the speed of the rising edge or falling edge of the signal at the input pin of the driver output pin, the capacitance value of the input pin of the target receiver, the load capacitance of the driver, the Elmore delay between the driver and the target receiver, the sum of the Elmore delays between the driver and other receivers, and the second-order moment of the impulse response; The second type of features includes: at least one of a driving strength and a number of receiving pins of each target line net in the target circuit; The third category of features includes at least one of: a pre-routing distance between a driver output pin location and a target receiver input pin location, a sum of routing requirements in a path traversed by pre-routing of a target net, and a maximum distance from a driver output pin to all receiver input pins; The receiver includes a circuit element for receiving or processing a signal transmitted through a wire network, and the target receiver includes a predetermined destination element for signal transmission.
6. The method according to claim 1, characterized in that The performing delay analysis on the target circuit based on the line delay timing characteristics and the process corner characteristics to obtain the arrival time of each node in the target circuit includes: Based on the line delay timing characteristics, the weights corresponding to the line delay timing characteristics, and the process corner characteristics, a delay analysis is performed on the target circuit to obtain the arrival time of each of the nodes in the target circuit.
7. The method according to claim 6, characterized in that The line delay timing characteristics include: the speed of the rising edge or falling edge of the signal of the target receiver input pin, the speed of the rising edge or falling edge of the signal of the input pin of the driver output pin, the average area of the overlapping area, the distance between the driver and the geometric center of the maximum overlapping area, the sum of the wiring requirements in the path traversed by the pre-wiring of the target line net, the capacitance value of the target receiver input pin, the maximum distance from the driver output pin to all receiver input pins, the product of the rectangular uniform line density of the target line net, the number of long-range neighboring line nets and the average distance between the driver and the geometric center of all overlapping areas; The weights corresponding to the speed of the rising or falling edge of the signal of the target receiver input pin, the speed of the rising or falling edge of the signal of the input pin of the driver output pin, the average area of the overlapping area, the distance between the driver and the geometric center of the maximum overlapping area, the sum of the wiring requirements in the path passed by the pre-wiring of the target line network, the capacitance value of the target receiver input pin, the maximum distance from the driver output pin to all receiver input pins, the product of the rectangular uniform line density of the target line network, the number of long-range neighboring line networks and the average distance between the driver and the geometric center of all overlapping areas decrease in sequence.
8. The method according to claim 1, characterized in that The performing delay analysis on the target circuit based on the line delay timing characteristics and the process corner characteristics to obtain the arrival time of each node in the target circuit, and obtaining the timing analysis result of each path in the target circuit based on the arrival time of each node in the target circuit and the required arrival time corresponding to each node, includes: Inputting the line delay timing characteristics and the process corner characteristics into a pre-trained target model, so as to use the target model to perform delay analysis on the target circuit, obtain the arrival time of each of the nodes in the target circuit, and obtain the timing analysis result of each path in the target circuit based on the arrival time of each of the nodes in the target circuit and the required arrival time corresponding to each of the nodes; The target model is obtained by training with target training samples, and the target training samples include target training information and annotation information corresponding to the target training information, the target training information includes the line delay timing characteristics of the training circuit, and the annotation information includes the timing analysis results of each path in the training circuit obtained based on the timing report generated in the sign-off stage.
9. The method according to claim 8, characterized in that During the training process of the target model, the target training samples are divided into a training set and a validation set, the target model is trained based on the training set, and the target model is validated based on the validation set; The line delay timing characteristics of the training circuit included in the target training information in the training set are at least partially different from the line delay timing characteristics of the training circuit included in the target training information in the validation set; and / or The process corner features of the training circuit included in the target training information in the training set are at least partially different from the process corner features of the training circuit included in the target training information in the validation set.
10. A computing device, characterized in that The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the timing analysis method according to any one of claims 1 to 9 when executing the computer program.
11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the timing analysis method according to any one of claims 1 to 9 is implemented.
Citation Information
Cited By
Line length and line time delay collaborative prediction method after digital integrated circuit layout
CN120012669A
Load capacitance optimization method and system based on dichotomy and dynamic variable control
CN120449817A
Load capacitance optimization method and system based on dichotomy and dynamic variable control
CN120449817B
Pre-wiring RC automatic calibration method and system based on time sequence driving
CN121279241A
