# US 2023/0267265 A1 · Using machine-trained network to perform drc check

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2023/0267265 A1
- Document ID: US20230267265A1
- Status: Published application
- Kind: A1
- Application number: 18/097,272
- Filed: 2023-01-15
- Published: 2023-08-24
- Earliest claimed: 2020-10-22
- Assignees: D2S, Inc.
- Applicants: D2S, Inc.
- Inventors: Donald Oriordan
- Web page: https://patentgrep.com/patent/US20230267265A1

## CPC

- G06F 30/398: Design verification or optimisation, e.g. using design rule check [DRC], layout versus schematics [LVS] or finite element methods [FEM] (optical proximity correction [OPC] design processes G03F1/36)
- G06F 30/27: using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/09: Supervised learning
- H01J 37/3177: {Multi-beam, e.g. fly's eye, comb probe}
- H01J 2237/31761: Patterning strategy
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- H01J: ELECTRIC DISCHARGE TUBES OR DISCHARGE LAMPS (spark-gaps H01T; electric arc lamps H05B31/00; particle accelerators H05H)

## Abstract

A method for performing pixel-based design rule checking (DRC) is described. This method is used to perform design rule checks for rectilinear and curvilinear designs. In some embodiments, the pixel-based approach is based on computational deep-learning. The pixel-based DRC method of some embodiments is more resilient to false positives than traditional geometric approaches, particularly for designs with curvilinear content, and the inference time remains constant, regardless of how many shapes exist in the design being checked, or how many polygon edges are needed to represent its curvature. The DRC method of some embodiments is implemented by highly parallel architectures (such as Graphics Processing Units (GPU) and Tensor Processing Units (TPU)) to improve processing throughput compared to traditional means.

## Claims

### Claim 1

A method for performing design rule checking (DRC) on a design comprising a plurality of shapes, the method comprising:
receiving a first description of the design in a first non-pixelized format;
producing, from the first description, a second description of the design in a second pixelized format;
using the second description to provide input to a machine-trained network to process in order to identify a DRC violation in the design; and
based on output produced by the machine-trained network, identifying DRC violations in the design.

### Claim 2 (depends on 1)

The method of claim 1 wherein the DRC violations are initially expressed in a pixel-based format, the method further comprising generating, for the identified DRC violations that are specified in the pixel-based format, contoured shapes to display with the design, and displaying the design with the contoured shapes in order to identify locations in the design that have DRC violations.

### Claim 3 (depends on 2)

The method of claim 2, wherein the contoured shapes are displayed along with the design by a geometry-based design editing or visualization tool.

### Claim 4 (depends on 1)

The method of claim 1, wherein the machine-trained network is a neural network.

### Claim 5 (depends on 1)

The method of claim 1, wherein the shapes comprise rectilinear shapes and curvilinear shapes.

### Claim 6 (depends on 5)

The method of claim 5, wherein each rectilinear shape is formed by Manhattan edges, each curvilinear shape is formed by at least one curvilinear edge, and the shapes further comprise shapes with at least one non-Manhattan rectilinear edge that has a 45-degree angle or another angle other than 0, 45, or 90.

### Claim 7 (depends on 1)

The method of claim 1, wherein said producing comprises using the first description to rasterize the design to obtain the second pixelized format in which pixel values are used to describe the design.

### Claim 8 (depends on 7)

The method of claim 7, wherein the first description comprises a description of shapes as polygons.

### Claim 9 (depends on 1)

The method of claim 1, wherein the machine-trained network implements a DRC checking process.

### Claim 10 (depends on 1)

The method of claim 1, wherein the machine-trained network partially implements a DRC checking process along with a pixel-based process.

### Claim 11

The method of 10, wherein the pixel-based process comprises a morphological image-processing process.

### Claim 12

A non-transitory machine-readable medium storing a program, which when executed by at least one processing unit of a computer, performs design rule checking (DRC) on a design comprising a plurality of shapes, the program comprising sets of instructions for:
receiving a first description of the design in a first non-pixelized format;
producing, from the first description, a second description of the design in a second pixelized format;
using the second description to provide input to a machine-trained network to process in order to identify a DRC violation in the design; and
based on output produced by the machine-trained network, identifying DRC violations in the design.

### Claim 13 (depends on 12)

The non-transitory machine-readable medium of claim 12 wherein the DRC violations are initially expressed in a pixel-based format, the program further comprising a set of instructions for generating, for the identified DRC violations that are specified in the pixel-based format, contoured shapes to display with the design, and displaying the design with the contoured shapes in order to identify locations in the design that have DRC violations.

### Claim 14 (depends on 13)

The non-transitory machine-readable medium of claim 13, wherein the contoured shapes are displayed along with the design by a geometry-based design editing or visualization tool.

### Claim 15 (depends on 12)

The non-transitory machine-readable medium of claim 12, wherein the machine-trained network is a neural network.

### Claim 16 (depends on 12)

The non-transitory machine-readable medium of claim 12, wherein the shapes comprise rectilinear shapes and curvilinear shapes.

### Claim 17 (depends on 16)

The non-transitory machine-readable medium of claim 16, wherein each rectilinear shape is formed by Manhattan edges, each curvilinear shape is formed by at least one curvilinear edge, and the shapes further comprise shapes with at least one non-Manhattan rectilinear edge that has a 45-degree angle or another angle other than 0, 45 or 90.

### Claim 18

The non-transitory machine-readable medium of 12, wherein the machine-trained network produces a single output representative of design violations for a single design rule.

### Claim 19

The non-transitory machine-readable medium of 12, wherein the machine-trained network produces multiple outputs representative of design violations for multiple design rules.

### Claim 20 (depends on 19)

The non-transitory machine-readable medium of claim 19, wherein the multiple design rules comprise at least two of DRC rule constraints regarding widths of shapes, DRC rule constraints regarding spacing between two shapes, and DRC rule constraints regarding an amount by which one shape encloses another.
