Transformer-based end-to-end hotspot detection method for integrated circuit layouts

A Transformer-based end-to-end hotspot detection method for integrated circuit layouts, comprising the following steps:

• step 1. converting existing integrated circuit layout data into binary image data and storing the converted binary image data;
• step 2. inputting the converted binary image data into a GPU-driven lithography simulator, and outputting a post-lithography printed image and storing;
• and
• step 3. constructing a Transformer-based end-to-end hotspot detector, and inputting both the post-lithography printed image and an original binary image into a query vector initialization module for performing metric measurement, thereby obtaining positional information of potential hotspots, and storing;
• wherein the obtaining the positional information of the potential hotspots comprises the following steps:
• step 3.1. dividing the input original binary image and the post-lithography printed image into a plurality of non-overlapping blocks;
• step 3.2. calculating, for each block B(i,j), a pixel value difference at each pixel location within the blocks:
• D
• ⁡
• (
• i
• ,
• j
• ,
• x
• ,
• y
• )
• =
• I
• input
• (
• p
• )
• -
• I
• print
• (
• p
• )
• wherein Iinput(p) and Iprint(p) denote pixel intensities at pixel location p in the input original binary image and the post-lithography printed image, respectively;
• step 3.3. calculating a metric loss for each patch:
• calculating a metric loss Lmetric(i,j) for each block by summing squared differences of all pixels within the patches, as defined by the following formula:
• L
• metric
• (
• i
• ,
• j
• )
• =
• ∑
• x
• =
• 0
• N
• -
• 1
• ∑
• y
• =
• 0
• N
• -
• 1
• D
• ⁡
• (
• i
• ,
• j
• ,
• x
• ,
• y
• )
• 2
• ;
• wherein N denotes a block size, and (i,j) denotes a block index;
• step 3.4. constructing a loss map:
• aggregating metric losses from all blocks, to construct a loss map Lmetric map, wherein the loss map comprises metric loss values of each block;
• step 3.5. selecting the potential hotspots:
• identifying, within the loss map, top k blocks exhibiting highest loss values as the potential hotspots, and preserving the positional information corresponding to the potential hotspots within the blocks;
• step 3.6. integrating positional prior information:
• conveying location data of potential hotspot-containing blocks to a decoder as prior knowledge, directing decoder's attention toward regions exhibiting significant lithographic disparitie, thereby enhancing hotspot detection accuracy and operational efficiency;
• step 4. constructing, according to prior lithography information obtained from the original binary image and the post-lithography printed image, a layout training set and a validation set of a layout, inputting the sets into the Transformer-based end-to-end hotspot detector for model training, thereby obtaining a trained hotspot detection model;
• and
• step 5. inputting an integrated circuit layout under test, according to the positional information of the potential hotspots in a binary image layout obtained through step 2 and step 3, into the trained hotspot detection model of step 4 for testing;
• and locating hotspots within the layout.