Computing parasitic values for semiconductor designs
Some embodiments provide a method for calculating parasitic parameters for a pattern to be manufactured on an integrated circuit (IC) substrate. The method receives a definition of a wire structure as input. The method rasterizes the wire structure (e.g., produces pixel-based definition of the wire structure) to produce several images. Before rasterizing the wire structure, the method in some embodiments decomposes the wire structure into several components (e.g., several wires, wire segments or wire structure portions), which it then individually rasterizes. The method then uses the images as inputs to a neural network, which then calculates parasitic parameters associated with the wire structure. In some embodiments, the parasitic parameters include unwanted parasitic capacitance effects exerted on the wire structure. Conjunctively, or alternatively, these parameters include unwanted parasitic resistance and/or inductance effects on the wire structure.
1 . A method for calculating a parasitic effect of a set of neighboring wires on a first wire defined in a region of a semiconductor design, the method comprising:
dividing the region into a plurality of tiles so that different segments of the first wire fall within different tiles and in the different tiles the different segments of the first wire have different subset of neighboring wire segments;
performing a rasterization operation to produce a pixel-based definition for each tile, wherein each tile's pixel-based definition comprises a pixel-based representation of each wire segment in the tile;
supplying the pixel-based definition of each tile to a machine-trained network to produce a tile-specific parasitic value representing a parasitic effect on the wire segment of the first wire that falls within the tile; and
computing an overall parasitic value from the produced parasitic values, the overall parasitic value representing an overall parasitic effect of the set of neighboring wires on the first wire.
2 . The method of claim 1 , wherein computing the overall parasitic value comprises computing a sum of the tile-specific parasitic values over all of the tiles that include a segment of the first wire.
3 . The method of claim 1 , wherein the pixel-based representation of each wire segment in each tile comprises a pixel-based representation of any segment of the first wire that falls within the tile and any segment of any neighboring wire that falls within the tile.
4 . The method of claim 1 , wherein the machine-trained network is a neural network comprising a plurality of machine-trained neurons.
5 . The method of claim 4 , wherein the machine-trained network is a first machine-trained network, the method further comprising supplying a first wire structure that includes the first wire and the neighboring wires to a second machine-trained network to produce a second wire structure that has curvilinear wire segments.
6 . The method of claim 5 , wherein the first wire and the neighboring wires in the first wire structure only have rectilinear wire segments that are straight, while the first wire and at least a subset of the neighboring wires in the second wire structure have curvilinear wire segments that are curved.
7 . The method of claim 5 , wherein the rasterization operation is performed on the first wire structure before the first wire structure is supplied to the second machine-trained network.
8 . The method of claim 7 , wherein the second machine-trained network is a neural network comprising a plurality of machine-trained neurons.
9 . The method of claim 1 , wherein the wire segments of the first wire comprise at least one curvilinear segment that includes at least one curved portion.
10 . The method of claim 9 , wherein the wire segments of the neighboring wires comprise curvilinear segments that include at least one curved portion.
11 . The method of claim 10 , wherein the wire segments of the first wire and the neighboring wires further comprise rectilinear segments that include only straight segments.
12 . A non-transitory machine-readable medium storing a program which when executed by at least one processing unit calculates a parasitic effect of a set of neighboring wires on a first wire defined in a region of a semiconductor design, the program comprising sets of instructions for:
dividing the region into a plurality of tiles so that different segments of the first wire fall within different tiles and in the different tiles the different segments of the first wire have different subset of neighboring wire segments;
performing a rasterization operation to produce a pixel-based definition for each tile, wherein each tile's pixel-based definition comprises a pixel-based representation of each wire segment in the tile; and
supplying the pixel-based definition of each tile to a machine-trained network to produce a tile-specific parasitic value representing a parasitic effect on the wire segment of the first wire that falls within the tile;
computing an overall parasitic value from the produced parasitic values, the overall parasitic value representing an overall parasitic effect of the set of neighboring wires on the first wire.
13 . The non-transitory machine-readable medium of claim 12 , wherein the set of instructions for computing the overall parasitic value comprises a set of instructions for computing a sum of the tile-specific parasitic values over all of the tiles that include a segment of the first wire.
14 . The non-transitory machine-readable medium of claim 12 , wherein the pixel-based representation of each wire segment in each tile comprises a pixel-based representation of any segment of the first wire that falls within the tile and any segment of any neighboring wire that falls within the tile.
15 . The non-transitory machine-readable medium of claim 12 , wherein the machine-trained network is a neural network comprising a plurality of machine-trained neurons.
16 . The non-transitory machine-readable medium of claim 15 , wherein the machine-trained network is a first machine-trained network, the program further comprising a set of instructions for supplying a first wire structure that includes the first wire and the neighboring wires to a second machine-trained network to produce a second wire structure that has curvilinear wire segments.
17 . The non-transitory machine-readable medium of claim 16 , wherein the first wire and the neighboring wires in the first wire structure only have rectilinear wire segments that are straight, while the first wire and at least a subset of the neighboring wires in the second wire structure have curvilinear wire segments that are curved.
18 . The non-transitory machine-readable medium of claim 12 , wherein the wire segments of the first wire comprise at least one curvilinear segment that includes at least one curved portion.
19 . The non-transitory machine-readable medium of claim 18 , wherein the wire segments of the neighboring wires comprise curvilinear segments that include at least one curved portion.
20 . The non-transitory machine-readable medium of claim 19 , wherein the wire segments of the first wire and the neighboring wires further comprise rectilinear segments that include only straight segments.