Routing non-preferred direction wiring layers of an integrated circuit by minimizing vias between these layers
Some embodiments of the invention provide an integrated circuit (IC) that has a novel non-preferred direction (NPD) wiring architecture. In some embodiments, the IC includes a substrate and multiple wiring layers, which include a first set of one or more wiring layers with no preferred wiring directions, and a second set of one or more wiring layers with preferred wiring directions. In some embodiments, the first set of wiring layers includes the third and fourth wiring layers, while the second set of wiring layers includes the fifth and higher metal layers with successive neighboring layers having different (e.g., alternating) preferred wiring directions. The first set of wiring layers in other embodiments includes the third wiring layer but not the fourth wiring layer, which in these embodiments has a preferred wiring direction.
1 . A method of performing routing for an integrated circuit (IC) design comprising (i) a plurality of non-preferred direction (NPD) routing layers including first and second NPD routing layers and (ii) a plurality of preferred direction (PD) routing layers, the method comprising
defining a first via cost for biasing path searches for the routing against defining a via between the first and second NPD routing layers and a second via cost for biasing the path searches against defining a via between an NPD routing layer and a PD routing layer, wherein the second via cost is more expensive than the first via cost in order to bias the path searches against using the PD routing layers;
using the first and second via costs to perform path searches to identify a plurality of paths to connect a plurality of nets; and
embedding the identified paths as routes that connect the plurality of nets.
2 . The method of claim 1 , wherein by biasing the path searches against defining vias between the first and second NPD routing layers, the first via cost biases the path searches to identify paths for the nets that reside completely or mostly on just one NPD routing layer.
3 . The method of claim 2 further comprising presenting the first via cost as a configurable parameter to a designer in order to allow the designer to modify the first via cost to increase or decrease the biasing towards using just one NPD routing layer for all of an identified path.
4 . The method of claim 1 , wherein the defining, using and embedding are part of a first routing operation that is performed for a first set of nets, the method further comprising performing a second routing operation to identify routes that traverse the PD routing layers to connect nodes associated with a second set of nets that are not part of the first set of nets.
5 . The method of claim 4 further comprising:
identifying a sorted order for a group of nets according to estimated lengths of routes for connecting the nets; and
selecting, based on the sorted order, a subgroup of nets with estimated lengths shorter than a particular value as a plurality of nets to try to route in the first routing operation.
6 . The method of claim 4 , wherein the second set of nets comprises selected nets that the first routing operation is not able to route.
7 . The method of claim 4 , wherein at least a subset of routes identified by the first routing operations are curvilinear routes with at least one curvilinear segment.
8 . The method of claim 4 , wherein at least some of the NPD routes identified by the first routing operation have route segments that traverse in two or more of eight preferred directions of routing.
9 . The method of claim 4 , wherein the first and second routing operations are detailed routing operations.
10 . The method of claim 4 , wherein the first and second routing operations are topological routing operations.
11 . The method of claim 10 further comprising performing a geometric routing operation after the first and second topological routing operations to identify geometric routes for the topological routes produced by the first and second topological routing operations.
12 . The method of claim 1 , wherein the first and second NPD routing layers are third and fourth routing layers of the IC design and the second plurality of routing layers comprises a plurality of routing layers above the fourth routing layer.
13 . The method of claim 12 further comprising defining a third via cost for routes traversing to the third routing layer from a first routing layer or a second routing layer of the IC design; and using the third via cost in performing the path searches.
14 . The method of claim 13 , wherein the first and second routing layers comprise a plurality of regions that have preferred direction rectilinear routing for intellectual property (IP) circuit blocks of the IC design, said path searches using space between said regions to identify NPD routes on the first and second NPD routing layers.
15 . A non-transitory machine readable medium storing a program for performing routing for an integrated circuit (IC) design comprising (i) a plurality of non-preferred direction (NPD) routing layers including first and second NPD routing layers and (ii) a plurality of preferred direction (PD) routing layers, the program comprising sets of instructions for:
defining a first via cost for biasing path searches for the routing against defining a via between the first and second NPD routing layers and a second via cost for biasing the path searches against defining a via between an NPD routing layer and a PD routing layer, wherein the second via cost is more expensive than the first via cost in order to bias the path searches against using the PD routing layers;
using the first and second via costs to perform path searches to identify a plurality of paths to connect a plurality of nets; and
embedding the identified paths as routes that connect the plurality of nets.
16 . The non-transitory machine readable medium of claim 15 , wherein by biasing the path searches against defining vias between the first and second NPD routing layers, the first via cost biases the path searches to identify paths for the nets that reside completely or mostly on just one NPD routing layer.
17 . The non-transitory machine readable medium of claim 16 , wherein the program further comprises a set of instructions for presenting the first via cost as a configurable parameter to a designer in order to allow the designer to modify the first via cost to increase or decrease the biasing towards using just one NPD routing layer for all of an identified path.
18 . The non-transitory machine readable medium of claim 15 , wherein the sets of instructions for defining, using and embedding are part of a first routing operation that is performed for a first set of nets, the program further comprising a set of instructions for performing a second routing operation to identify routes that traverse the PD routing layers to connect nodes associated with a second set of nets that are not part of the first set of nets.
19 . The non-transitory machine readable medium of claim 18 , wherein the program further comprises sets of instructions for:
identifying a sorted order for a group of nets according to estimated lengths of routes for connecting the nets; and
selecting, based on the sorted order, a subgroup of nets with estimated lengths shorter than a particular value as a plurality of nets to try to route in the first routing operation.
20 . The non-transitory machine readable medium of claim 18 , wherein the second set of nets comprises selected nets that the first routing operation is not able to route.