Multiple-layer contour searching method and apparatus for circuit building block placement

• US 5,363,313 A
• Filed: 02/28/1992
• Issued: 11/08/1994
• Est. Priority Date: 02/28/1992
• Status: Expired due to Term

• Current Assignee: Cadence Design Systems Incorporated (Cadence Group)
• Original Assignee: Cadence Design Systems Incorporated (Cadence Group)
• Inventors: Lee, Tsu-Chang
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): Walker, Tyrone
• Application Number:
  US07/843,712
• Time in Patent Office: 984 Days
• Field of Search: 364/491, 364/490, 364/489, 364/488
• US Class Current: 716/55
• CPC Class Codes: 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

Abstract

Apparatus and methods for two-dimensional compaction of object collections defines an initial set of unrefined objects and an initial compaction direction. One by one, each unrefined object is taken from the unrefined object configuration and placed within a refined configuration. Both unrefined and refined configurations possess profiles along which the selected object is placed. Pruning rules select which locations to investigate more closely, by eliminating those positions that clearly do not provide compaction improvement. Various orientations and/or shapes of an object can be tried to improve the compaction or the group of objects. Once a best location for the object is found, the object is included in the refined object set, new unrefined and refined profiles are constructed, and a new unrefined object is selected to place. Once all unrefined objects have been selected and placed, the compaction process for the chosen direction has completed. The process can continue for other compaction directions, and as many times as required, to compact the collection of objects within a certain tolerance. The cost function employed to determine best object positions can include costs of wire interconnection lengths.

Claims

1. A method for compacting a set of objects, comprising the steps of:

(a) defining a set of N unrefined objects, each of said unrefined objects having at least one object profile, each of said objects having at least one layer;

(b) defining a set of refined objects;

(c) defining an object area, said object area having a vertical axis defining an upper and a lower end, said object area having a horizontal axis having a base width;

(d) placing said set of unrefined objects within and at said upper end of said object area, said placement defining a first unrefined configuration;

(e) removing a first removed object from said set of unrefined objects by selecting a lowest unrefined object along a longest critical path;

said first unrefined configuration no longer including said first removed object, said first removed object no longer included in said set of unrefined objects;

(f) defining a first unrefined profile for at least one of said layers of said first unrefined configuration;

(g) defining a first refined profile for a first refined configuration, said first refined profile located within and at a lower end of said object area;

(h) finding a lowest cost for placing said first removed object along said first refined profile by iteratively positioning the first removed object along the first refined profile at matching points selected according to a matching of a curvature between a refined side of the first removed object and the first refined profile and a matching of a curvature between an unrefined side of the first removed object and the first unrefined profile; and

(i) placing said first removed object at a point along said first refined profile associated with said lowest cost.

2. A method as recited in claim 1 further comprising the steps of:

(j) removing an ith removed object from an (i-1)th set of unrefined objects;

(k) defining an ith unrefined configuration, said ith unrefined configuration comprising an (i-1)th unrefined configuration no longer including an ith removed object, said ith removed object no longer included in said (i-1)th set of unrefined objects;

(l) defining an ith unrefined profile for at least one of said layers of said ith unrefined configuration;

(m) defining an (i-1)th refined profile for at least one of said layers of said (i-1)th refined configuration;

(n) finding a lowest cost for placing said ith removed object along said (i-1)th refined profile at selected matching points;

(o) placing said ith removed object at the matching point along said (i-1)th refined profile associated with said lowest cost;

(p) defining an ith refined configuration, said ith refined configuration comprising an (i-1)th refined configuration now also including said ith removed object, said ith removed object now included in said set of refined objects; and

(q) and repeating steps (j) through (p) for all of said N objects.

3. A method as recited in claim 2 wherein said finding of said lowest cost further comprises the steps of:

(a) defining a lower object location curve for said ith removed object, said lower object location curve comprising a set of possible matching points for said ith removed object along said (i-1)th refined profile;

(b) defining an upper object location curve for said ith removed object, said upper object location curve comprising a set of possible matching points for said ith removed object along said ith unrefined profile;

(c) defining a cost for each possible matching position of said ith removed object comprising a chip size cost as a function Of the combined height of said upper object location curve and said lower object location curve at said matching position; and

(d) finding an actual position for said ith removed object having the lowest of said costs.

4. A method as recited in claim 3 wherein said finding an actual position for said ith removed object comprises the steps of:

(a) defining a current best refined profile location for said ith object, said current best refined profile location having a height defined by said lower object location curve;

(b) defining a current best unrefined profile location for said ith object, said current best unrefined profile location having a height defined by said upper object location curve;

(c) defining a current best combined cost as a function of a chip size cost and a connection net length cost;

(d) defining a set of first refined matching points for placing said ith object along said ith refined profile such that a convex begin point of said ith object meets a concave begin point of said ith refined profile;

(e) discarding any of said first refined matching points if the height of placing said ith object at said first refined matching points as defined by said lower object location curve is greater than said current best combined cost minus said height of said current best unrefined profile location;

(f) defining a set of second refined matching points for placing said ith object along said ith refined profile such that a convex end point of said ith object meets a convex begin point of said ith refined profile;

(g) discarding any of said second refined matching points if the height of placing said ith object at said second refined matching point as defined by said lower object location curve is greater than said current best combined cost minus said height of said current best unrefined profile location;

(h) defining a set of first unrefined matching points for placing said ith object along said ith unrefined profile such that a convex begin point of said ith object meets a concave begin point of said ith unrefined profile;

(i) discarding any of said first unrefined matching points if the height of placing said ith object at said first unrefined matching point as defined by said upper object location curve is greater than said current best combined cost minus said height of said current best refined profile location;

(j) defining a set of second unrefined matching points for placing said ith object along said ith unrefined profile such that a convex end point of said ith object meets a convex begin point of said ith unrefined profile;

(k) discarding any of said second unrefined matching points if the height of placing said ith object at said second unrefined matching point as defined by said upper object location curve is greater than said current best combined cost minus said height of said current best refined profile location;

(l) calculating a current combined cost of placing said ith object at one of said undiscarded matching points, said point defining a current undiscarded matching point;

(m) if said calculated current combined cost is less than said current best combined cost, then defining said current best combined cost as said calculated current combined cost;

(n) if said current undiscarded unrefined profile location is better than said current best unrefined profile location, then defining said current best unrefined profile location as said current undiscarded unrefined matching point;

(o) if said current undiscarded refined profile location is better than said current best refined profile location, then defining said current best refined profile location as said current undiscarded refined matching point; and

(p) repeating steps (l) through (o) until all of said refined matching points defined in steps (d) and (f) and all of said unrefined matching points defined in steps (h) and (j) have been explored.

5. A method as recited in claim 3 wherein said finding an actual position for said ith removed object comprises the steps of:

(a) defining a current best refined profile location for said ith object, said current best refined profile location having a height defined by said lower object location curve;

(b) defining a current best unrefined profile location for said ith object, said current best unrefined profile location having a height defined by said upper object location curve;

(c) defining a current best combined cost as a function of a chip size cost and a connection net length cost;

(d) defining a set of first refined matching points for placing said ith object along said ith refined profile such that a convex begin point of said ith object meets a convex end point of said ith refined profile;

(e) discarding any of said first refined matching points if the height of placing said ith object at said first refined matching points as defined by said lower object location curve is greater than said current best combined cost minus said height of said current best unrefined profile location;

(f) defining a set of second refined matching points for placing said ith object along said ith refined profile such that a convex end point of said ith object meets a concave end point of said ith refined profile;

(g) discarding any of said second refined matching points if the height of placing said ith object at said second refined matching point as defined by said lower object location curve is greater than said current best combined cost minus said height of said current best unrefined profile location;

(h) defining a set of first unrefined matching points for placing said ith object along said ith unrefined profile such that a convex begin point of said ith object meets a convex end point of said ith unrefined profile;

(i) discarding any of said first unrefined matching points if the height of placing said ith object at said first unrefined matching point as defined by said upper object location curve is greater than said current best combined cost minus said height of said current best refined profile location;

(j) defining a set of second unrefined matching points for placing said ith object along said ith unrefined profile such that a convex end point of said ith object meets a concave end point of said ith unrefined profile;

(k) discarding any of said second unrefined matching points if the height of placing said ith object at said second unrefined matching point as defined by said upper object location curve is greater than said current best combined cost minus said height of said current best refined profile location;

(l) calculating a current combined cost of placing said ith object at one of said undiscarded matching points, said point defining a current undiscarded matching point;

(m) if said calculated current combined cost is less than said current best combined cost, then defining said current best combined cost as said calculated current combined cost

(n) if said current undiscarded unrefined profile location is better than said current best unrefined profile location, then defining said current best unrefined profile location as said current undiscarded unrefined matching point;

(o) if said current undiscarded refined profile location is better than said current best refined profile location, then defining said current best refined profile location as said current undiscarded refined matching point; and

(p) repeating steps (l) through (o) until all of said refined matching points defined in steps (d) and (f) and all of said unrefined matching points defined in steps (h) and (j) have been explored.

6. A method as recited in claim 3 wherein said finding an actual position for said ith removed object comprises the steps of:

(a) defining a current best refined profile location for said ith object, said current best refined profile location having a height defined by said lower object location curve;

(b) defining a current best unrefined profile location for said ith object, said current best unrefined profile location having a height defined by said upper object location curve;

(c) defining a current best combined cost as a function of a chip size cost and a connection net length cost;

(d) defining a set of first refined matching points for placing said ith object along said ith refined profile such that a convex begin point of said ith refined profile meets a concave begin point of said ith object;

(e) discarding any of said first refined matching points if the height of placing said ith object at said first refined matching points as defined by said lower object location curve is greater than said current best combined cost minus said height of said current best unrefined profile location;

(f) defining a set of second refined matching points for placing said ith object along said ith refined profile such that a convex end point of said ith refined profile meets a convex begin point of said ith object;

(g) discarding any of said second refined matching points if the height of placing said ith object at said second refined matching point as defined by said lower object location curve is greater than said current best combined cost minus said height of said current best unrefined profile location;

(h) defining a set of first unrefined matching points for placing said ith object along said ith unrefined profile such that a convex begin point of said ith unrefined profile meets a concave begin point of said ith object;

(i) discarding any of said first unrefined matching points if the height of placing said ith object at said first unrefined matching point as defined by said upper object location curve is greater than said current best combined cost minus said height of said current best refined profile location;

(j) defining a set of second unrefined matching points for placing said ith object along said ith unrefined profile such that a convex end point of said ith unrefined profile meets a convex begin point of said ith object;

(k) discarding any of said second unrefined matching points if the height of placing said ith object at said second unrefined matching point as defined by said upper object location curve is greater than said current best combined cost minus said height of said current best refined profile location;

(l) calculating a current combined cost of placing said ith object at one of said undiscarded matching points, said point defining a current undiscarded matching point;

(m) if said calculated current combined cost is less than said current best combined cost, then defining said current best combined cost as said calculated current combined cost

(n) if said current undiscarded unrefined profile location is better than said current best unrefined profile location, then defining said current best unrefined profile location as said current undiscarded unrefined matching point;

(o) if said current undiscarded refined profile location is better than said current best refined profile location, then defining said current best refined profile location as said current undiscarded refined matching point; and

(p) repeating steps (l) through (o) until all of said refined matching points defined in steps (d) and (f) and all of said unrefined matching points defined in steps (h) and (j) have been explored.

7. A method as recited in claim 3 wherein said finding an actual position for said ith removed object comprises the steps of:

(a) defining a current best refined profile location for said ith object, said current best refined profile location having a height defined by said lower object location curve;

(b) defining a current best unrefined profile location for said ith object, said current best unrefined profile location having a height defined by said upper object location curve;

(c) defining a current best combined cost as a function of a chip size cost and a connection net length cost;

(d) defining a set of first refined matching points for placing said ith object along said ith refined profile such that a convex begin point of said ith, refined profile meets a convex end point of said ith object;

(e) discarding any of said first refined matching points if the height of placing said ith object at said first refined matching points as defined by said lower object location curve is greater than said current best combined cost minus said height of said current best unrefined profile location;

(f) defining a set of second refined matching points for placing said ith object along said ith refined profile such that a convex end point of said ith refined profile meets a concave end point of said ith object;

(g) discarding any of said second refined matching points if the height of placing said ith object at said second refined matching point as defined by said lower object location curve is greater than said current best combined cost minus said height of said current best unrefined profile location;

(h) defining a set of first unrefined matching points for placing said ith object along said ith unrefined profile such that a convex begin point of said ith unrefined profile meets a convex end point of said ith object;

(i) discarding any of said first unrefined matching points if the height of placing said ith object at said first unrefined matching point as defined by said upper object location curve is greater than said current best combined cost minus said height of said current best refined profile location;

(h) defining a set of second unrefined matching points for placing said ith object along said ith unrefined profile such that a convex end point of said ith unrefined profile meets a concave end point of said ith object;

(k) discarding any of said second unrefined matching points if the height of placing said ith object at said second unrefined matching point as defined by said upper object location curve is greater than said current best combined cost minus said height of said current best refined profile location;

(l) calculating a current combined cost of placing said ith object at one of said undiscarded matching points, said point defining a current undiscarded matching point;

(m) if said calculated current combined cost is less than said current best combined cost, then defining said current best combined cost as said calculated current combined cost

(n) if said current undiscarded unrefined profile location is better than said current best unrefined profile location, then defining said current best unrefined profile location as said current undiscarded unrefined matching point;

(o) If said current undiscarded refined profile location is better than said current best refined profile location, then defining said current best refined profile location as said current undiscarded refined matching point;

(p) repeating steps (l) through (o) until all of said refined matching points defined in steps (d) and (f) and all of said unrefined matching points defined in steps (h) and (j) have been explored.

8. A method as recited in claim 2 wherein said finding of said lowest cost further comprises the steps of:

(a) defining a lower object location curve for said ith removed object for each of said layers, each of said lower object location curves comprising a set of possible matching points for said ith removed object along said ith refined profile of said layer;

(b) defining an upper object location curve for said ith removed object for each of said layers, each of said upper object location curves comprising a set of possible matching points for said ith removed object along said unrefined profile of said layer;

(c) defining a cost for each possible position of said ith removed object as the sum of the maximum of said upper object location curves and the maximum of said lower object location curves for each layer at said position; and

(d) finding an actual position for said ith removed object having the lowest of said costs.

9. A method as recited in claim 1 wherein said objects comprise contiguous masked regions on a semiconductor mask.

10. A method as recited in claim 1 wherein said set of refined objects is initially empty.

11. A method as recited in claim 1 wherein said first unrefined profile comprises a line connecting the lowermost points of said first unrefined configuration of said set of said unrefined objects.

12. A method as recited in claim 1 wherein said first refined profile comprises a line along said lower end of said object area.

13. A method as recited in claim 2 wherein said ith unrefined profile comprises a line connecting the lowermost points of said ith unrefined configuration of said set of said unrefined objects.

14. A method as recited in claim 2 wherein said ith refined profile comprises a line connecting the uppermost points of said ith refined configuration of said set of said refined objects.

15. A method as recited in claim 1 wherein said objects are connected by a plurality of connection nets, said cost further comprising a new cost of a connection net length among said objects as a function of a previous connection net length and a change in a chip size along the vertical axis.

16. A method as recited in claim 15 wherein said new connection net length cost is calculated by a method comprising:

(a) defining a previous connection net length cost;

(b) defining a previous object net length, said previous object net length comprising a combined net length for all connection nets attached to said ith removed object at a previous ith object location;

(c) defining a new object net length, said new object net length comprising a combined net length for all connection nets attached to said ith removed object at a current ith object location;

(d) defining a jth net change in height, said jth net change comprising said jth upper object location curve added to said jth lower object location curve at a current jth removed object location minus the summation of said jth upper object location curve added to said jth lower object location curve at an original jth removed object location;

(e) defining a total ith net change in height as a summation of said jth net changes in height from j=l to j=i;

(f) dividing said plurality of connection nets into three types of connection nets, a first type comprising covariant connection nets, a second type comprising contravariant connection nets, a third type comprising invariant connection nets;

(g) dividing said plurality of connection nets into a set of object nets, said object nets including said connection nets attached to said ith removed object, and a set of non-object nets, said non-object nets including said connection nets not attached to said ith removed object;

(h) determining a margin M of said net change in height, said margin M dividing said plurality of connection nets into strict and non-strict nets, said strict nets remaining within their type as long as said total ith net change in height is less than margin M;

(i) if said total ith net change in height is less than margin M, then calculating said new connection net length cost as said previous connection net length minus said previous object net length plus said current object net length plus said total ith net change in height multiplied by the total of (the number of said non-object strict covariant connection nets minus the number of said non-object strict contravariant connection nets) plus a calculated connection length for said non-object non-strict connection nets; and

(j) if said total ith net change in height is greater than margin M, then determining which nets have changed type and recategorizing said changed type nets accordingly, calculating said new connection net length cost as said previous connection net length minus said previous object net length plus said current object net length plus said total ith net change in height multiplied by the total of (the number of said non-object strict covariant connection nets minus the number of said non-object strict contravariant connection nets) plus a calculated connection length for said non-object non-strict connection nets.

17. The method of claim 1 wherein an aspect ratio of the object area is altered by changing the base width before compaction.

18. The method of claim 2, further comprising the steps of:

reserving an area for a preplaced object; and

comparing the ith removed object with the reserved area to prevent the ith removed object from overlapping the reserved area.

19. The method of claim 1 wherein each object profile has an aspect ratio, and the lowest cost for placing said first removed object is the lowest cost of placing said first removed object using each object profile.

20. A system for compacting of a set of objects, comprising:

a logic processor;

an object register connected to said logic processor for storing data related to said set of objects, each of said objects having a position attribute for determining its position within a defined two-dimensional area;

an unrefined object register connected to said logic processor for storing a listing of a set of unrefined objects of said set of objects;

an unrefined profile register connected to said logic processor for storing an unrefined profile of said set of unrefined objects;

a refined object register connected to said logic processor for storing a listing of a set of refined objects of said set of objects;

a refined profile register connected to said logic processor for storing a refined profile of said set of refined objects; and

a cost optimizer operated on by said logic processor for selecting one of said unrefined objects stored in said unrefined object register, said cost optimizer determining a position attribute for said selected object by matching a curvature of said selected object with a curvature of said unrefined profile and by matching a curvature of said selected object with a curvature of said refined profile.

21. A system as recited in claim 20 further comprising:

a net register connected to said logic processor for storing connection net information comprising a net type and an object net length for each of said objects, said cost optimizer determining a position attribute for said selected object by using said unrefined profile, said refined profile and said connection net information.