Cell placement method for microelectronic integrated circuit combining clustering, cluster placement and de-clustering

• US 5,682,321 A
• Filed: 10/05/1994
• Issued: 10/28/1997
• Est. Priority Date: 10/05/1994
• Status: Expired due to Term

• Current Assignee: Bell Semiconductor, LLC (Hilco Inc. (d/b/a Hilco Global))
• Original Assignee: LSI Logic Corporation (Broadcom, Inc.)
• Inventors: Wang, Ting-Chi, Irwin, Mary Jane, Ding, Cheng-Liang
• Primary Examiner(s): Teska, Kevin J.
• Assistant Examiner(s): NGUYEN, TAN QUANG
• Application Number:
  US08/318,275
• Time in Patent Office: 1,119 Days
• Field of Search: 395/500, 395/921, 364/488, 364/489, 364/490, 364/491, 257/210, 257/211, 257/203, 257/208, 326/41, 326/38, 326/101, 437/51
• US Class Current: 716/122
• CPC Class Codes: G06F 30/392 Floor-planning or layout, e.g. partitioning or placement

Abstract

A large number of microelectronic circuit cells that are interconnected by a set of wiring nets are optimally placed on an integrated circuit chip such that all interconnects can be routed and the total wirelength of the interconnects is minimized. Cells are first grouped into disjoint clusters by an optimization-driven clustering technique, which uses both local and global connectivity information among the cells. This technique uses Rent'"'"'s rule for combining pairs of neighboring clusters, and selects among pairs of clusters having the same Rent'"'"'s exponent using distance information derived from global optimization processing. Clusters are prevented from growing to an excessive size by limiting the number of cells per cluster and the maximum area per cluster to predetermined maximum values. After the clusters are generated, they are placed using an optimization-driven placement technique, preferably "Gordian". Finally, the cells within each cluster are de-clustered and locally placed using a partitioning technique, preferably "min-cut".

Claims

1. A method of cell placement for a microelectronic circuit including a plurality of cells and interconnections between said cells, comprising the steps of:

(a) grouping said cells into clusters using an optimization-driven technique including a predetermined local function of said interconnections and a predetermined global function of said interconnections;

(b) placing said clusters using a global optimization and partitioning technique; and

(c) placing said cells within said clusters using a partitioning technique.

2. A method according to claim 1, in which said global optimization and partitioning technique in step (b) comprises the substep of:

(d) globally placing said clusters using linearly constrained quadratic programming; and

said partitioning technique in step (c) comprises the substep of;

(e) locally placing said clusters using min-cut partitioning.

3. A method according to claim 1, in which the global optimization and partitioning technique in step (b) comprises the "Gordian" technique.

4. A method according to claim 1, in which the partitioning technique of step (c) comprises min-cut partitioning.

5. A method as in claim 1, in which said predetermined global function in step (a) comprises linearly constrained quadratic programming.

6. A method according to claim 5, in which said predetermined local function in step (a) comprises merging pairs of clusters in accordance with Rent'"'"'s Rule.

7. A method according to claim 6, in which step (a) comprises the substeps of:

(d) selecting pairs of clusters for merging in ascending order of Rent'"'"'s exponent; and

(e) when more than one pair of clusters has substantially the same Rent'"'"'s exponent, selecting the pair having the smallest number of cells.

8. A method according to claim 7, in which step (a) further comprises the substep, following step (e), of:

(f) when said more than one pair of clusters has substantially said same Rent'"'"'s exponent and the same number of cells, selecting the pair having the smallest distance between their clusters.

9. A method according to claim 6, in which step (a) further comprises, when more than one pair of clusters has substantially the same Rent'"'"'s exponent, selecting the pair having the smallest distance between their clusters.

10. A method according to claim 6, in which step (a) further comprises merging only pairs of clusters having Rent'"'"'s exponents that are smaller than a predetermined value.

11. A method as in claim 6, in which step (a) comprises merging only pairs of clusters that have combined numbers of cells that are smaller than a predetermined value.

12. A method as in claim 6, in which step (a) comprises merging only pairs of clusters having combined areas that are smaller than a predetermined value.

13. A method determining an optimal placement for a plurality of objects having predetermined interconnections therebetween, comprising the steps of:

(a) grouping said objects into clusters using an optimization-driven technique including a predetermined local function of said interconnections and a predetermined global function of said interconnections;

(b) placing said clusters using a global optimization and partitioning technique; and

(c) placing said objects within said clusters using a partitioning technique.

14. A method according to claim 13, in which said global optimization and partitioning technique in step (b) comprises the substep of:

(d) globally placing said clusters using linearly constrained quadratic programming; and

said partitioning technique in step (c) comprises the substep of;

(e) locally placing said clusters using rain-cut partitioning.

15. A method according to claim 13, in which the global optimization and partitioning technique in step (b) comprises the "Gordian" technique.

16. A method according to claim 13, in which the partitioning technique of step (c) comprises min-cut partitioning.

17. A method as in claim 13, in which said predetermined global function in step (a) comprises linearly constrained quadratic programming.

18. A method according to claim 17, in which said predetermined local function in step (a) comprises merging pairs of clusters in accordance with Rent'"'"'s Rule.

19. A method according to claim 18, in which step (a) comprises the substeps of:

(d) selecting pairs of clusters for merging in ascending order of Rent'"'"'s exponent; and

(e) when more than one pair of clusters has substantially the same Rent'"'"'s exponent, selecting the pair including the smallest number of objects.

20. A method according to claim 19, in which step (a) further comprises the substep, following step (e), of:

(f) when said more than one pair of clusters has substantially said same Rent'"'"'s exponent and the same number of objects, selecting the pair having the smallest distance between their clusters.

21. A method according to claim 18, in which step (a) further comprises, when more than one pair of clusters has substantially the same Rent'"'"'s exponent, selecting the pair having the smallest distance between their clusters.

22. A method according to claim 18, in which step (a) comprises merging only pairs of clusters having Rent'"'"'s exponents that are smaller than a predetermined value.

23. A method as in claim 18, in which step (a) comprises merging only pairs of clusters that have combined numbers of objects that are smaller than a predetermined value.

24. A method as in claim 18, in which step (a) comprises merging only pairs of clusters having combined areas that are smaller than a predetermined value.