Method of cell placement for an itegrated circuit chip comprising integrated placement and cell overlap removal

• US 5,619,419 A
• Filed: 09/13/1994
• Issued: 04/08/1997
• Est. Priority Date: 09/13/1994
• 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: Boyle, Douglas B., D'Haeseleer, Patrik
• Primary Examiner(s): Teska, Kevin J.
• Assistant Examiner(s): LOUIS JACQUES, JACQUES H
• Application Number:
  US08/306,182
• Time in Patent Office: 938 Days
• Field of Search: 364/488, 364/489, 364/490, 364/491, 364/578, 364/468
• US Class Current: 716/123
• CPC Class Codes: G06F 30/392 Floor-planning or layout, e.g. partitioning or placement

Abstract

A method of cell placement for an integrated circuit chip includes performing a chaotic improvement operation on an initial cell placement. At least some of the cells are relocated to new locations that provide lower interconnect wirelength and congestion. For each cell, the centroid of the net of cells to which the cell is connected is computed. The cell is then moved toward the centroid by a distance that is equal to the distance from the current position of the cell to the centroid multiplied by a "chaos" factor λ. The value of λ is selected such that the cell relocation operations will cause the placement to converge toward an optimal configuration without chaotic diversion, but with a sufficiently high chaotic element to prevent the optimization operation from becoming stuck at local fitness maxima. The chaotic placement operation can generate illegal placements in which two or more cells can occupy one location, some locations can contain no cells and/or two or more cells can partially overlap. Cell overlap is reduced and the spatial distribution of the placement improved by computing a density map representing cell densities in incremental blocks of the placement respectively, computing, from the density map, a warp map representing first repulsive forces exerted at centers of said blocks by cell densities in surrounding blocks respectively, interpolating, from the warp map, second repulsive forces acting on selected cells in the placement and spatially distributing the selected cells in accordance with the second repulsive forces. Overlapping cells are then moved away from each other by a separate cell-to-cell repulsive force based on cell overlap to reduce any remaining overlap. The chaotic placement and spatial distribution phases are repeated, with the scale of movement for chaotic placement being progressively decreased and the scale of movement for spatial distribution being progressively increased. A final phase is then performed in which only the cell-to-cell repulsive force is used to eliminate the remaining overlap.

Claims

1. A computer implemented method of spatially distributing cells in a cell placement for an integrated circuit chip, comprising the steps of:

(a) computing a density map representing cell densities in incremental blocks of said placement respectively;

(b) computing, from the density map, a warp map representing first repulsive forces exerted at centers of said blocks by cell densities in surrounding blocks respectively;

(c) interpolating, from the warp map, second repulsive forces acting on selected cells in the placement; and

(d) spatially distributing said selected cells in accordance with said second repulsive forces.

2. A method as in claim 1, in which step (a) comprises computing said cell densities as a predetermined function of areas of said incremental blocks that are overlapped by cells.

3. A method as in claim 2, in which step (a) comprises summing areas of said incremental blocks that are overlapped by cells.

4. A method as in claim 1, in which step (b) comprises computing said first repulsive forces as a predetermined function of distances from said centers to said surrounding blocks respectively.

5. A method as in claim 1, in which step (b) comprises computing said first repulsive forces as being inversely proportional to distances from said centers to said surrounding blocks respectively.

6. A method as in claim 1, in which step (b) comprises computing said first repulsive forces as being inversely proportional to squares of distances from said centers to said surrounding blocks respectively.

7. A method as in claim 1, in which step (c) comprises linearly interpolating said second repulsive forces.

8. A method as in claim 1, further comprising the steps, performed between steps (a) and (b), of:

(e) computing an average cell density from said cell densities; and

(f) subtracting said average cell density from said cell densities.

9. A method as in claim 1, in which step (c) comprises selecting said selected cells as being overlapping cells.

10. A method as in claim 1, further comprising the steps, performed after step (d), of:

(e) identifying overlapping cells; and

(f) moving at least one of said overlapping cells to reduce overlap thereof.

11. A method as in claim 10, in which step (f) comprises moving two overlapping cells in opposite directions.

12. A method as in claim 10, in which step (f) comprises moving two overlapping cells in opposite directions along a line joining centers of said overlapping cells.

13. A method as in claim 10, in which step (f) comprises moving two overlapping cells along lines joining centers of said two overlapping cells and a center of an overlapping area of said two overlapping cells.

14. A method as in claim 10, in which step (f) comprises moving two overlapping cells along lines joining centers of said two overlapping cells and a center of an overlapping area of said two overlapping cells respectively.

15. A method as in claim 10, in which step (f) comprises moving two overlapping cells in directions that require minimum movement to reduce said overlap thereof.

16. A method as in claim 10, in which step (f) comprises the substeps of:

(g) calculating an amount of overlap of two overlapping cells; and

(h) moving at least one of said two overlapping cells by a distance corresponding to said amount of overlap.

17. A computer implemented method of improving a placement of cells for an integrated circuit chip, comprising the steps of:

(a) selecting and moving a plurality of first cells in accordance with a first predetermined function to reduce a cost factor of said placement;

(b) selecting and moving a plurality of second cells in accordance with a second predetermined function to improve a spatial distribution of said placement;

(c) altering said first and second predetermined functions to decrease a scale factor for movement of said first cells and increase a scale factor for movement of second cells respectively; and

(d) repeating steps (a) to (c).

18. A method as in claim 17, further comprising the steps, performed prior to step (a), of:

(e) selecting and moving a plurality of first cells in accordance with said first predetermined function to reduce a cost factor of said placement;

(f) selecting and moving a plurality of second cells in accordance with said second predetermined function to improve a spatial distribution of said cells;

(g) altering only said second predetermined function to increase said scale factor for movement of said second cells; and

(h) repeatedly performing steps (e) to (g).

19. A method as in claim 17, in which said first predetermined function in step (a) comprises chaotic placement.

20. A method as in claim 17, in which said first predetermined function in step (a) comprises performing the substeps of:

(e) computing centroids for said first cells in accordance with locations of cells to which said first cells are connected respectively;

(f) computing first distances from current locations of said first cells to said centroids respectively;

(g) computing second distances in accordance with a third predetermined function of said first distances respectively; and

(h) moving said first cells from said current locations toward said centroids by said second distances respectively.

21. A method as in claim 17, in which said second predetermined function in step (b) comprises selecting said second cells as being overlapping cells.

22. A method as in claim 17, further comprising the steps, performed between steps (b) and (c), of:

(e) identifying overlapping cells; and

(f) moving at least one of said overlapping cells to reduce overlap thereof.

23. A method as in claim 22, in which step (f) comprises moving two overlapping cells in opposite directions.

24. A method as in claim 22, in which step (f) comprises moving two overlapping cells in opposite directions along a line joining centers of said overlapping cells.

25. A method as in claim 22, in which step (f) comprises moving two overlapping cells along lines joining centers of said two overlapping cells and a center of an overlapping area of said two overlapping cells.

26. A method as in claim 22, in which step (f) comprises moving two overlapping cells along lines joining centers of said two overlapping cells and a center of an overlapping area of said two overlapping cells respectively.

27. A method as in claim 22, in which step (f) comprises moving two overlapping cells in directions that require minimum movement to reduce said overlap.

28. A method as in claim 22, in which step (f) comprises the substeps of:

(g) calculating an amount of overlap of two overlapping cells; and

(h) moving at least one of said two overlapping cells by a distance according to said amount of overlap.

29. A method as in claim 22, in which:

step (f) comprises moving said at least one of said overlapping cells in accordance with a third predetermined function; and

step (c) further comprises altering said third predetermined function to increase a scale factor for movement of said at least one of said overlapping cells.

30. A method as in claim 29, in which steps (a), (b), (c) and (f) are performed simultaneously in combination.

31. A method as in claim 29, further comprising the steps, performed after step (d), of:

(g) identifying overlapping cells;

(h) moving at least one of said overlapping cells to reduce overlap thereof; and

(i) repeatedly performing steps (g) and (h) until all overlap has been removed.

32. A method as in claim 31, in which steps (g) to (i) are integral with steps (a) to (f), and comprise adjusting said first, second and third predetermined functions in accordance with a fourth predetermined function.

33. A method as in claim 17, in which said cost factor comprises interconnect wirelength.

34. A method as in claim 17, in which steps (a) and (b) are performed simultaneously in combination.

35. A method of improving a placement of cells for an integrated circuit chip, comprising the steps of:

(a) selecting and moving a plurality of first cells in accordance with a first predetermined function to reduce a cost factor of said placement;

(b) selecting and moving a plurality of second cells in accordance with a second predetermined function to improve a spacial distribution of said placement;

(c) altering said first and second predetermined functions to decrease a scale factor for movement of said first cells and increase a scale factor for movement of second cells respectively; and

(d) repeatedly performing steps (a) to (c);

in which said second predetermined function in step (b) comprises performing the substeps of;

(e) computing a density map representing cell densities in incremental blocks of said placement respectively;

(f) computing, from the density map, a warp map representing first repulsive forces exerted at centers of said blocks by cell densities in surrounding blocks in the placement respectively;

(g) interpolating, from the warp map, second repulsive forces acting on said second cells; and

(h) moving said second cells in accordance with said second repulsive forces.

36. A method as in claim 35, in which step (e) comprises computing said cell densities as a predetermined function of areas of said incremental blocks that are overlapped by cells.

37. A method as in claim 35, in which step (e) comprises summing areas of said incremental blocks that are overlapped by cells.

38. A method as in claim 35, in which step (f) comprises computing said first repulsive forces as a predetermined function of distances from said centers to said surrounding blocks respectively.

39. A method as in claim 35, in which step (f) comprises computing said first repulsive forces as being inversely proportional to distances from said centers to said surrounding blocks respectively.

40. A method as in claim 35, in which step (f) comprises computing said first repulsive forces as being inversely proportional to squares of distances from said centers to said surrounding blocks respectively.

41. A method as in claim 35, in which step (c) comprises linearly interpolating said second repulsive forces.

42. A method as in claim 35, further comprising the steps, performed between steps (e) and (f), of:

(i) computing an average cell density from said cell densities; and

(j) subtracting said average cell density from said cell densities.