EE382V Fall 2006

VLSI Physical Design Automation

Lecture 8. Floorplanning (2)

Prof. David Pan

[email protected]
Office: ACES 5.434

10/22/08 1
 Linear Programming Approach
“An Analytical Approach to Floorplan Design and
Optimization”, Sutanthavibul, Shragowitz and Rosen,
IEEE Transaction on CAD, 10:761-769, June 1991.

2
 Mixed Integer Linear Program
• A mathematical program such that:
– The objective is a linear function.
– All constraints are linear functions.
– Some variables are real numbers and some are integers, i.e.,
“mixed integer”.
• It is almost like a linear program, except that some
variables are integers.
• Can you think of which variables may be integer?

3
 Problem Formulation
• Minimize the packing area:
– Assume that one dimension W is fixed.
– Minimize the other dimension Y.
• Need to have constraints W
so that blocks do not overlap.
• Associate each block Bi with 4
Y
variables:
– xi and yi: coordinates of its lower left
corner.
– wi and hi: width and height.

4
 Non-overlapping Constraints

For two non-overlapping blocks Bi and Bj, at least one

of the following four linear constraints must be
satisfied:

(1) x i + wi ≤ x j if B i is to the left of B j

or (2) xi − w j ≥ x j if B i is to the right of B j
or (3) y i + hi ≤ y j if B i is below B j
or (4) yi − h j ≥ y j if B i is above B j

hi Bi hj Bj
(xi, yi) wi (xj, yj) wj
5
 Integer Variables
• Use integer (0 or 1) variables xij and yij:
xij=0 and yij =0 if (1) is true.
xij=0 and yij =1 if (2) is true.
xij=1 and yij =0 if (3) is true.
xij=1 and yij =1 if (4) is true.
• Let W and H be upper bounds on the total width and
height. Non-overlapping constraints:
(1' ) xi + wi ≤ x j + W ( xij + yij )
(2' ) xi − w j ≥ x j − W (1 + xij − yij )
(3' ) yi + hi ≤ y j + H (1 − xij + yij )
(4' ) yi − h j ≥ y j − H (2 − xij − yij )
6
 Formulation

Min. Y
s.t. 0 ≤ xi , xi + wi ≤ W 1≤ i ≤ n
0 ≤ yi , yi + hi ≤ Y 1≤ i ≤ n
xi + wi ≤ x j + W ( xij + yij ) 1≤ i < j ≤ n
xi − w j ≥ x j − W (1 + xij − yij ) 1≤ i < j ≤ n
yi + hi ≤ y j + H (1 − xij + yij ) 1≤ i < j ≤ n
yi − h j ≥ y j − H (2 − xij − yij ) 1≤ i < j ≤ n
xij = 0 or 1 1≤ i < j ≤ n
yij = 0 or 1 1≤ i < j ≤ n

7
 Formulation with Hard Blocks
If the blocks can be rotated, use a 0-1 integer variable zi for each
block Bi s.t. zi = 0 if Bi is in the original orientation and zi = 1 if Bi is
rotated 90o .
Min. Y
s.t. 0 ≤ x i , x i + z i hi + (1 − z i ) wi ≤ W 1≤ i ≤ n
0 ≤ y i , y i + z i wi + (1 − z i )hi ≤ Y 1≤ i ≤ n
x i + z i hi + (1 − z i ) wi ≤ x j + W ( x ij + y ij ) 1≤ i < j ≤ n
x i − z j h j − (1 − z j ) w j ≥ x j − W (1 + x ij − y ij ) 1≤ i < j ≤ n
y i + z i wi + (1 − z i )hi ≤ y j + H (1 − x ij + y ij ) 1≤ i < j ≤ n
y i − z j w j − (1 − z j )h j ≥ y j − H (2 − x ij − y ij ) 1 ≤ i < j ≤ n
x ij = 0 or 1 1≤ i < j ≤ n
y ij = 0 or 1 1≤ i < j ≤ n
8
 Formulation with Soft Blocks

• If Bi is a soft block, wihi=Ai. But this constraint is

quadratic!
• Linearized by taking the first two terms of the Taylor
expression of hi=Ai/wi at wimax (max. width of block Bi).
hi =himin+λi(wimax-wi)
where himin =Ai/wimax and λi=Ai/wimax2

9
 Formulation with Soft Blocks
• If Bi is soft and Bj is hard:

(1) xi + wi ≤ x j + W ( xij + yij )

(2) xi − w j ≥ x j − W (1 + xij − yij )
(3) yi + himin + λi ( wimax − wi ) ≤ y j + H (1 − xij + yij )
(4) yi − h j ≥ y j − H (2 − xij − yij )

• If both Bi and Bj are soft:

(1) xi + wi ≤ x j + W ( xij + yij )
(2) xi − w j ≥ x j − W (1 + xij − yij )
(3) yi + himin + λi ( wimax − wi ) ≤ y j + H (1 − xij + yij )
(4) yi − h jmin − λ j ( w jmax − w j ) ≥ y j − H (2 − xij − yij )
10
Another way to linearize

11
 Solving Linear Program

• Linear Programming (LP) can be solved by classical

optimization techniques in polynomial time.
• Mixed Integer LP (MILP) is NP-Complete.
– The run time of the best known algorithm is exponential to the
number of variables and equations

12
 Complexity

• For a problem with n blocks, and for the simplest case,

i.e., all blocks are hard:
– 2n continuous variables (xi, yi)
– 2n(n-1) +n integer variables (xij, yij, zi)
– 4n2-2n linear constraints
• Practically, this method can only solve small size
problems.

13
 Successive Augmentation
A classical greedy approach to keep the problem size
small: repeatedly pick a small subset of blocks to
formulate a MILP, solve it together with the previously
picked blocks with fixed locations and shapes:

Next subset

Y Partial Solution

14
 Floorplan Representations
• Slicing
– Normalized Polish Expression: Wong & Liu [DAC-86]
• Mosaic (and General)
– Corner Block List (CBL): Hong et al. [ICCAD-00]
– Q-Sequence: Sakanushi & Kajitani [APCCAS-00]
– Twin Binary Sequence (TBS): Young, Chu, Shen [ISPD-02]
• General
– Polar graphs: Ohtsuki et al. [ICCST-70]
– Sequence pair: Murata et al. [ ICCAD-95]
– Bounded Slicing Grid (BSG): Nakatake [ICCAD-96]
– Transitive Closure Graph (TCG): Lin & Chang [DAC-01]
• Compacted
– O-tree: Guo et al. [DAC-99]
– B*-tree: Chang et al. [DAC-00]

15
 General Floorplanning
by Simulated Annealing:
Sequence-Pair Representation

Murata, Fujiyoshi, Nakatake, Kajitani,

“VLSI Module Placement based on
Rectangle-Packing by the Sequence-Pair”,
TCAD, 15:1518-1524, December 1996.

10/22/08 16
 Sequence-Pair vs. Polish Expression
• Sequence-Pair is a succinct representation of non-
slicing floorplans of rectangles
– Just like Polish Expression for slicing floorplans
• Represent a non-slicing floorplan by a pair of sequences
of blocks.
• Using Simulated Annealing to find a good sequence-pair
• Can only handle hard blocks
– i.e., cannot do things like shape-curve computation
• Essentially macro placement
• Techniques for soft block shaping exist (e.g.,
Lagrangian Relaxation) but are very slow

17
 Sequence Pair

• Positive step line • Negative step line

sequence: ecadfb sequence: fcbead

18
Positive Locus and Negative Locus
Positive Locus
of Block b

Negative Locus
of Block b

19
 Sequence-Pair

Positive Loci Negative Loci

Sequence-Pair = (abdecf, cbfade)

20
 Geometric Info of Sequence-Pair
Given a placement and the corresponding
sequence-pair (P, N):
• a is right to b iff a is after b in both P and N.
• a is left to b iff a is before b in both P and N.
• a is above b iff a is before b in P and after b in N.
• a is below b iff a is after b in P and before b in N.

21
 From Sequence-Pair to a Placement
Labeled grid for
(abdecf, cbfade)
• Given a sequence-pair,
e the placement with
d smallest area can be
a found in O(n2) time.
f
b • Algorithms of time O(n
c log log n) or O(n log n)
exist. But faster than
a O(n2) algorithm only
b when n is quite large.
d
e
c
f
22
 From Sequence-Pair to Placement
• Distance from left (bottom) edge can be found using
the longest path algorithm on the horizontal (vertical)
constraint graph.

Horizontal Constraint Graph Vertical Constraint Graph

23
 Sequence Pair (SP)
A floorplan is represented by a pair of permutations of
the module names:
e.g. 13245
35412
A sequence pair (s1, s2) of n modules can represent all
possible floorplans formed by the n modules by
specifying the pair-wise relationship between the
modules.

24
 Sequence Pair

Consider a pair of modules A and B. If the

arrangement of A and B in s1 and s2 are:
– (…A…B…, …A…B…), then the right boundary of A is on the
left hand side of the left boundary of B.
– (…A…B…, …B…A…), then the upper boundary of B is
below the lower boundary of A.

25
 Example

Consider the sequence pair:

(13245,41352 )

2
1 3

5
4

Any other SP that is also valid for this packing?

26
 Floorplan Realization
• Floorplan realization is the step to construct a floorplan
from its representation.
• How to construct a floorplan from a sequence pair?
• We can make use of the horizontal and vertical
constraint graphs (Gh and Gv).

27
 Floorplan Realization

• Whenever we see (…A…B…, …A…B…), add an edge from A to

B in Gh with weight wA.
• Whenever we see (…A…B…, …B…A…), add an edge from B to
A in Gv with weight hA.
• Add a source vertex s to Gh and Gv pointing, with weight 0, to all
vertices without incoming edges.
• Finally, find the longest paths from s to every vertex in Gh and Gv
(how?), which are the coordinates of the lower left corner of the
module in the packing.

28
 Example

Gh 3 1.1 2
1.2
1.1 1 1.2 1.1
1.2 1
0 1.2
2 1.2 5
2 1 3 s 0
4 2.4

5 2 Gv 3 2
1 4 1 2
2.4 1.2 1 1
1
5
(13245,41352 ) 4 0 0
s
29
 Constraint Graphs

• How many edges are there in Gh and Gv in total?

• Is there any transitive edges in Gh and Gv?
• How to remove the transitive edges?
• Can we reduce the size of Gh and Gv to linear, i.e., no.
of edges is of order O(n), by removing all the transitive
edges?

30
 Moves

• Three kinds of moves in the annealing

process:
M1: Rotate a module, or change the shape of a
module
M2: Interchange 2 modules in both sequences
M3: Interchange 2 modules in the first sequence
• Does this set of move operations ensure
reachability? Why?

31
 Pros and Cons of SP

• Advantages:
– Simple representation
– All floorplans can be represented.
– The solution space is finite. (How big?)
• Disadvantages:
– Redundant representation. The representation is not 1-to-1.
– The size of the constraint graphs, and thus the runtime to
construct the floorplan is quadratic

32
 Questions
• Can we improve the runtime to realize a floorplan from
its SP representation? (“FAST-SP: A Fast Algorithm for
Block Placement on Sequence Pair”, X. Tang and D.F. Wong,
ASP-DAC 2001, pp. 521-526.)

33
 Summary of Floorplanning
• Stockmeyer
– Slicing with given set of modules
– Dynamic programming (only keep irredundant solutions)
• Wong-Liu
– Slicing floorplan
– Nice polish expression
• Sequence pair
– Nonslicing
– Nice compact representation
• Mixed Integer Linear Programming
– Nonslicing
– Not scalable

34