www.jntuworld.

com
www.jntuworldupdates.org

D
L
R
O

Floorplanning
Inputs to the floorplanning problem:

m
o
c

A set of blocks, fixed or flexible.

.
ld

Pin locations of fixed blocks.

r
o
w

A netlist.

w
w
6

2
3

An optimal floorplan,
in terms of area

N
J

u
t
jn

.
w

U
T

Objectives: Minimize area, reduce wirelength for (critical) nets, maximize routability, determine shapes of flexible blocks

A nonoptimal floorplan
1

www.specworld.in

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Floorplan Design

w
w

Aspect ratio: r <= y/x <= s

Rotation:
Module connectivity
a

N
J

.
w
b

.
ld

r
o
w

u
t
jn

m
o
c
y

Area: A=xy

Modules:

U
T

3 6
d
5

f
2

www.specworld.in

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

.
ld

u
t
jn

U
T

Floorplanning: Terminology
Rectangular dissection: Subdivision of a given rectangle by a finite #
of horizontal and vertical line segments into a finite # of non-overlapping
rectangles.

m
o
c

Slicing structure: a rectangular dissection that can be obtained by

repetitively subdividing rectangles horizontally or vertically.

r
o
w

Slicing tree: A binary tree, where each internal node represents a vertical
cut line or horizontal cut line, and each leaf a basic rectangle.
Skewed slicing tree: One in which no node and its right child are the
same.

1
2

4
6

w
w

.
w

4
2

1 H

6 7 4 5

Nonslicing floorplan

N
J

1 V
6

H
7 V

4 5

A slicing tree (skewed) Another slicing tree

(nonskewed)

Slicing floorplan

3
www.specworld.in

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

Related work

D
L
R
O

Floorplan Design by Simulated Annealing

m
o
c

Wong & Liu, A new algorithm for floorplan design, DAC86.

Consider slicing floorplans.

r
o
w

.
ld

u
t
jn

U
T

Wong & Liu, Floorplan design for rectangular and L-shaped modules, ICCAD87.
Also consider L-shaped modules.

.
w

w
w

N
J

Wong, Leong, Liu, Simulated Annealing for VLSI Design, pp. 3171,
Kluwer academic Publishers, 1988.
Ingredients: solution space, neighborhood structure, cost function, annealing schedule?

4
www.specworld.in

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Solution Representation
An expression E = e1 e2 . . . e2n1 , where ei {1, 2, . . . , n, H, V }, 1 i
2n 1, is a Polish expression of length 2n 1 iff

m
o
c

1. every operand j, 1 j n, appears exactly once in E;

2. (the balloting property) for every subexpression Ei = e1 . . . ei, 1
i 2n 1, #operands > #operators.

.
ld

r
o
w

1 6 H 3 5 V 2 H V 7 4 H V
# of operands = 4 ....... = 7
# of operators = 2 ....... = 5

U
T

u
t
jn

Polish expression Postorder traversal.

ijH: rectangle i on bottom of j; ijV : rectangle i on the left of j.

w
w

V
H

6
1

N
J

.
w

2 7 5

6
E = 16H2V75VH34HV

E = 16+2*75*+34+*
Postorder traversal of a tree!
5
www.specworld.in

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Solution Representation (contd)

V

w
tu

Nonskewed
cases

w
w

....... HH ........

E = 123HV4V

U
T

nonskewed!

n
j
.
w

or

E = 123H4VV

.
ld

4
3

m
o
c

skewed!
V

N
J

3
1

....... VV ........

Question: How to eliminate ambiguous representation?

6
www.specworld.in

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Normalized Polish Expression

A Polish expression E = e1 e2 . . . e2n1 is called normalized iff E has no
consecutive operators of the same type (H or V ).

m
o
c

.
ld

Given a normalized Polish expression, we can construct a unique rectangular slicing structure.

w
w

H
3

N
J

u
t
jn

.
w3

6
1

r
o
w

U
T

E = 16H2V75VH34HV
A normalized Polish expression

7
www.specworld.in

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Neighborhood Structure
Chain: HV HV H . . . or V HV HV . . .

m
o
c

1 6 H 3 5 V 2 H V 7 4 H V

.
ld

r
o
w

chain

u
t
jn

.
w

3 types of moves:

U
T

Adjacent: 1 and 6 are adjacent operands; 2 and 7 are adjacent operands;

5 and V are adjacent operand and operator.

w
w

N
J

M 1 (Operand Swap): Swap two adjacent operands.

M 2 (Chain Invert): Complement some chain (V = H, H = V ).
M 3 (Operator/Operand Swap): Swap two adjacent operand and
operator.

8
www.specworld.in

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

4
3

D
L
R
O

Effects of Perturbation
3

.
ld

12V4H3V

M1

12V3H4V

M2

or

m
o
c
2

w
tu

M3

12H3H4V

4
3

12H34HV

U
T

Question: The balloting property holds during the moves?

n
j
.
w

M 1 and M 2 moves are OK.

Check the M 3 moves! Reject illegal M 3 moves.

N
J

w
w

Check M 3 moves: Assume that the M3 move swaps the operand ei

with the operator ei+1 , 1 i k 1. Then, the swap will not violate the
balloting property iff 2Ni+1 < i.
Nk : # of operators in the Polish expression E = e1 e2 . . . ek , 1 k 2n 1.

9
www.specworld.in

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Cost Function
= A + W .

A: area of the smallest rectangle

W : overall wiring length

.
ld

: user-specified parameter

W =

M1

.
w

w
w

ij cij dij .

M2

cij : # of connections between blocks i and j.

M3

2
1

4
3

A: 12H34HV

N
J

u
t
jn
4

r
o
w

U
T

m
o
c

dij : center-to-center distance between basic rectangles i and j.

10
www.specworld.in

10

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Cost Evaluation: Shape Curves

Shape curves correspond to different kinds of constraints where the
shaded areas are feasible regions.

hi

hi

hi

wi

y = six
Bounding
area

w
w

wi

u
t
n

y = six

j
.
w
y

r
o
w

y = ri x

corner
points

U
T

hi

.
ld

y = six

y = ri x

hi

wi

wi hi

y = six y = ri x y = 1
r x

N
J

wi

wi

feasible
region

m
o
c
y = s1 x
i

xi >= a, yi >= b, xi yi >= A

or
xi >= b, yi >= a, xi yi >= A

xi >= a, yi >= b

xi >= a, yi >= b
or
xi >= b, yi >= a

(a) rigid, fixed

orientation

(b) rigid, free (c) flexible, fixed (d) flexible, free

orientation
orientation
orientation

xi >= a, yi >= b
xi yi >= A

11
www.specworld.in

11

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Area Computation
2

{ (5,5) (9,4) }

{ (6,2) (3,3) }

{ (2,3) (3,2) } { (2,2) }

.
w

u2

u1
v

w
w
u1

u2
w

u
t
jn

max{u1, u2}

v+w
u2

4 { (1,2) (2,1) } { (2,2) }

{ (1,3) (3,1) } { (2,3) (3,2) }

.
ld

r
o
w

V { (3,2) }

m
o
c

{ (3,5) (6,,4) }

N
J

U
T

{ (2,5) (3,4) } H

u1+u2

u1

max{v, w}

Wiring cost?
12
www.specworld.in

12

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Incremental Computation of Cost Function

Each move leads to only a minor modification of the Polish expression.

m
o
c

At most two paths of the slicing tree need to be updated for each move.

u
t
jn

5
3

r
o
w

.
w

E = 12H34V56VHV

w
w

M1

H
V

2
3

E = 12H35V46VHV

N
J

U
T

.
ld
V

13
www.specworld.in

13

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Incremental Computation of Cost Function

(contd)

u
t
jn

E = 12H34V56VHV

.
w

w
w
1

M3

.
ld
V

r
o
w

m
o
c

E = 12H34V56HVH
V

N
J

M2

U
T

E = 12H34V56VHV

E = 123H4V56VHV
14
www.specworld.in

14

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Annealing Schedule
Initial solution: 12V 3V . . . nV .
1

m
o
c

.
ld

r
o
w

= 1, 2, 3, . . .; r = 0.85.

Ti =

riT0 , i

.
w

# of accepted moves < 5%,

temperature is low enough, or

w
w

run out of time.

N
J

u
t
jn

Terminate the annealing process if

U
T

At each temperature, try kn moves (k = 510).

15
www.specworld.in

15

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

.
ld

u
t
jn

U
T

Algorithm: Simulated Annealing Floorplanning(P, , r, k)

1 begin
2 E 12V 3V 4V . . . nV ; /* initial solution */
avg
; M M T uphill 0; N = kn;
3 Best E; T0 ln(P
)
4 repeat
5 M T uphill reject 0;
6 repeat
7
SelectMove(M );
8
Case M of
9
M1 : Select two adjacent operands ei and ej ; N E Swap(E, ei , ej );
10
M2 : Select a nonzero length chain C; N E Complement(E, C);
11
M3 : done F ALSE;
12
while not (done) do
13
Select two adjacent operand ei and operator ei+1 ;
14
if (ei1 6= ei+1 ) and (2Ni+1 < i) then done T RU E;
15
N E Swap(E, ei , ei+1 );
16
M T M T + 1; cost cost(N E) cost(E);
cost
17
if (cost 0) or (Random < e T )
18
then
19
if (cost > 0) then uphill uphill + 1;
20
E N E;
21
if cost(E) < cost(best) then best E;
22
else reject reject + 1;
23 until (uphill > N ) or (M T > 2N );
24 T = rT ; /* reduce temperature */
25 until ( reject
> 0.95) or (T < ) or OutOf T ime;
MT
26 end

.
w

w
w

www.specworld.in

16

www.jntuworld.com

N
J

r
o
w

m
o
c

16

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Floorplanning by Mathematical Programming

Sutanthavibul, Shragowitz, and Rosen, An analytical approach to floorplan design and optimization, 27th DAC, 1990.
Notation:

.
ld

r
o
w

wi, hi: width and height of module Mi.

m
o
c

u
t
jn

U
T

(xi, yi): coordinate of the lower left corner of module Mi.

ai wi/hi bi: aspect ratio wi/hi of module Mi. (Note: We defined
aspect ratio as hi/wi before.)

.
w

w
w

Linear constraints? Objective function?

N
J

Goal: Find a mixed integer linear programming (ILP) formulation for

the floorplan design.

17
www.specworld.in

17

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

wi

D
L
R
O
u
t
jn

U
T

Area = hi * wi
Aspect ratio = wi / hi

.
ld

Mi

(xi, yi)

r
o
w

.
w

w
w
www.specworld.in

18

www.jntuworld.com

m
o
c

N
J

hi

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Nonoverlap Constraints
Two modules Mi and Mj are nonoverlap, if at least one of the following linear constraints
is satisfied (cases encoded by pij and qij ):
pij
qij
Mi to the left of Mj :
xi + w i xj
0
0
Mi below Mj :
yi + h i y j
0
1
1
0
Mi to the right of Mj : xi wj xj
Mi above Mj :
yi h j yj
1
1

m
o
c

.
ld

r
o
w

Let W, H be upper bounds on the floorplan width and height, respectively.

u
t
jn

.
w

w
w

wi

xj + W (pij + qij )
yj + H(1 + pij qij )
xj W (1 pij + qij )
yj H(2 pij qij )

wj

wi

hj

N
J

xi + w i
yi + h i
xi w j
yi h j

U
T

Introduce two 0, 1 variables pij and qij to denote that one of the above inequalities is
enforced; e.g., pij = 0, qij = 1 yi + hi yj is satisfied.

hi

(xj, yj)
(xi, yi)
xi + wi > xj

(xj, yj)
(xi, yi)
xi + wi <= xj

18
www.specworld.in

19

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Cost Function & Constraints

Minimize Area = xy, nonlinear! (x, y: width and height of the resulting
floorplan)
How to fix?

.
ld

Fix the width W and minimize the height y!

r
o
w

Four types of constraints:

m
o
c

u
t
jn

U
T

1. no two modules overlap (i, j : 1 i < j n);

.
w

3. xi 0, yi 0, 1 i n;

w
w

4. pij , qij {0, 1}.

wi, hi are known.

N
J

2. each module is enclosed within a rectangle of width W and height H

(xi + wi W, yi + hi H, 1 i n);

19
www.specworld.in

20

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Mixed ILP for Floorplanning

Mixed ILP for the floorplanning problem with rigid, fixed modules.
min

subject to

xi + wi W,
yi + hi y,
xi + wi xj + W (pij + qij ),
yi + hi yj + H(1 + pij qij ),
xi wj xj W (1 pij + qij ),
yi hj yj H(2 pij qij ),
xi , yi 0,
pij , qij {0, 1},

.
ld

w
w

Size of the mixed ILP: for n modules,

(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)

N
J

.
w

1in
1in
1i<jn
1i<jn
1i<jn
1i<jn
1in
1i<jn

U
T

r
o
w

u
t
jn

m
o
c

# continuous variables: O(n); # integer variables: O(n2 ); # linear constraints:

O(n2 ).
Unacceptably huge program for a large n! (How to cope with it?)
Popular LP software: LINDO, lp solve, etc.
20
www.specworld.in

21

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Mixed ILP for Floorplanning (contd)

Mixed ILP for the floorplanning problem: rigid, freely oriented modules.
min

.
ld

subject to
xi + ri hi + (1 ri )wi W,
yi + ri wi + (1 ri )hi y,
xi + ri hi + (1 ri )wi xj + M (pij + qij ),
yi + ri wi (1 ri )hi yj + M (1 + pij qij ),
xi rj hj + (1 rj )wj xj M (1 pij + qij ),
yi rj wj (1 rj )hj yj M (2 pij qij ),
xi , yi 0,
pij , qij {0, 1},

w
w

1in
1in
1i<jn
1i<jn
1i<jn
1i<jn
1in
1i<jn

(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)

N
J

.
w

U
T

r
o
w

u
t
jn

m
o
c

For each module i with free orientation, associate a 0-1 variable ri :

ri = 0: 0 rotation for module i.
ri = 1: 90 rotation for module i.
M = max{W, H}.
21
www.specworld.in

22

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Flexible Modules
Assumptions: wi , hi are unknown; area lower bound: Ai .
Module size constraints: wi hi Ai ; ai
Hence, wmin =

Ai ai , wmax =

wi
hi

Ai bi , hmin =

Ai
,
bi

.
ld

hmax =

r
o
w

wi hi Ai nonlinear! How to fix?

m
o
c

bi .

Ai
.
ai

j
.
w

w
w

/ y = mx + c /
/ slope /
/ c = y0 mx0 /

N
J

u
t
n

h i = i w i + c i
hmax hmin
i =
wmin wmax
ci = hmax i wmin

U
T

Can apply a first-order approximation of the equation: a line passing through

(wmin , hmax ) and (wmax , hmin ).

Substitute i wi + ci for hi to form linear constraints (xi , yi , wi are unknown; i , j ,

ci , cj can be computed as above).

22
www.specworld.in

23

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

h
Ai = wi * hi

D
L
R
O

h max

hi =

i wi

+ ci

m
o
c

h min

.
w

w
w
www.specworld.in

U
T

r
o
w

u
t
jn

.
ld

wmax

24

www.jntuworld.com

N
J

wmin

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Reducing the Size of the Mixed ILP

Time complexity of a mixed ILP: exponential!

m
o
c

Recall the large size of the mixed ILP: # variables, # constraints: O(n2 ).
How to fix it?

.
ld

Key: Solve a partial problem at each step (successive augmentation)

r
o
w

Questions:

How to select next subgroup of modules? linear ordering based on connectivity.

u
t
jn

w
w

N
J

.
w

U
T

How to minimize the # of required variables?

23
www.specworld.in

25

www.jntuworld.com

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

Next group
of modules

www.specworld.in

26

www.jntuworld.com

N
J

w
w

U
T

.
w

.
ld

Partial
floorplan

r
o
w

u
t
jn

m
o
c

www.smartzworld.com

www.jntuworld.com
www.jntuworldupdates.org

D
L
R
O

.
ld

u
t
jn

U
T

Reducing the Size of the Mixed ILP (contd)

Size of each successive mixed ILP depends on (1) # of modules in the next group; (2)
size of the partially constructed floorplan.

m
o
c

Keys to deal with (2)

Minimize the problem size of the partial floorplan.

Replace the already placed modules by a set of covering rectangles.

r
o
w

# rectangles is usually much smaller than # placed modules.

.
w

w
w

Horizontal
cut edges

(a)

C3

C4
C2

(b)

R4

N
J

Dead space

R5

R3

R2

C1

R1

(c)

(d)

24
www.specworld.in

27

www.jntuworld.com

www.smartzworld.com