Floorplan Sizing 1. Floorplanning Algorithms Common Goals To minimize the total length of interconnect, subject to an upper bound on the floorplan area.

Floorplan Sizing

1

Floorplanning Algorithms

• Common Goals To minimize the total length of interconnect, subject to

an upper bound on the floorplan area

• or

· To simultaneously optimize both wire length and area

2

3

3.5.1Floorplan Sizing

Shape functions

Legal shapes

w

h

w

h

Block with minimum width and height restrictions

ha*aw A Ott

en, R

.: E

ffic

ien

t F

loor

pla

n O

pti

miz

atio

n. I

nt.

Con

f. o

n C

omp

ute

r D

esig

n, 4

99-5

02, 1

983

Legal shapes

4


Shape functions

w

h

Hard library block Ott

en, R

.: E

ffic

ien

t F

loor

pla

n O

pti

miz

atio

n. I

nt.

Con

f. o

n C

omp

ute

r D

esig

n, 4

99-5

02, 1

983

w

h

Discrete (h,w) values

5


Corner points

52

2

5

2 5

2

5

w

h

مرتضي صاحب الزماني

6

Slicing Floorplan Sizing

Floorplan Sizing• Algorithm:

finds the minimum floorplan area for a given slicing floorplan in polynomial time.

For non-slicing floorplans, the problem is NP-hard.

Construct the shape functions of all individual blocks

• Bottom up: Determine the shape function of the top-level floorplan

from the shape functions of the individual blocks

• Top down: From the corner point (minimum top-level floorplan

area), trace back to each block’s shape function

− Find that block’s dimensions and location

7

8

4

2

2

4Block B:

Block A :5

5

3

3

Step 1: Construct the shape functions of the blocks

3.5.1Floorplan Sizing – Example

9

4

2

2

4Block B:

Block A :5

5

3

3


2

4

h

6

w2 64

5

3


10

4

2

2

4Block B:

Block A :5

5

3

3


2

4

h

w2 64

6

3

5


11

4

2

2

4Block B:

Block A :5

5

3

3

w2 6

2

4

h

4

6

hA(w)



12

4

2

2

4Block B:

Block A :5

5

3

3

hB(w)

w2 6

2

4

h

4

6

hA(w)



13

w2 6

2

4

h

4

6

hB(w)hA(w)

8


Step 2: Determine the shape function of the top-level floorplan (vertical)

14

w2 6

2

4

h

4

6

hB(w)hA(w)

8



15

w2 6

2

4

h

4

6

w2 6

2

4

h

4

6

hB(w)hA(w)

hB(w)hA(w)

hC(w)

88



16

w2 6

2

4

h

4

6

w2 6

2

4

h

4

6

hB(w)hA(w)

hB(w)hA(w)

hC(w)

5 x 5

88



17

w2 6

2

4

h

4

6

w2 6

2

4

h

4

6

hB(w)hA(w)

hB(w)hA(w)

hC(w)

3 x 9

4 x 7

5 x 5

88



18

w2 6

2

4

h

4

6

w2 6

2

4

h

4

6

hB(w)hA(w)

hB(w)hA(w)

hC(w)

3 x 9

4 x 7

5 x 5

88

Minimimum top-level floorplan

with vertical composition



19

w2 6

2

4

h

4

6

w2 6

2

4

h

4

6

hA(w)hB(w) hC(w)hA(w)hB(w)

9 x 3

7 x 4

5 x 5

88


Minimimum top-level floorplan with horizontal composition


20


Step 3: Find the individual blocks’ dimensions and locations

w2 6

2

4

h

4

6

8

(1 )Minimum area floorplan: 5 x 5

Horizontal composition

21

w2 6

2

4

h

4

6


(2 )Derived block dimensions : 2 x 4 and 3 x 5

8




22

2 x 4 3 x 5

5 x 5



w2 6

2

4

h

4

6


(2 )Derived block dimensions : 2 x 4 and 3 x 5

8


23

2 x 4 3 x 5

5 x 5

Resulting slicing tree

B

V

AB A


24

DeFer:Deferred Decision Making Enabled

Fixed-Outline Floorplanner

DAC 2008

Jackey Z. Yan Chris Chu

25

Fixed-Outline Floorplanning

26

Constraints of Ordinary Slicing Tree

A

B

C

EDHH

V

BA C V

D E

• Block orientation

• Slice line direction (H/V)• Left-right or top-bottom relative order

27

DDM in Shape/Orientation Defer the decision for subfloorplan (i.e., block) orientation until

the end

HH

V

BA C V

D E

Pick any point!

Fixed-outlineregion

A

B

C

ED

H

W

A

A

28

DDM in Relative Order The relative order (left-right / top-down) between

subfloorplans does not affect the shape curves

A B

V

BA A

V

ABB

Same shape curve

Decision on subfloorplan relative order will be made at the end to minimize HPWL

H

W

ABVH

W

BAV

V

BA

V

BA

V

B Aand

29

DDM in Slice Line Direction Two combine operators in ordinary slicing tree

AB

H

BA

A B

V

BA

AB

H

BA

A B

V

BABAand

Generalized combine operator

30

Floorplans by Generalized Slicing Tree

A B

• Block orientation

A B

A B

A BA B

AB

AB

ABAB

A

B

A

BA

BA

B

A

B

A

BA

BA

B

• Slice line direction• Left-right or top-bottom relative order

Deferred Decision Making (DDM)

31

Shape Curve Operation for

H

W

A B

)i ( Addition

Ch

Cv

)iii ( Merging

H

W

W=H

Cv

C

A B

CA B C

Ch

)ii ( Flipping

H

W

W=HCh

DeFer

Binary and source code in

− http://www.public.iastate.edu/~zijunyan/

32

Cluster Growth• Cluster growth:

Iteratively add blocks to the cluster until all blocks are assigned

− Merged either horizontally, vertically, or diagonally with the cluster.

Only the different orientations of the blocks (instead of the shape/aspect ratio) are taken into account

33

w

h

w

h

w

h

a a a

b bc

Growth direction

2

4

6

4

6

4

©20

11

Spr

inge

r V

erla

g

Cluster Growth• Cluster growth:

Blocks b and c are placed so that the increase to the floorplan’s dimensions is minimum.

• Linear ordering to minimize total wirelength of connections between

blocks

34

w

h

w

h

w

h

a a a

b bc

Growth direction

2

4

6

4

6

4

©20

11

Spr

inge

r V

erla

g

Linear Ordering

• Linear-ordering algorithms often invoked to produce initial placement solutions for

iterative-improvement placement algorithms.• Linear-ordering objective:

Arrange blocks in a single row so as to minimum total wirelength of connections

Initial block:

− Either randomly− Or based on number of connections to others

35

Linear Ordering

• Categories of nets for a block: Terminating nets:

− have no other incident blocks that are unplaced.

New nets

− have no pins on any block from the partially-constructed ordering.

Continuing nets

− have at least one pin on a block from the partially-constructed ordering and at least one pin on an unordered block.

36

Linear Ordering

37

…

Terminating nets New nets

Continuing nets

Cluster Growth Example

• cluster growth.ppt

38

Cluster Growth Example• Given:

(1) blocks a-e (2) linear ordering <a b d e c>.

• a: (wa = 2, ha = 3) or (wa = 3, ha = 2)

• b: (wb = 2, hb = 1) or (wb = 1, hb = 2)

• c: (wc = 2, hc = 4) or (wc = 4, hc = 2)

• d: (wd = 3, hd = 3)

• e: (we = 6, he = 1) or (we = 1, he = 6)

• Task: find a floorplan with minimum global bounding box area.

39

Growth direction


40


41

Floorplan Sizing 1. Floorplanning Algorithms Common Goals To minimize the total length of interconnect, subject to an upper bound on the floorplan area.

Documents

w h slide

floorplan sizing example

slicing floorplan sizing

hbwhbw w

w h discrete h

floorplan sizing algorithm

level floorplan vertical

blocks slide