You are on page 1of 15

0'1 I

81 r.rr

(a) (tIJ

Fipre Ul Ch~el connectivity graph. (a) A building-block !.ayouL (b) Corresponding channel connectIVIty grapb.

COl£- ~ n nR £ ~?'t? ac-e"J

'" v<' .e Y-'" Jc/ ''''' hr'" r:»

(.1

?YJ1t? •

Fipft 6.U Bottleneck graph. (a) A .. lb)

grapb. building-.block layout, (b) Corr .

esponding bottleneck

. , .

~~._+ ~ . ; ~_ _~&.~~___ • t.t

r-~_t~~-· ·-~·r-& ~ .. ~~~-~-

----t .... ····f----

.... :----- &&&·t·---

. . .

'-'-f--' .... +--- .... j ... --

~~"'~1----- ~&&·T~·-- ----·r-rr--

Fipre 6.14 Grid graph. (a) A two-dimensional grid. (b) Corresponding grid grapb where eacb global ceO (3 )( 2 grid cells) i!l modeUed by a vertex.

,

/DU //OV /~ / :" d<k5foa?J

lflU) a;.u fi:Jc<CPv> c/o £;t.._ .. ~ ,~.

&nce/ f

o I 2 3 4 5 6 (= n)

o I 2 3 4 5 6 (= n)

.---~----~----.----p-~~-- .. -- 111 II " II II • II

!. ~~~ .:(Ell4~ ~~ ~~~~:~;:$~ ~. i

~. "'{47 ~ :~~~hH4~:~i ;s. ~ ;.~~::~:.1:~~7.a~:J~~x~:~:,.~

• • • • II • ..

l1li 10 • • II • II

!' a:~ n€ a~ ,~~:"'~751~XT:i!: ,- ~

It • ,. ... • .. ..

~ . ,. .. . .. .

- ~ ,

~,..;..~~e;.~:

3 •• ..~

• .. • III ..

.. • ... III .. • ,.

4 :. ~:~r·l:u.;: 4~~ ,~~ 3r0Sl7 E~: ~. ~

• • II" III • • "

5 (= m) :. ... .:... ••.. ~ ... _.:. - - _!. -- _!. - - -:

(a)

(b)

Figure 9.1' The partitioning of the standard-cell layout area by a grid for the sake of global

routing (a) and a rectilinear Steiner tree embedded in this grid (b), ~

/, 10 r"iJa-> ~ .r: ,,~ -1'

./' . ,#"# cJ I "- J rv»

(/1 VIC/ ,lrl0' ) } /'

(/ l/~~rL/ AI'?le/)

:j:,e" tJ I - - J n I /

- u,,"~_/ ~6f- Vv/i) . hi ":>:

/'V) r., 0/' i .. ~ t" .L:,,) /s .: ,/

(~lV-5~;' ?n,n) ~C)J£d~/ /?u

"YP7 If r ,41 tf,

• /1'0 -:» e. -i,L/cL.,,,J ~~ »": j

In

2 Pv- (;-)

(-::;

f

k / 1;;:; / I. / /. -J -: H' /; < t: ~ no-j

• £fIf£ .o -vfy (" /Y>§ ;iz<,,, !1 '-/1. (5) - 1/ ,£ " ~ f~ j

~ 1/" zz: 7n~,,," /..-d#TZlCl/ c~ c.ui<I/-A (:hi

/cc-..i A7.( tJiteJ) /,~ (~ .... Rt&-- t.'

Oc:~pn => L", /-' 1" j

FIJ

\t/ S7~AaI3bn'7

/-. (tfflS/~JM.-5. fifo 0<'.-/ c; /;{y ;?""J,

~ d muA. ,.;/ y~

f e:7/tr~.kK k7,4~

,kyf?J ='~7

C.~~/ d",~?

/L£/ RocJ?-

A/?ProeJ/0

I
'T -

J <II

£-- J
--I 5~ ArC.'

!1Jll _~?Y'lrJ /~

') j /v;,~i"'~ e.: ~'-'"7 ~~. (rti? ~ r).

Ai),/, §' If;?, ~l-

~?T ----#-

I1~Spf /JeT>~ L

pIP/:; u/ t9- ~uJ!U

\1 ('(' \/('7 tft> 01 rnc(f-e FJ<cvf/J C<h'/N' (s).

_ ;?e jlPc:£! IfOC.(?61 A .A~J ~~ahJ

_/7// >,,0 /:': 'rr/,=u"C~1 (ivJcc i"'f/,vcX ~-I =:

- L (/4 SJ ?.p I r; 1>& '''/! !p., /.J I~ c;<_

:? ~ I~ P it. ¥7 «.:., ~

- tJ=f.J .: !1 ror rr ) ""-.- ~~ ~ ~

,/1 /-,rrC'nn"cA ~dI /,//,,1< t r'n P ",~t!

1M, i o~ /?!J" -/' /,p,,,/J In ffi ~e.

(re) ~f' t1r/fl-!~ fbtnA c: rI?I'4R ~ 3) nol fr jaAr pomh ff> 'n/u-onnrd, k-,.{_~ fY ".o",/; On -;4!&)

1 Ike j/t Ir,j ~r /lc /u u.--ii;, Ih~1'7P; .

/LJ nm-(l fhmls f! r ?:- v. ~

r:dl/'d /#i/)if »:" ~ cI~/; :;

II I' ~ !1 R s; I (aA> t.v<- M'" I~fl

'A ~ /2J1nft; fI._, 11J?>7 IJ

~r)'

I I I I I I I

I I 1 I I I I

o .00 0







• 000 •





o o. 0 0

(b)

Figure 9.20 A set of points P for whi .

Hanan points (b) or which the Steiner-tree should be

. constructed (a) and its

(a)

II ~~ / /ovrVdi / ieu->,/f ~ u.¥' g;J ~ ah£- d ""'051 71-2- s~ pm;' /: c< fplnl sv! F - -- p-",,,,,! c:~ ffi I,; /1",/ e=J ~

~ /" fl .~ C_/ f" rftM In .;-~

· HI? 5/T (f) 15 a/lrox.

SJelIJev -tr~~

(set of struet vertex, set of stmct edge) steinenset of struct vertex P)

I

sd of strvct vertex T; set ofstruct edge E, F; int gain;

E +- prime P);

(T, F, gain) - l-steiner(P, E); wbile (gain> 0) {

P +- T; E -F;

(T, F. gain) +- l·steil'ler(P, E);

)

return (P, E); }

(set of stntct vertex, set of strud edge, int) t-steinertset of struet vertex V. set of struc! edge £)

set of slruct vertex W; set of slrud edge F; struct vertex maxpoint: inl gain, maxgain;

rnaxgain +- 0;

for each S E "Hannan points of V" I

(W, F, gain) +- spanning.updatet V. E. r): if (gain> maxgain) {

maxgain <- gain;

max point <- s;

Figun 9.19 The iterated l -Steiner heuristic for Steiner-tree construction.

(8)

(b)

(e)

I I I I I I I

II I " I I

I I I I I I I

(d)

(e)

(0

I I I I I I I

I I I I I I I

I I II I I I

} ,

if(maxgain > 0) I

(W, F, gain) +- spanning.updaterV. E, s); return (W, F, maxgain);

else return (V, E, 0); }

Fignre 9.21 The pseudo- code of the I-Steiner algorithm.

(set of struet vertex, set of struct edge, int)

spannmg.updatetset of struet vertex V, set of struet edge E. struct vertex s)

{

Int deka;

struct vertex u, v, w;

delta <- 0:

V-VU{s};

for each d E (north. east. south, west) f u +- closest.poimt V, S, d);

delta ._ delra- distancets, u);

E +- E U (s, u)};

if{cycle(V, E) I

(v, w) +- Iargest.cycle.segmenn V • E); E +- E \ {(v. w»);

delta +- delta + distancetu, w);

I )

return (V, E, delta); )

Fignre 9.22 The function thai incrementally computes a spanning tree when a new point is added to the original point set.

I I I I I I I

I I I I I I I

I I I I I I I

(a)

(c)

I I I I I I I

I I I I I I I

~ I I I I I I

(d)

(e)

(1)

I I I I I I I

I I I I I I I

I I I I I I I

(g)

(h)

(0

Figure 9.23 The different steps in the incremental computation of a spanning tree.

I I I I I I I

P2., P3·

PI tis

.-0-0-.

I P4

Figure 9.24 The solution found by the iterated I-Steiner heuristic for the problem instance

Cc7yry/IK/ I; "

- 5r"",,,? (/irWe " e(y

.i - S;/u"", T~ -f?t(1 <cJ/J!p S,fAp":;r

LfW'.e ffr /!-'h S~ po,,>~

=0 &fr'~

- ~k,n'r ~~1-(> &/n) raA h $- Sk,." =f7 (3 In)

I I I I I I I I I I I I I I I I I I I I I

• °l 01.
Lo


(a) (b)
I ! I I I I I I I I I
I I I I I

• ] :J
I


(c)
I I I I I I I I I I I I I I

• •
•• Lo_. I
.-0-.
L:..o..l I
I
• •
(d)
Figuno 9..25 local transformations for Steiner trees.

You might also like