Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
13
,
No.
3
,
Ma
rch
201
9
, p
p.
1048
~
1055
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
3
.i
3
.pp
1
048
-
1
055
1048
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
Des
i
gn o
f modifie
d booth b
ase
d m
ulti
pli
er
with
car
ry
pre
-
comp
utation
Chaitany
a C
V
S
1
,
Su
nd
ares
an C
2
,
P
R Ven
ka
t
esw
ar
an
3
, Ke
erth
ana Pr
as
ad
4
1,2,4
School
of
In
f
orm
at
ion
Sci
ences,
Manip
al Academ
y
of
Higher
Educ
a
ti
on,
Indi
a
3
Bhara
t
Hea
v
y
E
le
c
tri
c
al
s
Li
m
it
e
d,
Indi
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Oct
1
1,
2018
Re
vised
D
ec
9
,
2018
Accepte
d Dec
2
2
, 201
8
Arithmeti
c
un
it
is
the
m
ost
important
component
of
m
oder
n
embedde
d
computer
s
y
s
te
m
s.
Arithmeti
c
uni
t
gen
erall
y
inc
lu
des
floating
poin
t
and
fix
ed
-
point
arithm
et
i
c
oper
ations
and
tr
igonometri
c
fun
ct
ions.
Mul
ti
pl
iers
unit
s
are
the
m
ost
import
ant
h
ard
ware
str
uct
ure
s
in
a
co
m
ple
x
arithm
et
i
c
uni
t.
W
it
h
inc
re
ase
in
chi
p
fre
quency
,
th
e
d
esigne
r
m
ust
be
abl
e
to
find
the
best
set
of
tra
de
-
offs
.
The
abi
lit
y
for
fast
e
r
comput
at
ion
i
s
essential
to
a
chi
ev
e
high
per
fo
rm
anc
e
in
m
an
y
DS
P
and
G
rap
hic
proc
essing
al
gor
it
hm
s
and
is
wh
y
th
ere
is
at
l
ea
st
on
e
de
dic
a
te
d
Mult
ipl
i
er
unit
in
a
ll
of
t
he
m
oder
n
comm
erc
ia
l
DS
P
proc
essors
.
Trem
endous
adva
n
c
es
in
VLSI
tech
nolog
y
ov
er
the
past
sev
era
l
y
e
ars
resul
te
d
in
an
inc
r
ea
sed
ne
ed
for
h
igh
spe
e
d
m
ult
ipl
i
ers
an
d
compelled
the
design
ers
to
go
for
tra
d
e
-
offs
among
spee
d,
p
ower
consum
pti
on
and
ar
ea.
A
novel
m
odifi
e
d
booth
m
ult
ipl
i
er
design
for
hi
gh
spee
d
VLSI
appl
i
ca
t
ions
using
pre
-
computat
ion
logi
c
h
as
be
en
pr
ese
nt
ed
in
th
is
p
ape
r
.
Th
e
proposed
arc
hi
te
c
ture
m
odel
ed
using
Veri
log
HD
L,
sim
ul
at
ed
using
Cadence
NCS
IM
and
s
y
nth
esized
using
Cade
nc
e
R
TL
Com
pil
er
wi
t
h
65nm
TSMC
l
ibra
r
y
.
The
proposed m
ult
iplier ar
chitecture
i
s
compare
d
with
the
exi
sting m
ul
ti
pliers
and
the
r
esult
s show
signifi
c
a
nt im
prove
m
ent
in
spee
d
and
pow
er
diss
ipa
ti
on
.
Ke
yw
or
d
s
:
Bi
nar
y M
ulti
pl
ic
at
ion
Ca
rr
y P
re
-
C
ompu
ta
ti
on
Mod
ifie
d
B
oo
t
h
Mult
ipli
er
Mult
ipli
er Arc
hitec
ture
Ved
ic
M
ulti
plier
Copyright
©
201
9
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights
reserv
ed
.
Corres
pond
in
g
Aut
h
or
:
Chait
anya CV
S
School
of In
for
m
at
ion
Science
s,
Ma
nip
al
Acad
e
m
y of
Higher
Ed
ucati
on,
Ma
nip
al
576104, Ka
rn
at
a
ka,
I
nd
ia
.
Em
a
il
:
chait
anya.cvs@
m
anip
al
.ed
u
1.
INTROD
U
CTION
Pr
oc
esso
rs
are i
m
po
rtant p
a
rt o
f
integ
rated
ci
rcu
it
s
(I
C)
.
La
r
ge
nu
m
ber
s of f
unct
io
nalit
ie
s
are p
acke
d
in
an
IC
tha
nk
s
to
trem
endous
gro
wth
i
n
de
ns
it
y
of
i
nteg
r
at
ion
in
rece
nt
tim
es.
As
the
nu
m
ber
of
f
unct
ion
s
increases
,
t
he
need
for
com
pu
ta
ti
on
al
so
grow
s
.
W
it
h
t
he
adv
e
nt
of
ne
w
pr
ocess
te
ch
nolo
gies,
s
hr
in
k
ing
of
featur
e
siz
e
an
d
a
vaila
bili
ty
of
m
od
er
n
CA
D
to
ols,
a
dev
el
op
m
ent
of
c
om
plex
integrat
ed
ci
rcu
it
s
f
or
var
i
ou
s
app
li
cat
io
ns
is
po
s
sible.
E
xa
m
ples
of
su
ch
app
li
cat
ions
inclu
de
di
gital
sign
al
processi
ng
[1
]
,
[
2]
,
m
ob
il
e
com
pu
ta
ti
on
s
a
nd
c
omm
un
ic
at
ion
s,
m
ultim
e
dia
ap
plica
ti
ons
and
proce
ssing
re
qu
ire
d
for
sci
entifi
c
com
pu
ti
ng
and
ap
plica
ti
ons
et
c.
The
s
peed
an
d
e
ff
ic
ie
ncy
of
proc
essor
in
su
c
h
IC
is
ve
ry
cr
uc
ia
l
fo
r
m
eet
i
ng
the
requirem
ents
of
the
a
pp
li
cat
io
ns
s
upporte
d
by
the
IC.
T
he
s
peed
of
proce
s
so
r
a
nd
ef
fici
ency
of
proce
ss
or
i
n
-
tur
n
de
pe
nds
upon
a
n
arit
hm
et
ic
log
ic
unit
[3]
w
hich
is
co
ns
ide
red
as
the
m
ai
n
com
pu
ta
ti
onal
unit
of
th
e
process
or.
More
ov
e
r,
t
he
m
ul
ti
plier
un
it
s
[4]
are
t
he
m
os
t
i
m
po
rtant
ha
rdwar
e
struct
ures
in
a
c
om
ple
x
arit
hm
et
ic
un
it
.
Th
e
m
ulti
plier
un
it
s
a
re
capa
ble
of
pe
rfor
m
ing
op
e
r
at
ion
s
on
oper
ands
of
va
rio
us
data
ty
pes
s
uch
as
cal
culat
ing
r
unning
su
m
of
products.
As
m
ulti
plica
ti
on
is
a
cru
ci
al
a
rithm
e
ti
c
op
e
rati
on
i
n
pr
ocess
or
s
[
5]
a
nd
dig
it
al
com
pu
t
er
syst
em
s,
m
ulti
pliers
a
re
t
he
c
ore
buil
din
g
blo
c
k
for
m
any
al
gorith
m
s
in
a
wi
de
var
ie
ty
of
com
pu
ti
ng
a
pp
li
cat
ion
s.
Alt
houg
h
m
ulti
plier
s
are
m
ai
n
ari
thm
e
ti
c
co
m
ponen
ts
us
e
d
for
processi
ng
sci
entifi
c
data,
t
he
e
xces
sive
power
c
onsu
m
ption
an
d
delay
at
tract
s
at
te
ntion
from
the
researc
h
c
omm
un
i
ty
.
Usu
al
ly
,
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Desig
n of
modi
fi
ed
boot
h base
d mu
lt
ipli
er wi
th carry
pr
e
-
c
omp
uta
ti
on
(
Ch
aitan
y
a
CV
S
)
1049
m
ul
ti
ple
arit
hm
et
ic
cor
es
w
orkin
g
in
pa
rall
el
are
us
e
d
so
as
t
o
proces
s
la
rg
e
am
ou
nts
of
data
wit
h
re
la
ti
vely
low p
ow
e
r
a
nd
d
el
ay
.
Howe
ver,
i
n
t
he
hi
gh
-
s
pee
d
proces
sors
w
hic
h
a
re
ope
rati
ng
at
higher
cl
oc
k
fr
e
quencies
,
the
e
xisti
ng
m
ul
ti
plier
ta
ke
s
m
or
e
delay
f
or
e
xecu
ti
on
of
the
in
structio
ns.
T
he
e
xisti
ng
m
ulti
plier
unit
s
that
c
on
s
um
e
m
or
e
powe
r
are
no
t
su
it
able
to
be
i
ncor
porated
in
the
process
ors
wh
ic
h
are
u
se
d
in
wi
reless
a
nd
portable
de
vices.
Th
us
, p
ow
e
r
sa
vings is a
n
im
po
rta
nt ar
ea
for
i
m
pr
ovem
ent.
In
order
t
o
ad
dr
ess
t
he
lo
w
powe
r
com
pu
t
at
ion
al
ong
wi
th
high
perfor
m
ance,
a
ne
w
appr
oach
t
o
m
ul
ti
plier
design
base
d
on
a
nc
ie
nt
Ve
dic
Ma
them
atics
has
been
ex
pl
ored
.
The
m
at
he
m
atical
op
erati
ons
us
i
ng
Ved
ic
m
at
he
m
at
ic
s
are
ver
y
f
ast
and
re
quire
le
ss
hardw
a
re.
This
aspect
of
Ved
ic
m
at
hem
at
ic
s
can
be
util
iz
ed
to
in
crease
the
com
pu
ta
ti
onal
s
peed
of
m
ult
ipli
ers.
This
pa
per
descr
i
bes
the
desi
gn
a
nd
i
m
ple
m
entat
io
n
of
a
Ved
ic
m
ulti
pli
er
base
d
on
U
rd
hv
a
-
Tiry
agbhya
m
S
utra
[
6]
-
[9
]
.
T
he
num
ber
of
ste
ps
require
d
to
pe
rfor
m
a
m
ul
ti
plica
ti
on
operati
on
by
us
in
g
Ur
dh
v
aT
iry
agbhya
m
Su
tra
are
c
on
si
der
a
bly
le
ss
com
par
ed
to
th
e
conve
ntion
a
l
m
ul
ti
plica
ti
on
te
chn
i
qu
e
s
.
In
t
his
pa
pe
r,
we
ha
ve f
urt
her
ex
pl
or
e
d a
novel
m
et
ho
d t
o en
ha
nce
the
sp
ee
d
of
a
Ve
di
c
m
ulti
plier
by
pre
-
com
pu
ti
ng
the
carries
w
hich
are
us
e
d
du
r
ing
s
umm
a
ti
on
of
par
ti
al
pro
duct
s
.
The
im
ple
m
ent
at
ion
of
pr
e
-
c
om
pu
ta
ti
on
log
i
c
us
in
g
m
ulti
plexer
-
ba
sed
ca
r
ry
-
lo
ok
ahea
d
l
og
ic
a
nd
X
OR
log
ic
resu
lt
ed
in
re
duct
ion
of
dela
y.
The
pro
po
s
ed
ca
rr
y
pre
-
c
om
pu
ta
ti
on
is
us
e
d
al
on
g
wi
th
Mo
difie
d
B
oo
t
h
m
et
ho
dolo
gy re
su
lt
ed
i
n
f
urt
he
r
re
duct
ion o
f dela
y i
n per
for
m
ing
m
ulti
plication op
e
rati
on
.
The
st
ru
ct
ur
e
of
the
pa
per
is
div
ide
d
a
s
f
oll
ow
s:
The
m
et
ho
dolo
gy
a
nd
th
e
arc
hitec
ture
of
the
pro
pose
d
m
ul
ti
pliers are give
n
in
secti
on
2.
Res
ults ar
e presente
d
in
sect
ion
3. Final
ly
, co
ncl
us
io
n
i
s g
i
ven in secti
on 4.
2.
RESEA
R
CH MET
HO
D
2.1.
C
arry
Pre
-
C
omp
u
t
at
i
on
B
as
ed Bi
n
ar
y Multipli
er
An
8
-
bit
Bi
nary
Ved
ic
Mult
ip
li
er
has
be
en
pro
posed
with
A
an
d
B
as
in
puts
an
d
P
as
t
he
final
16
-
bit
pro
du
ct
.
The
blo
c
k
diagr
am
for
8
-
bit
m
ult
ipli
cat
ion
is
s
how
n
in
F
ig
ure
1
.
In
the
propose
d
m
ulti
pl
ie
r
the
op
e
ra
nd
s
A an
d
B a
re
div
id
ed
into Hig
he
r
a
nd L
ower
pa
rts
with
4
-
bits eac
h.
A
= {
A
H,
AL}
(
1
)
B = {B
H, BL}
(2
)
AL
*BH
AH*BH
AL
*A
L
AH*
AL
Pr
od
uct
Figure
1
.
Bl
oc
k Diag
ram
o
f
8
-
bit M
ulti
plica
ti
on
In
t
his
ty
pe
of
m
ulti
plier
an
8
-
bit
Bi
na
ry
m
ulti
plicatio
n
is
reali
zed
us
in
g
4
-
bit
bin
a
ry
ve
dic
m
ul
ti
plica
ti
on
us
in
g ca
rr
y
pr
e
-
com
pu
ta
ti
on
log
ic
s
how
n i
n
F
igure
2
,
w
here
A
3
,
A
2
,
A
1
,
A
0
&
B
3
,
B
2
,
B
1
,
B
0
are
4
-
bit
bi
nar
y i
nputs
and
P
7
,
P
6
,
P
5
,
P
4
,
P
3
,
P
2
,
P
1
,
P
0
a
re t
he b
inary
ou
t
pu
t
bi
ts.
The
pa
rtia
l
pro
du
ct
ge
ner
at
or
is
the
first
blo
c
k
of
the
m
ulti
p
li
er
to
w
hich
t
he
4
-
bit
m
ulti
plica
nd
a
nd
m
ul
ti
plier
are
gi
ven
as
i
nputs.
At
this
ju
nctu
r
e,
the
m
ulti
plication
te
c
hn
i
qu
e
us
e
d
is
Urd
hva
-
Tiry
ag
bhya
m
.
The
4
-
bit
m
ulti
plication
r
esults
in
a
total
of
16
pa
rtia
l
products
(
pp
1
-
pp
16
)
.
T
he
resu
lt
of
m
ultip
ly
ing
a
ny
on
e
bin
a
ry
bit wit
h
a
nothe
r
is ei
the
r
a
ze
r
o or a
on
e
whi
ch
is si
m
ply t
he
log
ic
of
ANDin
g of t
he
t
w
o bit
s.
The
seco
nd
sta
ge
i
n
the
bl
ock
dia
gr
am
is
the
car
ry
ge
ner
at
ion
ci
rc
uit.
He
r
e,
we
ha
ve
inte
gr
at
e
d
pre
-
com
pu
ta
ti
on
l
og
ic
al
ong
wit
h
the
Urdhva
-
Tiry
agbhyam
m
ul
ti
plica
ti
on
te
chn
iq
ue.
T
he
carry
eq
uatio
ns
a
re
gen
e
rated
sepa
ratel
y
for
eac
h
colum
n
of
par
t
ia
l
products
a
nd
the
in
pu
ts
f
or
these
e
quat
ions
are
ta
ken
f
rom
the
pr
e
vious c
olum
n.
Th
e e
qu
at
i
on
s
for p
re
-
co
m
pu
te
d
carries
are give
n
.
c
2
= pp
5
&
pp
2
;
(
3
)
c
3
t
1
= (pp
6
&
pp
3
)
|
(pp
9
&
(pp
3
| pp
6
));
(4
)
c
3
t
2
= (pp
9
&
~
pp
6
) |
(pp
3
&
~
pp
9
)
| (~
pp
3
&
pp
6
);
(5
)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
3
,
Ma
rc
h 201
9
:
1
048
–
1
055
10
50
c
31
= c
2
?
c
3
t
2
:c
3
t
1
;
(6
)
c
32
= pp
2
&
pp
5
&
pp
3
&
pp
6
& pp
9
;
(7
)
c
41
t
1
= p
p
13
?
(
(
pp
10
& ~
pp
7
) |
(pp
4
&
~p
p
10
)
|
(
~p
p
4
&
pp
7
))
:( (
pp
7
&
pp
4
)
|
(pp
10
&
(
pp
4
| p
p
7
)
))
;
(8
)
c
41
t
2
= p
p
13
?
(
(
~pp
7
& ~
pp
4
) |
(~pp
10
&
(~pp
4
| ~pp
7
)
)
)
:(
(
~
pp
7
&
pp
4
)
| (
pp
10
& ~p
p
4
)
|
(~pp
10
&
pp
7
));
(9
)
c
41
= c
31
?
c
41
t
2
:c
41
t
1
;
(1
0)
c
42
= ((
c
31
&
pp
13
)
&
(
(
pp
10
&
(
pp
7
|
pp
4
))
|
(pp
7
&
pp
4
)
)) | (
(
pp
10
&
pp
7
&
pp
4
)
&
(c
31
|
pp
13
)
);
(11
)
c
51
t
1
= c
32
?
(
(pp
14
& ~
pp
11
) |
(
pp
8
& ~
pp
14
)
|
(~pp
8
&
pp
11
)
) :
(
(
pp
11
&
pp
8
) |
(pp
14
&
(pp
8
| p
p
11
)));
(12
)
c
51
t
2
= c
32
?
(
(~
pp
11
& ~
pp
8
) |
(~pp
14
&
(~pp
8
| ~pp
11
)
)
)
:(
(
~
pp
11
&
pp
8
)
|
(pp
14
& ~
pp
8
) |
(~pp
14
&
pp
11
));
(13
)
c
51
= c
41
?
c
51
t
2
:c
51
t
1
;
(14
)
c
52
= ((
c
41
&
c
32
)
&
((p
p
14
&
(
pp
11
|
pp
8
))
|
(pp
11
&
pp
8
)
))
|
((pp
14
&
pp
11
&
pp
8
)
&
(c
41
|
c
32
));
(15
)
c
6
t
1
= (pp
12
&
pp
15
)
| (c
42
&
(
pp
12
|
pp
15
));
(16
)
c
6
t
2
= (c
42
& ~
pp
15
) |
(pp
12
&
~c
42
)
| (
pp
15
&~
pp
12
);
(17
)
c
61
= c
51
?
c
6
t
2
:c
6
t
1
;
(18
)
c
62
=
c
51
& c
42
& pp
12
&
pp
15
;
(19
)
c
71
= (c
52
&
pp
16
)
| (c
61
&
(
c
52
| pp
16
)
);
(20
)
A
3
A
2
A
1
A
0
X
B
3
B
2
B
1
B
0
pp
4
pp
3
pp
2
pp
1
pp
8
pp
7
pp
6
pp
5
pp
12
pp
11
pp
10
pp
9
pp
16
pp
15
pp
14
pp
13
c
32
c
31
c
2
c
42
c
41
c
52
c
51
c
62
c
61
c
71
P
8
P
7
P
6
P
5
P
4
P
3
P
2
P
1
Figure
2
.
Ca
rr
y
Pr
e
-
Com
pu
ta
ti
on Based
Mult
ipli
er
The
a
rch
it
ect
ur
e of the
4
-
bit
m
ul
ti
plier can be
unde
rstood
from
the
blo
c
k diag
ram
as
sh
own
in Fi
gure
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Desig
n of
modi
fi
ed
boot
h base
d mu
lt
ipli
er wi
th carry
pr
e
-
c
omp
uta
ti
on
(
Ch
aitan
y
a
CV
S
)
1051
Figure
3
.
A
rch
i
te
ct
ur
e
of
Ca
rry
Pr
e
-
C
om
pu
ta
ti
on
base
d
Mul
ti
plier
The
thir
d
sta
ge
in
the
bl
ock d
ia
gr
am
involve
s
the
u
se o
f
XOR
l
og
ic
f
or
th
e
pa
rtia
l
pro
du
ct
s
an
d
car
ry
gen
e
rated
in
ea
ch
c
olu
m
n.
Th
e
outp
ut
of
t
his
sta
ge
giv
es
the
fi
nal
16
-
bit
pr
oduct
w
hich
is
ob
ta
ine
d
in
a
pa
rall
el
m
echan
ism
instea
d
of
se
qu
entia
l
m
echan
i
s
m.
The
pro
duct
s
of
AL
*B
L,
A
H
*BL,
A
L*BH,
AH*B
H
are
determ
ined
us
ing
a
bove
4
-
bit
carry
pre
-
c
ompu
ta
ti
on
-
base
d
m
ulti
plier
and
the
res
ults
of
al
l
su
b
m
ulti
pli
ers
a
r
e
add
e
d
t
o deter
m
ine the f
i
nal
pro
du
ct
.
Th
e
b
l
ock of t
he 8
-
bit
m
ult
ipli
er
a
s s
how
n
in
F
ig
ure
4
.
Figure
4
.
Bl
oc
k Diag
ram
o
f
8
-
bit M
ulti
plier
Using
4
-
bit Ca
rr
y P
re
-
C
om
puta
ti
on
Base
d
M
ulti
plier
2.2.
Modifie
d
B
oot
h Ba
se
d Bina
r
y Multipli
er
w
ith
C
arr
y
Pre
-
Co
m
pu
tation
An
8
-
bit
Bi
na
r
y
Mo
dified
Bo
oth
ba
sed
m
ultip
li
er
with
car
ry
pre
-
c
om
pu
ta
ti
on
has
bee
n
pr
opos
e
d.
T
he
arch
it
ect
ure
of
the 8
-
bit m
ult
i
plier ca
n be
under
st
ood from
the b
l
ock d
ia
gra
m
as
show
n
i
n
Fi
gure
5
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
3
,
Ma
rc
h 201
9
:
1
048
–
1
055
1052
Figure
5
.
Mo
dified
B
oo
t
h
M
ul
ti
plier w
i
th
Ca
rr
y P
re
-
C
om
puta
ti
o
n
L
og
ic
The
m
od
ifie
d
booth
enc
odin
g
is
the
first
bl
ock
of
the
m
ul
ti
plier
to
w
hic
h
the
8
-
bit
m
ulti
plica
nd
a
nd
m
ul
ti
plier
are
gi
ven
as
in
puts.
The
m
od
ifie
d
boot
h
enc
oder
ge
ner
at
es
4
pa
rtia
l
pr
od
ucts
bas
ed
on
the
fo
ll
owin
g
T
able
1.
Co
ns
ide
rin
g
that
both
i
nputs
A
a
nd
B
a
re
of
8
bits.
As
in
puts
are
8
bits,
f
our
par
ti
al
pro
du
ct
s
will
be
gen
e
rated
as s
how
n
i
n
F
i
gure
6
.
Table
1
.
M
od
i
f
ie
d
bo
oth
e
n
c
odin
g
Xi+1
Xi
Xi
-
1
Actio
n
0
0
0
0
×
Y
0
0
1
1
×
Y
0
1
0
1
×
Y
0
1
1
2
×
Y
1
0
0
-
2
×
Y
1
0
1
-
1
×
Y
1
1
0
-
1
×
Y
1
1
1
0
×
Y
Figure
6
.
Parti
al
Products
of
Mod
ifie
d
B
oo
t
h
Mult
ipli
er
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
s1
s1
s1
s1
s1
s1
s1
p18
p17
p16
p15
p14
p13
p12
p11
p10
s2
s2
s2
s2
s2
p28
p27
p26
p25
p24
p23
p22
p21
p20
s3
s3
s3
p38
p37
p36
p35
p3
4
p33
p32
p31
p30
s4
p48
p47
p46
p45
p44
p43
p42
p41
p40
c
1
3
2
c
1
2
2
c
1
1
2
c
1
0
2
c
9
2
c
8
2
c
7
2
c
6
2
c
5
2
c
4
2
c
4
1
c3
c2
c
1
4
1
c
1
3
1
c
1
2
1
c
1
1
1
c
1
0
1
c
9
1
c
8
1
c
7
1
c
6
1
c
5
1
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Desig
n of
modi
fi
ed
boot
h base
d mu
lt
ipli
er wi
th carry
pr
e
-
c
omp
uta
ti
on
(
Ch
aitan
y
a
CV
S
)
1053
Partia
l pro
duct
–
1:
p10
t
o p18 a
nd s
1
is t
he
sign exte
ns
io
n
Partia
l pro
duct
–
2:
p20
t
o p28 a
nd s
2
is t
he
sign exte
ns
io
n
Partia
l pro
duct
–
3:
p30
t
o p38 a
nd s
3
is t
he
s
ign exte
ns
io
n
Partia
l pro
duct
–
4:
p40
t
o p48 a
nd s
4
is t
he
sign exte
ns
io
n
In
F
ig
ur
e
,
c2
is ca
rr
y
ge
ne
rated
from
co
lum
n
2
c3
is ca
rr
y
ge
ne
rated
from
co
lum
n
3
c41 an
d
c
42
is
carries
ge
ner
at
ed fr
om
co
lum
n 4
c51 an
d
c
52
is
carries
ge
ner
at
ed fr
om
column
5
c61 an
d
c
62
is
carries
ge
ner
at
ed fr
om
co
lum
n 6
c71 an
d
c
72
is
carries
ge
ner
at
ed fr
om
co
lum
n 7
c81 an
d
c
82
is
carries
ge
ner
at
ed fr
om
co
lum
n 8
c91 an
d
c
92
is
carries
ge
ner
at
ed fr
om
co
lum
n 9
c101
a
nd c
102
is carries
ge
nerat
ed
f
ro
m
co
lu
m
n
10
c111
a
nd
c
112
is carries
ge
nerat
ed
f
ro
m
co
lu
m
n
11
c121
a
nd c
122
is carries
ge
nerat
ed
f
ro
m
co
lu
m
n
12
c131
a
nd c
132
is carries
ge
nerat
ed
f
ro
m
co
lu
m
n
13
c141
is
car
ry ge
ner
at
e
d
f
r
om
colum
n
14
The
seco
nd
sta
ge
i
n
the
bl
ock
dia
gr
am
is
the
car
ry
ge
ner
at
ion
ci
rc
uit.
He
r
e,
we
ha
ve
inte
gr
at
e
d
pre
-
com
pu
ta
ti
on
l
og
ic
al
ong
with
the
m
od
ifie
d
boot
h
m
ulti
pl
ic
at
ion
te
ch
nique.
The
car
ry
values
are
ge
ne
rate
d
separ
at
el
y
f
or
each
col
um
n
of
pa
rtia
l
produ
ct
s
and
t
he
in
puts
f
or
th
ese
e
qu
at
io
ns
a
re
ta
ken
from
the
pr
evi
ous
colum
n.
T
he
ca
rr
y
pre
-
c
om
pu
ta
ti
on
lo
gic
disc
us
se
d
i
n
sect
io
n
2.1
is
us
e
d
to
gen
e
rate
t
he
va
lues.
T
he
t
hir
d
sta
ge
in
the
blo
c
k
dia
gr
am
involves
t
he
us
e
of
X
OR
log
ic
for
the
pa
rtia
l
pro
du
ct
s
a
nd
car
r
y
ge
ner
a
te
d
in
eac
h
c
ol
um
n.
The
outp
ut
of
this
sta
ge
gi
ve
s
the
final
16
-
bit
pro
duct
w
hi
ch
is
ob
ta
i
ned
in
a
pa
rall
el
m
echan
ism
instea
d
of
seq
uen
ti
al
m
ec
han
ism
.
I
n
pro
po
s
ed
m
ulti
pli
er,
a
s
Ca
rr
y
P
re
-
C
om
pu
ta
ti
on
unit
com
pu
t
es
al
l
the
car
ries
in
par
al
le
l
us
i
ng
carry
lo
ok
a
he
ad
lo
gic
a
nd
r
e
m
ov
e
dep
e
nd
encies
betwee
n
c
olu
m
ns
,
the
total
tim
e
require
d
to
gen
e
r
at
e the
pr
oduct
will
b
e
r
edu
ce
d.
3.
RESU
LT
S
A
ND AN
ALYSIS
The
pro
posed
arc
hitec
ture
m
od
el
ed
us
i
ng
Ve
rilog
H
DL
,
sim
ulate
d
usi
ng
Ca
de
nce
NCSI
M
a
nd
synthesiz
ed
us
i
ng
C
ade
nce
R
TL
Com
piler
with
65nm
TSMC
li
br
ary.
Di
ff
e
ren
t
im
ple
m
entat
ion
m
et
hodo
l
og
y
has
bee
n
ta
ke
n
an
d
im
ple
m
ented
in
sam
e
te
chnolo
gical
env
ir
onm
ent
and
t
hen
c
ompare
d
the
perf
or
m
ance
par
am
et
ers.
F
or
the
c
om
par
iso
n
point
of
view
the
idea
s
ha
ve
been
c
onsidere
d
from
the
re
fe
ren
ces
a
nd
sim
ulate
d
and
pe
rfo
rm
ance
pa
ram
et
ers
wer
e
com
pu
te
d
us
in
g
t
he
sam
e
MOSF
ET
te
c
hnol
og
y
file
.
I
nput
data
was
ta
ken
in
a
regular
fas
hio
n
f
or
e
xperim
ental
pur
pose.
The
delay
an
d
the
powe
r
m
easur
e
d
us
in
g
th
e
worst
-
cas
e
pa
tt
ern
and f
ro
m
the
outp
ut whe
re the
d
el
ay
is m
axim
u
m
.
It
is
obse
rv
e
d
t
hat
the
pro
pose
d
car
ry
pre
-
co
m
pu
ta
ti
on
-
ba
s
ed
m
ulti
plier
and
m
od
ifie
d
bo
oth
m
ulti
plier
with
c
ar
ry
pre
-
com
pu
ta
ti
on
ba
sed
offer
e
d
substanti
al
reducti
on
of
pro
pag
at
io
n
de
la
y
an
d
t
otal
po
w
e
r
consum
ption
s.
Fr
om
T
able
1
and
T
able
2,
it
can
be
obser
ve
d
that
the
pro
po
s
ed
ca
rr
y
pre
-
com
pu
ta
ti
on
bas
e
d
m
ul
ti
plier
design
offe
r
ed
~2
3%
,
~6
4%,
~5
7%
,
~8
3%
w
hen
com
par
ed
with
ar
ray
m
ulti
pl
ie
r,
wall
ace
m
ulti
plier,
colum
n
base
d
m
ul
ti
plier,
an
d
Nikhil
am
based
base
d
m
ultip
li
ers
r
espe
ct
ively
,
an
d
m
od
ifie
d
boot
h
m
ulti
plier
with
ca
rr
y
pre
-
com
pu
ta
ti
on
offe
re
d
~
83%,
~
92%,
~
91%,
~
97%
wh
e
n
c
om
par
ed
with
ar
ray
m
ulti
plier,
wall
ace
,
colum
n
base
d
,
and N
i
kh
il
am
b
ase
d
m
ulti
plier
s
r
especti
vely
.
Table
2
.
Su
m
m
ary o
f
sy
nth
esi
s r
es
ults o
f 8
-
bi
t
m
u
lt
ipli
er ar
chite
ct
ur
es
S.
No
Architectu
re
(
8
-
b
it
)
Delay
(ns
)
Dy
n
a
m
i
c
Po
wer
(u
W
)
Static Powe
r
(uW)
Total Po
wer
(uW)
Po
wer
-
D
elay
Prod
u
ct (
p
J)
1
Arr
a
y
Based
Multi
p
lier
[
9
]
1
.5
1
5
.09
6
2
1
.09
3
1
.63
2
W
allac
e Based
Multip
lier
[
2
]
1
.2
6
.27
4
9
.91
3
5
6
.18
4
6
7
.42
3
Co
lu
m
n
Based
Multip
lier
[
6
]
1
.95
2
6
.74
2
.8
2
9
.54
5
7
.6
4
Nik
h
ila
m
Based
M
u
ltip
lier
[
7
]
3
.2
4
2
.56
4
.3
4
6
.86
1
4
9
.95
5
Pre
-
Co
m
p
u
tatio
n
Bas
ed
M
u
ltip
lier
0
.75
2
5
.77
7
.45
3
3
.23
2
4
.23
6
Mod
if
ied
Bo
o
th
'
s Multip
lier
with
Carr
y
Pr
e
-
Co
m
p
u
t
atio
n
0
.45
9
.4
3
.99
1
3
.39
6
.02
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
3
,
Ma
rc
h 201
9
:
1
048
–
1
055
1054
Fr
om
the
T
able
2
a
nd
T
able
3,
it
can
be
obser
ved
t
hat
Mo
dif
ie
d
Bo
oth
base
d
m
ulti
plier
with
car
ry
pr
e
-
com
pu
ta
ti
on
c
on
s
um
es
le
ss
delay
w
he
n
c
om
par
ed
to
car
r
y
pre
-
c
om
pu
ta
ti
on
-
base
d
m
ulti
plier
with
the
delay
trade
off.
P
r
opose
d
Mo
difie
d
Boo
t
h
bas
ed
m
ulti
plier
with
c
arr
y
pre
-
com
pu
ta
ti
on
ga
ve
th
e
bette
r
power
-
delay
pro
du
ct
wh
e
n
com
par
ed
to
pro
posed
ca
rr
y
pre
-
com
pu
ta
ti
on
-
base
d
m
ulti
plier
and
e
xisti
ng
m
ulti
plier
f
ro
m
li
te
ratur
e.
Table
3
.
Su
m
m
ary o
f
sy
nth
esi
s r
es
ults o
f 1
6
-
bit m
ulti
plier a
rch
it
ect
ures
S.
No
Architectu
re
(
1
6
-
b
it)
Delay (ns
)
Dy
n
a
m
i
c Powe
r
(uW)
Static Powe
r
(uW)
Total Po
wer
(uW)
Po
wer
-
Delay
Prod
u
ct
1
Arr
a
y
Based
Multi
p
lier
[
9
]
2
.89
3
0
.18
12
4
2
.18
1
2
1
.90
2
W
allac
e Based
Multip
lier
[
2
]
2
.46
1
2
.54
9
9
.82
6
1
1
2
.366
2
7
6
.42
3
Co
lu
m
n
Based
Multip
lier
[
6
]
3
.82
5
2
.48
5
.4
5
7
.88
2
2
1
.10
4
Nik
h
ila
m
Based
M
u
ltip
lier
[
7
]
5
.96
8
0
.65
8
.1
8
8
.75
5
2
8
.95
5
Pre
-
Co
m
p
u
tatio
n
Bas
ed
M
u
ltip
lier
1
.4
5
1
.54
1
4
.9
6
6
.44
9
3
.01
6
Mod
if
ied
Bo
o
th
'
s Multip
lier
with
Carr
y
Pr
e
-
Co
m
p
u
t
atio
n
0
.84
1
7
.23
8
.3
2
5
.53
2
1
.44
4.
CONCL
US
I
O
N
In
this
pap
e
r,
a
Ve
dic
m
ath
em
atics
-
base
d
m
ulti
plier
ha
s
bee
n
pro
posed
w
hich
us
e
s
Ca
rr
y
pre
-
com
pu
ta
ti
on
a
nd
ope
rand
de
com
po
sit
ion
m
et
hodo
l
og
y.
T
he
pro
po
se
d
ar
chite
ct
ur
es
co
m
bin
e
the
be
ne
fits
of
Ved
ic
m
et
ho
d
and
pa
rall
el
pre
-
com
pu
ta
ti
on
of
ca
rr
ie
s
t
her
e
by
res
ulti
ng
i
n
reducti
on
of
powe
r
-
delay
pro
du
ct
.
The
pro
pag
at
i
on
delay
of
carry
pre
-
com
pu
ta
ti
on
-
base
d
m
ult
ipli
er
f
or
cal
culat
io
n
of
8
bit
a
nd
16
-
bit
m
ul
ti
plica
ti
on
was 0
.75
ns
a
nd
1.4n
s
w
hile
power
c
onsu
m
ption
was
33.23
uW
an
d
66.44
uW.
T
he p
r
op
a
ga
ti
on
delay
of
m
od
ifi
ed
boot
h
m
ultip
li
er
with
ca
rr
y
pre
-
c
om
pu
ta
ti
on
f
or
cal
culat
i
on
of
8
bit
a
nd
16
-
bit
m
ulti
pli
cat
ion
was
0.45ns
a
nd
0.84ns
wh
il
e
powe
r
c
on
s
um
pt
ion
was
13.39
uW
a
nd
25.53
uW.
T
he
de
la
y
of
m
ulti
pl
ic
at
io
n
was
dec
rease
d
by
~8
5%
a
nd
powe
r
c
onsu
m
p
ti
on
wer
e
re
duced
by
~8
8%
w
he
n
c
om
par
ed
t
o
Nikhil
am
base
d
Ved
ic
m
ulti
plier.
REFERE
NCE
S
[1]
Xiangui
Kang
,
Anjie
Peng,
Xian
y
uXu,
Xiao
chun
Cao,
"
Perform
ing
Scalable L
oss
y
Com
pre
ss
ion
o
n
Pixel E
nc
r
y
pted
Im
age
s,
"
EURASIP
Journal
on
I
mage
and
V
ide
o
Proce
ss
ing
,
pp
.
1
-
6
,
2013
.
[2]
Nikolay
Ponom
a
ren
ko,
Serge
y
Kr
ive
nko,
Vlad
imir
Luki
n
,
Kare
n
E
gia
z
ari
an
,
Jaa
kk
o
T
.
,
As
tol
a
,
"
Lo
ss
y
Com
pre
ss
ion
of
Nois
y
Im
ages
Based
on
Visual
Qual
ity
:
A
C
om
pre
hensive
St
ud
y
,
"
EURA
SI
P
Journal
on
A
dvanc
es
in
S
ign
al
Proce
ss
ing
,
pp
.
1
-
13
,
2010
.
[3]
L.
-
K.
W
ang,
M
.
A.
Erl
e
,
C.
Tse
n,
E.
M.
Schw
ar
z,
and
M.
J
.
Sch
ult
e
,
"
A
surve
y
of
ha
rdware
d
esigns
for
de
ci
m
al
ari
thmet
ic
,
"
IBM
Journal
o
f Re
se
arch
and
De
ve
lo
pment
,
vo
l.
54(2
),
pp
.
8:1
-
8
:15
,
2
010
.
[4]
M.
Jee
v
it
ha
,
R.
Muthaiah,
P.
Sw
aminat
han,
"
Eff
icient
Mult
ip
li
er
Archi
tectur
e
in
VLSI
Des
ign,
"
Journal
o
f
Theoreti
cal and Appli
ed
Informa
ti
on
Te
chnol
ogy
,
vol.
38(2), pp. 1
96
-
201.
2
,
2012
.
[5]
J.
R.
Boddi
e,
G.
T.
Dar
y
an
ani
,
I
.
I
.
El
dum
ia
t
i,
R.
N
,
Gad
enz,
J.
S.
T
hom
pson,
S.
M.
W
al
te
rs,
"
Digital
Signal
Proc
essor:
Archi
tect
ure
and
Perform
anc
e
,
"
Be
ll Sy
st
em
Tec
hnic
al
Journal
,
vol.
60(7)
,
pp
.
1
449
-
1462
,
1981
.
[6]
Bhara
t
iKrsnaT
ir
tha
ji
,
V
.
S Agra
wala
,
"
Vedic Ma
the
m
at
i
cs
"
,
13th
Edi
ti
on
,
Mo
ti
la
l Banars
idass
,
2010.
[7]
P.
Saha,
A.
Ban
erj
e
e,
A.
Danda
pat
,
and
P.
Bh
attac
h
ar
yy
a
,
"
AS
IC
design
of
a
h
igh
-
spee
d
low
p
ower
ci
r
cui
t
fo
r
fac
tor
ia
l
calc
u
la
t
ion
using
an
ci
en
t
Vedic
m
a
the
m
atics,
"
ELSEV
IE
R
Mic
roel
ectronic
s
Jour
nal
,
vol
.
4
2
(
12
)
,
pp.
1343
-
1352,
De
c
2011
.
[8]
MD
.
Bel
a
l
Rash
id,
B
al
a
ji
B
.
S
an
d
Prof.
M.B
.
An
anda
ra
ju,
"
VLSI
Design
and
Im
ple
m
ent
a
ti
on
of
B
ina
r
y
Mul
ti
plier
base
d
on
Urdhva
Ti
r
y
agbh
y
am
Su
tra
with
r
educed
Delay
and
Are
a
,
"
Int
ernati
onal
J
ournal
of
Engi
ne
ering
Re
search
and
Technol
og
y
,
vol. 6
(
2
)
,
pp.
26
9
-
278,
Mar
2013
.
[9]
Ko
-
Chi
Kuo,
C
hi
-
W
en
Chou,
"
Low
Pow
er
and
High
-
Speed
Multi
pl
ie
r
Design
with
Row
B
y
p
a
ss
ing
and
Par
all
el
Archi
tectur
e
,
"
M
ic
roel
ec
troni
cs
J
ournal
,
vol
.
41
,
pp.
639
-
650
,
20
10
.
[10]
Constant
inos E
fstat
hiou
,
N.
Mos
hopolous,
N.
Axe
los,
K.
Pekm
estzi,
Eff
i
cient
Modulo
2n
+1
Mul
ti
p
l
y
and
Mul
ti
p
l
y
-
Add Unit
s Ba
se
d
on
Modifi
ed
B
ooth
En
codi
ng,
I
nte
gra
ti
on,
th
e
V
LSI
Journal, 47 (
2014),
pp
.
140
-
1
47.
[11]
Mana
s
Ranjan
Mehe
r,
Ch
ing
C
huen
Jong,
and
Chip
-
Hong
Cha
ng,
“
A
High
Bit
Rat
e
Ser
ial
-
Serial
Multi
p
li
er
with
On
-
the
-
Fl
y
Ac
cumulat
ion
b
y
As
y
nc
hronous
Counte
r
s”,
I
EE
E
tr
ans.
On
VLSI
s
y
st
ems
,
Vol.
19,
No.
1
0,
pp
.
1733
-
174
5,
Octobe
r, 2011.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Desig
n of
modi
fi
ed
boot
h base
d mu
lt
ipli
er wi
th carry
pr
e
-
c
omp
uta
ti
on
(
Ch
aitan
y
a
CV
S
)
1055
BIOGR
AP
HI
ES OF
A
UTH
ORS
Chai
ta
n
y
a
CVS
recei
v
ed
h
i
s
Bac
he
lor
Degr
e
e
in
E
le
c
tronic
s
and
Com
m
unic
a
ti
on
Engi
ne
eri
ng
in
2006
from
JN
TU,
H
y
d
era
bad
a
nd
his
MS
degr
ee
in
VLSI
-
CAD
from
Manipa
l
Univer
s
ity
in
2
0
007.
In
2010
,
he
start
ed
h
is
caree
r
as
As
sistant
Profess
or
in
School
of
Inform
at
ion
S
ci
en
ce
s
,
Manipa
l
.
Cur
ren
tly
,
h
e is
doing P
h.
D
at
Man
ipa
l
Univer
sit
y
.
His
r
ese
arc
h
intere
st
i
ncl
udes
High
Perform
anc
e
Com
pute
r
Arithmeti
c
,
Advan
ce
d
Com
pute
r
Archi
tectur
e
,
Lo
w
-
power
VLSI Design,
Elec
tron
ic
D
esign A
utomati
on,
and
Para
l
le
l
Algori
t
hm
s/Archit
ec
tur
es.
Dr.
C
Sundar
esa
n
completed
B
a
che
lor
degr
ee
in
Elec
tron
ic
s
and
Com
m
unic
at
ion
in
2000
from
Madura
i
Kam
ara
j
Univ
ersity
and
MS
degr
e
e
in
VLSI
CAD
in
2003
from
Manipa
l
Univer
sit
y
and
PhD
in
2018
from
Manipa
l
Aca
dem
y
of
High
er
Educ
a
ti
on.
He
st
art
ed
h
is
ca
re
er
as
R
&
D
engi
ne
er
a
t
Apl
ab
Lt
d
.
Curre
n
tly
h
e
is
working
as
As
sistant
Profess
or
in
School
Of
Inform
at
ion
Scie
n
ce
s.
His
rese
arc
h
in
t
ere
sts
include
C
om
pute
r
Arithmetic,
Low
-
Pow
er
VLSI De
sign,
Log
ic S
y
n
t
hesis,
Sta
ti
c
T
iming Ana
l
y
sis.
Dr.
P.
R.
Venka
t
eswara
n
obt
ai
ne
d
his
bac
h
el
or’s
degr
ee
in
El
e
ct
ro
nic
s
and
Instrum
ent
a
ti
on
Engi
ne
eri
ng
fro
m
Nati
onal
En
gine
er
ing
Col
lege,
Kovilpatti
i
n
1998
and
M
aste
rs
in
Instrum
ent
at
ion
and
Control
En
gine
er
ing
from
Te
chn
ic
a
l
T
ea
c
her
s’
Tra
in
ing
I
nstit
ute,
Chandi
gar
h
in
2
002.
He
completed
h
is doctoral r
ese
arc
h
in
2008
from
Manipa
l
U
nive
rsi
t
y
,
Manipa
l
.
He
st
ar
te
d
his
ca
re
er
as
t
ea
ch
ing
fa
cul
t
y
a
t
Sethu
Insti
tut
e
of
Te
chno
log
y
,
Madura
i
and
cont
inu
ed
h
i
s
teac
hing
c
areer
with
T
ec
hni
ca
l
Te
a
che
rs’
Traini
ng
Instit
u
te,
Ch
a
ndiga
rh
and
l
at
e
r
a
t
Man
ipa
l
Instit
u
te
of
Te
chno
log
y
,
Ma
nipa
l
.
Present
l
y
,
he
is
working
a
s
Senior
Engi
ne
er
(Cont
rol
and
Instru
m
ent
at
ion)
at
W
el
ding
Rese
a
rch
Instit
u
te,
BHEL,
Ti
ruc
h
ira
ppa
ll
i
a
nd
is
associ
at
ed
in
t
he
ar
ea
s
of
W
el
ding
Autom
at
ion
and
W
el
d
i
ng
Pow
er
Source
s.
Hi
s
areas
of
int
er
est
are
li
ne
ar
Contro
l
th
eor
y
,
Elec
tron
ic
Instrum
ent
at
ion
and
Soft
Com
puti
ng
Tec
hnique
s.
He
h
as
bee
n
a
r
evi
ewe
r
for
journals
like
IE
EE
SM
C,
El
sevi
er
,
AM
SE
et
c. He
is
a
m
ember
o
f
pr
ofe
ss
iona
l
bod
ies
of
ISTE
,
IW
S
and
IE
.
Dr.
Ke
ert
han
a
Prasad
is
worki
ng
as
Profess
or
in
School
of
I
nform
at
ion
Sc
ience
s,
a
consti
tue
n
t
insti
t
uti
on
o
f
Man
ipa
l
Univer
si
t
y
.
Her
rese
arc
h
int
er
est
s a
re
imag
e anal
y
sis
and
it
s
applic
at
ions
in
m
edi
c
ine
and
high
-
p
erf
or
m
anc
e
computi
ng
appr
oa
ch
fo
r
image
proc
essing.
Evaluation Warning : The document was created with Spire.PDF for Python.