Hig
hH
igh --
Leve
l Dec
isio
n D
iagr
ams
base
d Le
vel D
ecis
ion
Dia
gram
s ba
sed
Ver
ifica
tion
with
PS
L A
sser
tions
Ver
ifica
tion
with
PS
L A
sser
tions
Mak
sim
JE
NIH
HIN
Jaan
RA
IK
Ale
xand
er S
UD
NIT
SÕ
N
(( Pre
viou
s:P
revi
ous:
Aut
omat
ed R
easo
ning
for H
ardw
are
Ver
ifica
tion
Aut
omat
ed R
easo
ning
for H
ardw
are
Ver
ifica
tion ,
A, A.. S
udni
tsS
udni
tsõõ nn
))
2H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007Out
line
Out
line
Intro
duct
ion
Form
al v
erifi
catio
nS
imul
atio
n-ba
sed
verif
icat
ion
Ass
ertio
n ba
sed
verif
icat
ion
(PS
L, th
e pa
per)
3H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007Es
toni
aEs
toni
a
~ 45
,000
km
2 ~
1,30
0,00
0in
habi
tant
sca
pita
l: Ta
llinn
(~40
0,00
0)
Sum
mer
: av
r: 20
ºC /
70ºF
m
ax: (
35ºC
/ 95
ºF)
day
~ 18
hou
rs
Win
ter:
av
r: -1
0ºC
/ 15
ºF
min
: (-3
5ºC
/ -3
0ºF)
day
~ 6
hour
s
4H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Talli
nn U
nive
rsity
of T
echn
olog
yTa
llinn
Uni
vers
ity o
f Tec
hnol
ogy
The
only
tech
nica
l uni
vers
ity in
E
ston
ia12
,000
stu
dent
s25
% a
re a
t IT
facu
ltyIT
Fac
ulty
con
sist
s of
6
Dep
artm
ents
:»
Com
pute
r Eng
inee
ring
»C
ompu
ter S
cien
ce»
Ele
ctro
nics
»C
ompu
ter C
ontro
l»
Info
rmat
ics
and
Rad
io»
Com
mun
icat
ion
Eng
inee
ring
5H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Dep
t. of
Com
pute
r Eng
inee
ring
Dep
t. of
Com
pute
r Eng
inee
ring
32 e
mpl
oyee
s:5
prof
esso
rs, 6
ass
ocia
te
prof
esso
rs, 1
0 re
sear
cher
s,10
PhD
stu
dent
sP
artic
ipat
ion
in R
&D
pro
ject
s:»
8 E
U p
roje
cts
sinc
e 19
93»
Bila
tera
l res
earc
h pr
ojec
ts w
ith S
wed
en,
Pol
and,
Ger
man
y
Aro
und
80 p
aper
s pu
blis
hed
annu
ally
Res
earc
h to
pics
:»
Har
dwar
e de
sign
(SoC
, NoC
, FP
GA
)»
Har
dwar
e ve
rific
atio
n (s
tatic
/ dy
nam
ic)
»H
ardw
are
test
ing
(BIS
T, B
ound
ary
Sca
n,et
c)
6H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Verif
icat
ion
is c
heck
ing
if th
e ci
rcui
t was
des
igne
dco
rrec
tly
Valid
atio
nis
sim
ilar t
o ve
rific
atio
n bu
t it i
s pe
rform
ed o
n ph
ysic
al p
roto
type
Test
ing
is c
heck
ing
ever
ym
anuf
actu
red
circ
uit f
or
its c
orre
ctne
ss (a
bsen
ce o
f man
ufac
turin
g de
fect
s)
Wha
t is
verif
icat
ion?
Wha
t is
verif
icat
ion?
7H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Dig
ital c
ircui
ts (i
.eA
SIC
, SoC
)
»no
t sof
twar
e!
»no
t ana
log,
RF,
mix
ed-s
igna
l!
Func
tiona
l ver
ifica
tion
of a
mod
el (i
.e. .
vhdl
)
The
scop
eTh
e sc
ope
8H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Ver
ifica
tion
take
s ro
ughl
y 70
-85%
of d
esig
n co
sts
Som
e co
mpa
nies
hav
e 2-
4 ve
rific
atio
n en
gine
ers
for e
very
des
ign
engi
neer
A n
eed
to in
crea
se v
erifi
catio
n ef
fect
iven
ess
»D
esig
n-fo
r-V
erifi
abili
ty (D
FV)
Ass
ertio
n-ba
sed
Ver
ifica
tion
(AB
V)
Inte
rnat
iona
l Tec
hnol
ogy
Roa
dmap
for S
emic
ondu
ctor
s re
port
http
://w
ww
.itrs
.net
Mot
ivat
ion
Mot
ivat
ion
9H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Intr
oduc
tion
to h
ardw
are
verif
icat
ion
1 In
trod
uctio
n to
har
dwar
e ve
rific
atio
n 1
10H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Intr
oduc
tion
to h
ardw
are
verif
icat
ion
2In
trod
uctio
n to
har
dwar
e ve
rific
atio
n 2
11H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Form
al v
s. S
imul
atio
nFo
rmal
vs.
Sim
ulat
ion --
base
d ve
rific
atio
n ba
sed
verif
icat
ion
Form
al v
erifi
catio
n (s
tatic
)»
Inte
lligen
t (m
athe
mat
ical
) pro
of o
f cor
rect
ness
»C
onst
rain
ed a
pplic
atio
n
Sim
ulat
ion-
base
d (d
ynam
ic)
»S
imul
atio
n of
inpu
t vec
tors
(ran
dom
or d
eter
min
istic
)»
The
mos
t com
mon
ly u
sed
12H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Dec
isio
n D
iagr
ams
cano
nica
l for
mTh
e id
ea:
»C
onst
ruct
DD
for t
he tw
o ci
rcui
ts to
be
com
pare
d»
Man
ipul
ate
them
to p
rove
that
they
are
equ
ival
ent
Form
al v
erifi
catio
n: E
quiv
alen
ce c
heck
ing
1Fo
rmal
ver
ifica
tion:
Equ
ival
ence
che
ckin
g 1
13H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Form
al v
erifi
catio
n: E
quiv
alen
ce c
heck
ing
2Fo
rmal
ver
ifica
tion:
Equ
ival
ence
che
ckin
g 2
d =
f ⊕g
»A
ssig
n th
e va
riabl
es o
f boo
lean
form
ula
to e
valu
ate
it to
TR
UE
(sat
isfia
ble,
fan
d g
are
not e
quiv
alen
t)»
OR
pro
ve th
at it
eva
luat
es to
FA
LSE
for a
ll po
ssib
le
assi
gnm
ents
(not
sat
isfia
ble,
fan
d g
are
equi
vale
nt)
⊕1
spec
ifica
tion
( f )
impl
emen
tatio
n ( g
)⊕
14H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
SAT
SAT
SA
T(B
oole
an s
atis
fiabi
lity
prob
lem
)2-
SA
T is
sol
vabl
e in
pol
ynom
ial t
ime
3-S
AT
is N
P-c
ompl
ete
»n-
SA
T ca
n be
redu
ced
to 3
-SA
T in
pol
ynom
ial t
ime
Is f(
a,b,
c)so
lvab
le?
( =1
)Th
e so
lutio
n is
: »
a =
1, b
= 0
, c =
0Th
e w
orst
cas
e is
to tr
y 2n
optio
ns
15H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Mod
el c
heck
ing
prov
es o
r dis
prov
es th
at a
pro
perty
(p
art o
f spe
cific
atio
n) h
olds
for t
he c
ircui
tE
xhau
stiv
ely
sear
ches
the
entir
e st
ate
spac
e »
In re
al li
ve th
e sp
ace
can
be c
onst
rain
ed
Typi
cally
the
prop
ertie
s ar
e de
scrib
ed in
CTL
(C
ompu
tatio
nal T
ree
Logi
c)»
[AG(P→((EX.Q)^(EX¬Q)))]
Form
al v
erifi
catio
n: M
odel
che
ckin
gFo
rmal
ver
ifica
tion:
Mod
el c
heck
ing
16H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Sim
ulat
ion
Sim
ulat
ion --
base
d ve
rific
atio
nba
sed
verif
icat
ion
Cov
erag
e m
etric
sU
sual
ly it
is n
ot fe
asib
le to
sim
ulat
e al
l pos
sibl
e in
put
com
bina
tions
It is
nec
essa
ry to
mea
sure
how
muc
h fu
nctio
nalit
y gi
ven
stim
uli (
inpu
t dat
a) c
over
s3
type
s of
cov
erag
e m
etric
s in
har
dwar
e ve
rific
atio
n:»
Cod
e co
vera
ge»
(nex
t slid
e)»
Par
amet
er c
over
age
»D
epen
ds o
n im
plem
enta
tion,
use
d fo
r par
amet
ers
»Fu
nctio
nal c
over
age
»D
epen
ds n
ot o
n im
plem
enta
tion
but o
n sp
ecifi
catio
ns»
Diff
icul
t to
mea
sure
17H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Cod
e co
vera
ge m
etric
Cod
e co
vera
ge m
etric
How
goo
d co
de e
ntiti
es a
re s
timul
ated
by
sim
ulat
ions
Dep
ends
on
impl
emen
tatio
n»
It is
pos
sibl
e to
hav
e 10
0% c
ode
cove
rage
on
com
plet
ely
wro
ng im
plem
enta
tion
Eas
y to
cal
cula
te
Sta
tem
ent c
over
age
Blo
ck c
over
age
Pat
h co
vera
ge
18H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Ass
ertio
nA
sser
tion --
base
d Ve
rific
atio
nba
sed
Verif
icat
ion
AB
V b
enef
its:
»D
ynam
ic–
bette
r obs
erva
bilit
y
dete
ctin
g bu
gs e
arlie
r and
clo
ser t
o th
eir o
rigin
»St
atic
–be
tter c
ontro
llabi
lity
dire
ct v
erifi
catio
n to
the
area
of i
nter
est
19H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Ass
ertio
nA
sser
tion --
base
d Ve
rific
atio
nba
sed
Verif
icat
ion
Com
plet
enes
s pr
oble
m»
Who
/wha
t and
whe
n sh
ould
spe
cify
ass
ertio
ns?
»W
hen
is it
eno
ugh?
In p
ract
ice
desi
gn e
ngin
eer w
rites
them
for V
HS
(V
erifi
catio
n H
ot S
pots
). S
uch
spot
:»
cont
ains
a g
reat
num
ber o
f seq
uent
ial s
tate
s;
»de
eply
hid
den
in th
e de
sign
, mak
ing
it di
fficu
lt to
con
trol f
rom
the
inpu
ts
»ha
s m
any
inte
ract
ions
with
oth
er s
tate
mac
hine
s an
d ex
tern
al a
gent
s
»ha
s a
com
bina
tion
of th
ese
prop
ertie
s
20H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Que
staS
IMQ
uest
aSIM
from
fr
om M
ento
rGra
phic
sM
ento
rGra
phic
s
21H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
PS
L =
Pro
perty
Spe
cific
atio
n La
ngua
ge»
Bas
ed o
n IB
M’s
Sug
ar, d
evel
oped
by
Acc
elle
ra»
IEE
E 1
850
Sta
ndar
d in
200
5Fl
avor
s:V
HD
L, V
erilo
g, S
yste
mV
erilo
g, G
DL,
Sys
tem
C4
laye
rs:
Boo
lean
laye
r–bo
olea
n ex
pres
sion
s in
HLD
: (a&
&(b
||c))
Tem
pora
l lat
er–
sequ
ence
s of
boo
lean
exp
ress
ions
ove
r m
ultip
le c
lock
cyc
les,
sup
ports
SE
RE
: ({A
[*3]
;B}|-
>{C
})V
erifi
catio
n la
yer-
dire
ctiv
es fo
r ver
ifica
tion
tool
telli
ng w
hat
to d
o w
ith s
peci
fied
prop
ertie
sM
odel
ling
laye
r–m
odel
s en
viro
nmen
t
PSL
PSL
22H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
PSL
(con
t.)PS
L (c
ont.)
23H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Pro
pose
d an
d de
velo
ped
in T
UT
HLD
Ds
are
prov
ed to
spe
ed-u
p si
mul
atio
n»
By
up to
fact
or 1
0 co
mpa
red
to c
omm
erci
al s
imul
ator
s
Hig
hH
igh --
Leve
l Dec
isio
n D
iagr
ams
(HLD
D)
Leve
l Dec
isio
n D
iagr
ams
(HLD
D)
+ …
reg1
'1'
=0
=1
mux1_addr
reg2_ena reg2
in1 re
g2_e
nare
g2m
ux1_
addr
reg2
reg1
+'1'
in1
0
0 1
1
24H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
VER
TIG
OVE
RTI
GO
Eur
opea
n C
omm
issi
on
6thFr
amew
ork
Pro
gram
me
Res
earc
h P
roje
ct»
ST
Mic
roel
ectro
nics
(coo
rdin
ator
)
»A
erie
logi
c an
d Tr
ansE
DA
»U
nive
rsiti
es fo
rm
Talli
nn (E
ston
ia)
Link
öpin
g(S
wed
en)
Sou
tham
pton
(UK
)
Ver
ona
(Ital
y)
2006
-20
08
25H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Wor
kW
ork --
inin-- p
rogr
ess
rese
arch
at T
UT
prog
ress
rese
arch
at T
UT
HLD
D-b
ased
ver
ifica
tion
flow
TLM
(Sys
tem
C)
RTL
(VH
DL)
PSL
Inte
rface
Val
idat
ion
patte
rns
Stim
uli &
Te
stbe
nch
(VH
DL)
Des
ign
mod
el
Pro
perti
es, a
sser
tions
PSL
(sim
ple
subs
et)
HLD
D In
terfa
ce
Sys
tem
re
pres
enta
tion
HLD
DM
odel
Pro
perti
es,
asse
rtion
s
HLD
D
Prop
ertie
s
HLD
D S
imul
atio
n
HLD
D C
over
age
Ana
lysi
s &
Test
Gen
erat
ion
Dyn
amic
Ver
ifica
tion:
A
sser
tion
Cov
erag
e
Pro
perty
Che
cker
TLM
(Sys
tem
C)
RTL
(VH
DL)
PSL
Inte
rface
Val
idat
ion
patte
rns
Stim
uli &
Te
stbe
nch
(VH
DL)
Des
ign
mod
el
Pro
perti
es, a
sser
tions
PSL
(sim
ple
subs
et)
HLD
D In
terfa
ce
Sys
tem
re
pres
enta
tion
HLD
DM
odel
Pro
perti
es,
asse
rtion
s
HLD
D
Prop
ertie
s
HLD
D S
imul
atio
n
HLD
D C
over
age
Ana
lysi
s &
Test
Gen
erat
ion
Dyn
amic
Ver
ifica
tion:
A
sser
tion
Cov
erag
e
Pro
perty
Che
cker
26H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Ass
ertio
ns a
utom
atic
tran
slat
ion
is a
com
plex
pr
oces
s em
ploy
ing:
»N
onde
term
inis
tic F
inite
Aut
omat
on
»D
eter
min
istic
Fin
ite A
utom
aton
Onl
y su
bset
of P
SL
prop
ertie
s is
tran
slat
able
An
optio
n to
TU
T in
tern
al s
olut
ion
is e
xter
nal
trans
lato
r lik
e Fo
Cs
form
IBM
PSL
inte
grat
ion
to H
LDD
PSL
inte
grat
ion
to H
LDD
27H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
PROC
ESS
(clk)
BEGI
NIF
( ( c
lk=
'1' )
) THE
Nfo
cs_o
k<=
( foc
s_vo
ut(4
) OR
NOT(
c ) )
;EL
SE focs
_ok<
= '1'
;EN
D IF
;EN
D PR
OCES
S;PR
OCES
S... VA
RIAB
LE fo
cs_v
out:
std_
logi
c_ve
ctor
(4 D
OWNT
O 0)
;BE
GIN
WAI
T UN
TIL
(clk'
EVEN
TAN
D clk
= '1'
);... fo
cs_v
out(4
DOW
NTO
0) :=
reve
rse(
( ( (
( ( (
focs
_v(0
) AND
a ) )
& (
( foc
s_v(
1) A
ND '1
' )) )
& (
( foc
s_v(
2) A
ND '1
' ) ) )
& (
(fo
cs_v
(3) A
ND b
) ) )
& (
( foc
s_v(
4) A
NDNO
T( c
) ) )
) );
...EN
D PR
OCES
S;
Tran
slat
ion
usin
g Fo
Cs
from
IBM
Tran
slat
ion
usin
g Fo
Cs
from
IBM
FoC
s .v
hdlc
heck
er
rese
t
0
a 1’
a 0
1 re
set
0
a 2’
a 1
0
rese
t
0
a 3’
a 2
0
re
set
0
a 4’
a 3
0
rese
t
0
b 1’
b 0
a 4
b 1
p 1
0
c 0
0 1 0 1 1
1 1 1
1
HLD
D c
heck
er
p: assert always ({a; [*2] ;b} |=> {c});
28H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Con
clus
ions
Con
clus
ions
This
pre
sent
atio
n ha
s gi
ven
a br
ief o
verv
iew
of
hard
war
e ve
rific
atio
nW
ork-
in-p
rogr
ess
of P
h.D
. res
earc
h w
as
pres
ente
d»
PS
L as
serti
ons
are
used
as
HLD
D s
imul
atio
n ch
ecke
rs
in v
erifi
catio
n flo
w
29H
igh
-Lev
el D
ecis
ion
Dia
gram
s ba
sed
Ver
ific
atio
n w
ith
PSL
Ass
erti
ons,
Cyp
rus,
Ju
ly 2
-6, 2
007
Bib
liogr
aphy
and
con
tact
Bib
liogr
aphy
and
con
tact
Boo
k:H
ardw
are
Des
ign
Verif
icat
ion:
Sim
ulat
ion
and
Form
al M
etho
d-B
ased
App
roac
hes,
Will
iam
K. L
am,
Sun
Mic
rosy
stem
s 20
05P
aper
s:H
igh-
Leve
l Dec
isio
n D
iagr
ams
(HLD
D) a
nd D
EC
IDE
Rby
Jaa
n R
aik
and
Prof
.Rai
mun
d U
barf
rom
IE
EE
xplo
reC
onta
ct:
pape
rs o
n P
SL
and
asse
rtion
sM
aksi
m J
enih
hin
–m
aksi
m@
pld.
ttu.e
eTa
llinn
Uni
vers
ity o
f Tec
hnol
ogy,
ES
TON
IA