I
nte
rna
t
io
na
l J
o
urna
l o
f
Rec
o
nfig
ura
ble a
nd
E
m
bedd
e
d Sy
s
t
em
s
(
I
J
RE
S)
Vo
l.
11
,
No
.
1
, M
a
r
ch
20
22
,
p
p
.
93
~
1
02
I
SS
N:
2089
-
4
8
6
4
,
DOI
:
1
0
.
1
1
5
9
1
/ijre
s
.
v
11
.
i
1
.
pp
93
-
1
02
93
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
r
es.ia
esco
r
e.
co
m
An ef
ficien
t
mul
ti
-
lev
el cache
sy
ste
m for geo
metri
ca
lly
interconne
cted
m
a
ny
-
co
re c
hip m
u
ltiproces
so
r
T
irum
a
le
Ra
m
esh
,
K
ha
lid
Abed
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
& C
o
m
p
u
t
e
r
En
g
i
n
e
e
r
i
n
g
a
n
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
J
a
c
k
so
n
S
t
a
t
e
U
n
i
v
e
r
s
i
t
y
,
M
i
ssi
ss
i
p
p
i
,
U
S
A
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
l 3
0
,
2
0
2
1
R
ev
is
ed
Sep
11
,
2
0
2
1
Acc
ep
ted
J
an
10
,
2
0
2
2
M
a
n
y
-
c
o
re
c
h
ip
m
u
lt
ip
r
o
c
e
ss
o
r
o
ffe
rs
h
ig
h
p
a
ra
ll
e
l
p
r
o
c
e
ss
in
g
p
o
we
r
fo
r
b
i
g
d
a
ta
a
n
a
ly
t
ics
;
h
o
we
v
e
r,
th
e
y
re
q
u
i
re
e
fficie
n
t
m
u
lt
i
-
le
v
e
l
c
a
c
h
e
a
n
d
in
terc
o
n
n
e
c
ti
o
n
to
a
c
h
ie
v
e
h
ig
h
s
y
ste
m
th
ro
u
g
h
p
u
t.
Us
in
g
o
n
-
c
h
i
p
first
lev
e
l
L1
a
n
d
se
c
o
n
d
lev
e
l
L
2
p
e
r
c
o
re
fa
st
p
ri
v
a
te
c
a
c
h
e
s
is
e
x
p
e
n
siv
e
fo
r
larg
e
n
u
m
b
e
r
o
f
c
o
re
s.
In
th
is
p
a
p
e
r,
fo
r
m
o
d
e
ra
te
n
u
m
b
e
r
o
f
c
o
re
s
fr
o
m
1
6
t
o
6
4
,
we
p
re
se
n
t
a
c
o
st
a
n
d
p
e
rf
o
rm
a
n
c
e
e
fficie
n
t
m
u
lt
i
-
lev
e
l
c
a
c
h
e
sy
ste
m
with
p
e
r
c
o
re
L1
a
n
d
las
t
lev
e
l
s
h
a
re
d
b
u
s
c
a
c
h
e
o
n
e
a
c
h
b
u
s
li
n
e
o
f
a
c
o
st
-
e
fficie
n
t
g
e
o
m
e
tri
c
a
ll
y
b
u
s
-
b
a
se
d
i
n
terc
o
n
n
e
c
ti
o
n
.
I
n
o
u
r
a
p
p
r
o
a
c
h
,
we
e
x
trac
ted
c
a
c
h
e
h
i
t
a
n
d
m
iss
c
o
n
c
u
rre
n
c
ies
a
n
d
a
p
p
li
e
d
c
o
n
c
u
rre
n
t
a
v
e
ra
g
e
m
e
m
o
ry
a
c
c
e
s
s
ti
m
e
to
m
o
re
a
c
c
u
ra
tely
d
e
term
in
e
th
e
c
a
c
h
e
sy
ste
m
p
e
rfo
rm
a
n
c
e
.
We
c
o
n
d
u
c
te
d
lea
st
re
c
e
n
tl
y
u
se
d
c
a
c
h
e
p
o
l
icy
-
b
a
se
d
sim
u
latio
n
fo
r
c
a
c
h
e
sy
ste
m
wit
h
L
1
,
with
L
1
/L
2
,
a
n
d
with
L
1
/
sh
a
re
d
b
u
s
c
a
c
h
e
.
Ou
r
sim
u
latio
n
re
su
lt
s
s
h
o
w
t
h
a
t
a
n
a
v
e
ra
g
e
s
y
ste
m
th
ro
u
g
h
p
u
t
imp
ro
v
e
m
e
n
t
o
f
2
.
5
x
c
a
n
b
e
a
c
h
iev
e
d
b
y
u
sin
g
sy
ste
m
with
L
1
/s
h
a
re
d
b
u
s
c
a
c
h
e
sy
ste
m
c
o
m
p
a
re
d
t
o
u
sin
g
o
n
ly
first
lev
e
l
L
1
o
r
L1
/
L2
.
F
u
rth
e
r,
we
sh
o
w
th
a
t
t
h
e
th
ro
u
g
h
p
u
t
d
e
g
ra
d
a
ti
o
n
f
o
r
th
e
p
ro
p
o
se
d
c
a
c
h
e
sy
ste
m
is
o
n
ly
with
in
5
%
fo
r
a
sin
g
le b
u
s fau
lt
,
s
u
g
g
e
sti
n
g
a
g
o
o
d
b
u
s fau
lt
to
lera
n
c
e
.
K
ey
w
o
r
d
s
:
B
ig
d
ata
B
u
s
ca
ch
e
Geo
m
etr
ical
Hete
r
o
g
en
eo
u
s
Ma
n
y
-
co
r
e
T
h
r
o
u
g
h
p
u
t
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Kh
alid
Ab
ed
Dep
ar
tm
en
t o
f
E
lectr
ical
&
C
o
m
p
u
ter
E
n
g
in
ee
r
in
g
a
n
d
C
o
m
p
u
ter
Scien
ce
,
J
ac
k
s
o
n
State
Un
iv
er
s
ity
1
4
0
0
J
o
h
n
R
L
y
n
ch
Stre
et
(
J
SU Bo
x
1
7
0
9
8
)
,
J
ac
k
s
o
n
,
MS.
3
9
2
1
7
,
USA
E
m
ail:
k
h
alid
.
h
.
a
b
ed
@
js
u
m
s
.
ed
u
1.
I
NT
RO
D
UCT
I
O
N
I
n
r
ec
en
t
y
ea
r
s
,
m
a
n
y
c
o
r
es
a
r
e
tr
en
d
in
g
as
a
on
-
ch
i
p
co
m
p
u
tin
g
p
latf
o
r
m
[
1
]
–
[
3
]
th
at
c
an
p
r
o
v
id
e
m
ass
iv
e
co
m
p
u
tatio
n
al
p
o
wer
f
o
r
a
h
eter
o
g
e
n
o
u
s
co
m
p
u
tin
g
en
v
ir
o
n
m
en
t
f
o
r
b
ig
d
ata
[
4
]
an
d
o
th
e
r
co
m
p
u
te
in
ten
s
iv
e
em
b
ed
d
e
d
ar
tific
ial
in
tellig
en
ce
ap
p
licatio
n
s
[
5
]
.
So
m
e
r
ec
en
t
wo
r
k
[
6
]
–
[
9
]
o
n
h
ig
h
p
er
f
o
r
m
an
c
e
co
m
p
u
tin
g
f
o
r
b
ig
d
ata
h
a
v
e
f
o
cu
s
ed
o
n
p
r
o
ce
s
s
in
g
f
r
am
ewo
r
k
,
ar
c
h
itectu
r
e
s
y
n
th
esis
an
d
u
tili
za
tio
n
o
f
m
u
ltip
le
co
r
e
s.
W
ith
i
n
cr
ea
s
ed
v
er
y
l
a
r
g
e
-
s
ca
le
i
n
teg
r
atio
n
(
VL
SI)
d
en
s
ity
,
it
m
ay
b
e
s
till
m
an
ag
ea
b
le
to
p
r
o
v
id
e
h
eter
o
g
en
eo
u
s
co
m
p
u
tin
g
u
s
in
g
c
o
s
t
ef
f
ec
tiv
e
on
-
ch
ip
in
ter
co
n
n
ec
tio
n
a
n
d
ca
c
h
e
m
em
o
r
y
s
y
s
tem
.
Fro
m
p
ast
r
esear
ch
o
n
b
u
s
-
b
ased
in
ter
co
n
n
ec
tio
n
f
o
r
lar
g
e
p
ar
allel
p
r
o
ce
s
s
in
g
s
y
s
tem
s
[
1
0
]
,
it
was
d
eter
m
in
ed
th
at
r
eg
u
lar
b
u
s
c
o
n
n
ec
ted
m
u
ltip
le
-
b
u
s
in
ter
c
o
n
n
ec
tio
n
th
at
u
s
es
n
u
m
b
er
o
f
b
u
s
es
eq
u
al
to
o
n
e
-
h
alf
o
f
th
e
c
o
r
es
o
r
m
em
o
r
y
m
o
d
u
les
,
g
iv
es
co
m
p
ar
a
b
le
m
em
o
r
y
b
an
d
wid
th
.
H
o
wev
e
r
,
th
e
r
ed
u
ce
d
b
u
s
in
ter
co
n
n
ec
tio
n
is
co
s
tly
f
o
r
c
h
ip
m
u
lt
i
p
ro
c
e
ss
o
r
(
C
MP)
d
u
e
t
o
lar
g
e
n
u
m
b
e
r
o
f
b
u
s
-
co
r
e/m
em
o
r
y
c
o
n
n
ec
tio
n
s
.
I
n
o
u
r
ea
r
lier
r
esear
ch
,
we
p
r
o
p
o
s
ed
a
c
o
s
t
-
ef
f
ec
tiv
e
in
ter
co
n
n
ec
tio
n
u
s
in
g
g
e
o
m
etr
ical
p
atter
n
s
f
o
r
b
u
s
-
co
r
e/m
em
o
r
y
co
n
n
ec
tio
n
s
[
1
1
]
with
r
ed
u
ce
d
n
u
m
b
er
o
f
b
u
s
es
.
T
h
e
ap
p
r
o
ac
h
in
[
1
1
]
was
ex
ten
d
ed
to
s
y
s
tem
lev
el
co
n
f
ig
u
r
atio
n
d
ef
in
ed
with
th
r
ee
g
eo
m
et
r
ical
s
y
s
tem
co
n
f
ig
u
r
atio
n
s
ter
m
e
d
a
s
g
eo
m
etr
ical
b
u
s
in
ter
co
n
n
ec
tio
n
(
GB
I
)
[
1
2
]
f
o
r
b
u
s
-
m
em
o
r
y
co
n
n
ec
tio
n
s
u
s
in
g
r
h
o
m
b
ic
co
n
n
ec
tio
n
p
atte
r
n
as
th
e
b
ase.
W
e
ac
h
iev
ed
c
o
s
t
s
av
in
g
s
f
r
o
m
1
.
8
to
2
.
4
with
GB
I
co
m
p
ar
e
d
to
r
eg
u
lar
r
e
d
u
ce
d
b
u
s
in
ter
co
n
n
ec
tio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
9
-
4
8
6
4
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
,
Vo
l.
11
,
No
.
1
,
M
ar
c
h
20
22
:
93
-
1
0
2
94
Ho
wev
er
,
as
th
e
o
v
er
all
th
r
o
u
g
h
p
u
t
o
f
th
e
m
a
n
y
-
co
r
e
C
MP
i
s
al
s
o
d
eter
m
in
ed
b
y
th
e
ca
ch
e
s
y
s
tem
p
er
f
o
r
m
an
ce
,
a
ch
ie
v
in
g
h
i
g
h
o
v
er
all
C
MP
th
r
o
u
g
p
u
t
with
co
s
t
an
d
p
er
f
o
r
m
an
ce
e
f
f
icien
t
in
ter
co
n
n
ec
tio
n
an
d
ca
ch
e
s
y
s
tem
is
h
ig
h
ly
d
esira
b
le
to
d
ay
.
Pro
v
id
in
g
an
a
d
eq
u
ate
an
d
s
u
s
tain
ed
m
an
y
-
co
r
e
C
MP
th
r
o
u
g
h
p
u
t
b
ec
o
m
es
m
o
r
e
ch
alle
n
g
in
g
as
it
also
r
eq
u
ir
es
ef
f
icien
t
ca
ch
e
s
y
s
tem
s
o
lu
tio
n
.
T
o
war
d
s
th
is
ch
allen
g
e,
o
u
r
f
o
c
u
s
is
to
p
r
esen
t
a
co
s
t
-
ef
f
ec
tiv
e
m
u
lti
-
lev
el
ca
ch
e
s
y
s
tem
to
im
p
r
o
v
e
t
h
e
o
v
e
r
all
m
an
y
-
co
r
e
C
MP
th
r
o
u
g
h
p
u
t
u
s
in
g
c
o
m
p
ar
ab
le
m
em
o
r
y
b
an
d
wid
th
r
esu
lts
f
r
o
m
c
o
s
t
-
ef
f
ec
tiv
e
GB
I
[
1
2
]
.
A
ty
p
ical
g
en
er
al
m
u
lti
-
lev
el
ca
ch
e
s
y
s
tem
h
ier
ar
ch
y
f
o
r
m
u
lti
-
co
r
e
s
y
s
tem
s
as
s
h
o
wn
i
n
Fig
u
r
e
1
h
as
L
1
an
d
L
2
p
r
i
v
ate
ca
ch
e
p
e
r
co
r
e
at
l
ev
el
s
1
an
d
2
,
an
d
a
s
h
ar
e
d
ca
ch
e
L
3
as
a
last
lev
el
ca
ch
e
(
L
L
C
)
at
lev
el
3
.
Fo
r
ex
am
p
le,
s
o
m
e
o
f
th
e
cu
r
r
en
t
m
ain
s
tr
ea
m
co
m
m
er
cial
m
u
lti
-
co
r
e
p
r
o
ce
s
s
o
r
s
u
c
h
as
In
tel
®
Co
re
™
i
5
p
r
o
c
e
ss
o
r
h
a
s
th
r
ee
lev
els
o
f
ca
ch
e
with
p
er
co
r
e
L
1
with
a
s
ep
ar
ate
in
s
tr
u
ctio
n
an
d
d
ata
ca
ch
e
,
a
p
er
c
o
r
e
L
2
u
n
if
ie
d
(
in
s
tr
u
ctio
n
/d
ata)
ca
ch
e
a
n
d
a
s
h
ar
ed
L
3
ca
ch
e
as
L
L
C
(
s
h
ar
ed
b
y
all
co
r
es
)
.
Fig
u
r
e
1
.
T
r
ad
itio
n
al
m
u
lti
-
lev
el
ca
ch
e
s
y
s
tem
with
L
1
,
L
2
a
n
d
L
3
f
o
r
m
u
lti
-
co
r
e
C
MP
Ad
d
in
g
a
lar
g
e
n
u
m
b
er
o
f
p
e
r
co
r
e
f
ast
o
n
-
ch
ip
p
r
iv
ate
L
1
an
d
L
2
ca
ch
e
s
with
a
s
h
ar
ed
L
3
m
ay
in
cr
ea
s
e
ca
ch
e
s
y
s
tem
co
s
t.
A
s
a
r
esu
lt,
we
p
r
o
p
o
s
e
a
n
alter
n
ativ
e
s
o
lu
tio
n
b
y
co
m
b
i
n
in
g
L
1
with
a
r
elativ
ely
s
lo
wer
s
h
ar
ed
b
u
s
ca
c
h
e
(
SB
C
)
as
L
L
C
ad
d
e
d
to
ev
e
r
y
b
u
s
lin
e
o
f
GB
I
[
1
2
]
in
wh
ich
th
e
d
ata
r
eq
u
est
o
f
all
co
r
es
is
s
h
ar
ed
v
ia
G
B
I
.
I
n
ad
d
itio
n
,
o
u
r
p
r
o
p
o
s
ed
ca
ch
e
s
y
s
tem
s
o
lu
tio
n
m
ay
also
p
r
o
v
id
e
th
e
ab
ilit
y
to
in
cr
ea
s
e
th
e
ca
ch
e
lev
els an
d
s
izes w
ith
in
th
e
ca
ch
e
h
ier
ar
ch
y
u
p
o
n
ca
ch
e
r
ec
o
n
f
i
g
u
r
atio
n
in
o
r
d
e
r
to
o
p
tim
ize
th
e
s
y
s
tem
f
o
r
co
s
t,
p
e
r
f
o
r
m
an
ce
an
d
p
o
wer
co
n
s
u
m
p
tio
n
.
So
m
e
ea
r
lier
r
esear
ch
[
1
3
]
–
[
1
6
]
h
av
e
ad
d
r
ess
ed
v
ar
io
u
s
ca
ch
e
s
y
s
tem
ar
ch
itectu
r
e,
is
s
u
es
an
d
s
o
lu
tio
n
s
f
o
r
im
p
r
o
v
e
d
p
er
f
o
r
m
an
ce
.
I
n
[
1
3
]
,
th
e
a
u
th
o
r
s
ad
d
r
ess
ed
an
aly
zin
g
m
em
o
r
y
p
e
r
f
o
r
m
a
n
ce
f
o
r
tiled
m
an
y
-
co
r
e
C
MP.
L
in
et
a
l
.
[
1
4
]
s
u
g
g
ested
h
y
b
r
i
d
ca
ch
e
s
y
s
tem
s
th
at
in
clu
d
ed
lay
e
r
s
f
o
r
ca
ch
e
ar
ch
itectu
r
e
f
r
o
m
m
em
o
r
y
t
o
d
ata
b
ase
t
o
im
p
r
o
v
e
p
er
f
o
r
m
an
ce
in
s
p
ec
if
ic
r
elatio
n
al
d
ata
b
ase
q
u
er
y
f
o
r
b
ig
d
ata
ap
p
licatio
n
s
.
C
h
ar
les
et
a
l
.
[
1
5
]
lo
o
k
ed
at
ca
ch
e
r
ec
o
n
f
ig
u
r
a
tio
n
f
o
r
n
etwo
r
k
-
on
-
ch
ip
(
N
o
C
)
b
ased
m
an
y
-
c
o
r
e
C
MP.
Saf
ay
en
ik
o
o
et
a
l
.
[
1
6
]
s
u
g
g
ested
an
en
er
g
y
-
ef
f
icien
t
ca
ch
e
ar
ch
itectu
r
e
to
ad
d
r
ess
th
e
p
r
o
b
lem
o
f
in
cr
ea
s
ed
leak
ag
e
p
o
wer
r
esu
ltin
g
f
r
o
m
lar
g
e
ar
ea
o
f
L
L
C
(
as
m
u
ch
as
5
0
%
o
f
th
e
ch
ip
ar
ea
)
d
u
e
to
its
in
cr
ea
s
ed
s
ize.
Mo
s
t
o
f
th
e
wo
r
k
r
ep
o
r
te
d
in
[
1
3
]
–
[
1
6
]
m
ay
r
e
q
u
ir
e
co
m
p
le
x
ca
ch
e
d
esig
n
p
r
o
ce
s
s
.
Ou
r
p
r
o
p
o
s
ed
ca
ch
e
s
y
s
tem
s
o
lu
tio
n
is
s
im
p
le
an
d
d
o
n
o
t
ad
d
an
y
ex
tr
a
o
r
d
if
f
icu
lt
ca
ch
e
d
esig
n
p
r
o
ce
s
s
.
Ou
r
m
ain
c
o
n
tr
ib
u
tio
n
i
n
th
is
p
a
p
er
ar
e
as:
i)
Pro
p
o
s
e
a
s
h
ar
ed
b
u
s
ca
ch
e
(
SB
C
)
with
in
a
m
u
lti
-
lev
el
ca
ch
e
s
y
s
tem
;
ii)
Pre
s
en
t
a
leas
t
r
ec
en
tly
u
s
ed
(
L
R
U)
m
u
lti
-
lev
el
ca
ch
e
s
y
s
tem
s
im
u
latio
n
to
ex
tr
ac
t
h
it
an
d
m
is
s
co
n
cu
r
r
en
cies
;
iii)
Ap
p
ly
c
o
n
cu
r
r
en
t
av
er
a
g
e
m
em
o
r
y
ac
ce
s
s
tim
e
(C
-
AM
AT
)
[
1
7
]
to
ac
cu
r
ately
d
eter
m
in
e
th
e
s
y
s
tem
th
r
o
u
g
h
p
u
t
p
er
f
o
r
m
an
ce
an
d
p
r
esen
t
o
u
r
r
esu
lts
;
an
d
iv
)
P
r
o
v
id
e
co
n
clu
s
io
n
an
d
p
r
esen
t
s
o
m
e
in
s
ig
h
t in
to
f
u
t
u
r
e
r
esear
ch
.
2.
L1
-
SB
C
CA
CH
E
SY
ST
E
M
Fig
u
r
e
2
s
h
o
ws
a
s
y
s
tem
with
L
1
an
d
s
h
ar
e
b
u
s
ca
ch
e
at
e
v
er
y
b
u
s
lin
e
o
f
GB
I
[
1
2
]
.
W
e
ter
m
th
e
m
em
o
r
y
s
y
s
tem
u
s
in
g
L
1
p
r
i
v
ate
ca
ch
e
as
L1
,
with
L
1
an
d
L
2
as
L1
2
,
with
L1
a
n
d
s
h
ar
e
d
b
u
s
ca
ch
e
as
L
1
-
SB
C
th
r
o
u
g
h
o
u
t th
is
p
ap
e
r
.
Fig
u
r
e
2
.
L1
-
SB
C
L1
Co
r
es
(1
-
n)
O
f
f
-
ch
i
p
G
l
o
b
a
l
M
em
o
r
y
L2
S
h
ar
e
d
L3
I
n
t
er
co
n
n
ec
t
b
S
h
a
r
e
d
B
u
s
C
a
c
h
e
s
(
S
B
C
)
G
BI
m
Off
-
c
h
i
p
M
e
m
o
r
y
M
o
d
u
l
e
s
n
c
o
r
e
s
a
n
d
n
L
1
c
a
c
h
e
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
I
SS
N:
2089
-
4
8
6
4
A
n
efficien
t m
u
lti
-
leve
l c
a
ch
e
s
ystem
fo
r
g
eo
metrica
lly
in
te
r
co
n
n
ec
ted
m
a
n
y
-
co
r
e
… (
Tir
u
ma
le
R
a
mesh
)
95
2
.
1
.
Co
ncurre
nt
a
v
er
a
g
e
m
emo
r
y
a
cc
ess
t
im
e
(C
-
AM
AT
)
S
o
m
e
ca
ch
e
tech
n
iq
u
es
[
1
8
]
–
[
2
0
]
we
r
e
s
u
g
g
ested
ea
r
lier
f
o
r
im
p
r
o
v
in
g
tr
ad
itio
n
al
av
e
r
ag
e
m
em
o
r
y
ac
ce
s
s
tim
e
f
o
r
m
u
lti
-
lev
el
ca
c
h
e
s
y
s
tem
s
.
I
n
[
1
8
]
,
h
a
r
d
war
e
p
r
ef
etch
in
g
was
c
o
n
s
id
er
ed
to
ex
p
lo
it
s
p
atial
a
n
d
tem
p
o
r
al
lo
ca
lity
o
f
r
e
f
er
en
c
es.
I
n
[
1
9
]
,
m
u
lti
-
lev
el
ca
ch
es
wer
e
co
n
s
id
er
e
d
as
p
r
im
ar
y
an
d
s
ec
o
n
d
ar
y
m
em
o
r
ies
f
o
r
p
r
o
x
y
s
er
v
er
s
t
o
ac
ce
s
s
web
co
n
ten
t.
I
n
[
2
0
]
,
an
L
R
U
r
e
p
lace
m
en
t
p
o
licy
was
p
r
o
p
o
s
ed
th
a
t
m
ak
es
u
s
e
o
f
th
e
awa
r
en
ess
o
f
th
e
ca
ch
e
m
is
s
-
p
en
alty
t
o
en
s
u
r
e
m
em
o
r
y
ac
ce
s
s
laten
cy
is
b
alan
ce
d
f
o
r
m
em
o
r
y
s
y
s
tem
b
u
ilt
with
d
if
f
er
en
t
m
em
o
r
y
tech
n
o
lo
g
ies
te
r
m
ed
as
“h
y
b
r
id
”
s
y
s
tem
.
T
h
e
wo
r
k
ad
d
r
ess
ed
in
[
1
8
]
–
[
2
0
]
wer
e
s
p
ec
if
ic
ca
c
h
e
tech
n
iq
u
es
attem
p
ted
t
o
r
ed
u
ce
av
er
a
g
e
m
em
o
r
y
a
cc
ess
tim
e
with
o
u
t
co
n
s
id
er
in
g
an
y
co
s
t
im
p
licatio
n
s
.
Ou
r
ap
p
r
o
ac
h
is
to
o
p
tim
ize
ca
ch
e
an
d
in
ter
co
n
n
ec
tio
n
co
s
t
ac
r
o
s
s
th
e
ca
ch
e
lev
els
an
d
a
p
p
ly
C
-
A
MA
T
f
o
r
ex
p
lo
itatio
n
o
f
p
ar
allel
co
n
cu
r
r
e
n
cy
in
ca
ch
e
h
it
an
d
mis
s
e
s
th
at
ac
cu
r
ately
d
eter
m
in
e
th
e
av
er
ag
e
m
em
o
r
y
ac
ce
s
s
tim
e
ac
r
o
s
s
all
lev
els
f
o
r
d
ata
ac
ce
s
s
.
An
an
aly
tical
m
eth
o
d
f
o
r
d
eter
m
in
in
g
C
-
AM
AT
is
b
r
ief
ly
p
r
o
v
id
ed
b
elo
w.
A
tr
ad
itio
n
al
av
e
r
ag
e
m
em
o
r
y
ac
ce
s
s
tim
e
(
AM
AT
)
with
a
m
u
lti
-
lev
el
ca
ch
e
s
y
s
te
m
is
g
iv
en
in
(
1
)
a
n
d
(
2
)
f
o
r
L
1
an
d
L
1
2
ca
c
h
e
s
y
s
tem
s
r
esp
ec
tiv
ely
.
1
=
1
ℎ
1
+
(
1
−
ℎ
1
)
(
1
)
2
=
1
ℎ
1
+
(
1
−
ℎ
1
)
(
2
ℎ
2
+
(
1
−
ℎ
2
)
(
2
)
W
h
er
e
t
1
an
d
t
2
ar
e
th
e
ca
ch
e
a
cc
ess
tim
e
f
o
r
lev
el
1
an
d
lev
el
2
ca
ch
es,
h
1
an
d
h
2
ar
e
ca
ch
e
h
it
r
atio
s
f
o
r
lev
el
1
an
d
lev
el
2
ca
ch
es
an
d
t
m
is
th
e
g
lo
b
al
m
em
o
r
y
ac
ce
s
s
tim
e.
I
n
o
u
r
ap
p
r
o
ac
h
,
we
ex
p
lo
it
p
ar
allel
co
n
cu
r
r
en
cy
f
o
r
co
r
e
an
d
SB
C
h
it
an
d
m
is
s
co
n
cu
r
r
en
cy
f
o
r
SB
C
s
u
p
p
o
r
ted
b
y
GB
I
an
d
ap
p
ly
C
-
AM
AT
f
o
r
p
er
f
o
r
m
an
ce
ev
alu
atio
n
.
T
h
e
h
it
co
n
cu
r
r
e
n
cy
will
im
p
r
o
v
e
p
er
f
o
r
m
an
ce
wh
ile
a
ca
ch
e
m
is
s
m
ay
im
p
ac
t
th
e
m
em
o
r
y
s
y
s
tem
p
er
f
o
r
m
an
ce
,
d
ep
en
d
in
g
o
n
h
it
co
n
cu
r
r
en
c
y
.
T
ak
in
g
a
d
v
an
tag
e
o
f
m
u
lti
p
le
b
u
s
es
with
m
is
s
co
n
cu
r
r
en
cy
,
h
ig
h
er
s
y
s
tem
p
er
f
o
r
m
a
n
ce
ca
n
b
e
ac
h
iev
ed
.
Ho
wev
er
,
th
e
ap
p
licatio
n
o
f
C
-
AM
AT
n
ee
d
to
en
s
u
r
e
th
at
th
e
m
is
s
co
n
cu
r
r
e
n
cy
d
o
n
o
t
ex
ce
ed
th
e
in
ter
c
o
n
n
ec
tio
n
b
an
d
wid
th
with
r
e
d
u
ce
d
n
u
m
b
er
o
f
b
u
s
e
.
T
h
u
s
,
we
r
e
-
wr
ite
(
1
)
an
d
(
2
)
as
(
3
)
an
d
(
4
)
.
−
1
=
1
ℎ
1
ℎ1
+
(
1
−
ℎ
1
)
/
(
3
)
−
2
=
1
ℎ
1
ℎ1
+
(
1
−
ℎ
1
)
(
2
ℎ
2
ℎ2
+
(
1
−
ℎ
2
)
/
(
4
)
W
h
er
e
ℎ
1
an
d
ℎ
2
ar
e
th
e
av
e
r
ag
e
h
it
cy
cle
co
n
cu
r
r
en
c
y
at
lev
el
s
1
an
d
2
a
n
d
is
th
e
av
er
ag
e
m
is
s
cy
cle
co
n
cu
r
r
e
n
cy
.
I
n
t
h
is
p
ap
er
,
we
e
v
alu
ate
L
1
,
L
1
2
an
d
L
1
-
SB
C
s
y
s
tem
s
.
W
e
s
elec
ted
m
in
im
u
m
n
u
m
b
er
o
f
L
1
an
d
SB
C
ca
ch
e
b
lo
ck
s
to
m
ee
t
t
h
e
f
o
llo
win
g
cr
iter
io
n
f
o
r
h
it
a
n
d
m
is
s
co
n
cu
r
r
e
n
cy
g
i
v
en
as
(
5
)
.
c
ℎ
≤
,
c
≤
/
2
(
5
)
Sin
ce
th
e
GB
I
in
ter
co
n
n
ec
tio
n
p
r
o
v
id
es
a
m
em
o
r
y
b
an
d
wi
d
th
o
f
/
2
,
we
ca
n
also
ap
p
r
o
x
im
ate
(
4
)
b
y
m
is
s
co
n
cu
r
r
en
cy
s
u
p
p
o
r
ted
b
y
th
e
GB
I
m
em
o
r
y
b
an
d
wid
th
as
(
6
)
.
−
=
1
ℎ
1
ℎ1
+
(
1
−
ℎ
1
)
(
2
ℎ
2
ℎ2
+
(
1
−
ℎ
2
)
2
/
(
6
)
W
h
en
th
e
is
less
th
an
/
2
,
th
e
in
ter
co
n
n
ec
tio
n
b
an
d
wid
th
is
n
o
t
f
u
lly
u
tili
ze
d
.
T
h
e
C
-
AM
AT
g
iv
en
in
(
6
)
is
s
m
aller
co
m
p
ar
ed
to
co
n
s
er
v
ativ
e
m
is
s
co
n
cu
r
r
e
n
cy
g
iv
e
n
in
(
4
)
.
T
h
e
p
er
ce
n
tag
e
d
e
v
iat
io
n
f
r
o
m
(
4
)
to
(
6
)
v
a
r
ies
f
r
o
m
4
t
o
3
0
%
ac
r
o
s
s
all
ca
ch
e
s
y
s
te
m
s
.
W
e
s
ee
a
h
ig
h
er
d
ev
iatio
n
f
o
r
L
1
-
SB
C
s
y
s
te
m
wh
ich
is
attr
ib
u
ted
to
th
e
f
ac
t
th
at
th
e
m
is
s
co
n
cu
r
r
en
cy
d
ec
r
ea
s
es
as
a
r
esu
lt
o
f
h
ig
h
er
h
it
co
n
cu
r
r
en
cy
u
s
in
g
b
u
s
ca
ch
e
d
u
r
in
g
r
ea
d
cy
cle.
I
n
th
is
p
ap
e
r
,
we
o
n
ly
i
n
clu
d
e
co
n
s
er
v
ativ
e
r
esu
lts
f
r
o
m
(
3
)
an
d
(
4
)
f
o
r
L
1
a
n
d
L
1
2
ca
ch
e
s
y
s
tem
s
r
esp
ec
tiv
el
y
an
d
at
t
h
e
s
am
e
tim
e
en
s
u
r
in
g
cr
iter
io
n
(
5
)
.
2
.
2
.
G
eo
m
et
ric
a
l
bu
s
inte
rc
o
nn
ec
t
io
n
(
G
B
I
)
[
1
2
]
c
o
s
t
T
ab
le
1
g
iv
es
t
h
e
av
e
r
ag
e
n
o
r
m
alize
d
i
n
ter
co
n
n
ec
tio
n
co
s
t
o
f
GB
I
co
m
p
ar
e
d
to
f
u
ll
y
r
ed
u
ce
d
m
u
ltip
le
b
u
s
s
y
s
tem
[
1
0
]
.
W
e
n
o
tice
a
r
ed
u
ctio
n
o
f
ab
o
u
t
3
0
%
in
co
s
t
ac
r
o
s
s
th
e
n
u
m
b
er
o
f
co
r
es
f
r
o
m
1
6
to
64.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
9
-
4
8
6
4
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
,
Vo
l.
11
,
No
.
1
,
M
ar
c
h
20
22
:
93
-
1
0
2
96
T
ab
le
1
.
No
r
m
alize
d
av
er
ag
e
GB
I
co
s
t c
o
m
p
ar
ed
to
f
u
lly
r
e
d
u
ce
d
b
u
s
s
y
s
tem
[
1
0
]
N
o
.
o
f
C
o
r
e
s
N
o
r
mal
i
z
e
d
C
o
st
16
0
.
6
9
32
0
.
6
6
64
0
.
6
4
2
.
3
.
SB
C
i
m
pa
ct
o
n C
-
AM
A
T
I
n
th
e
p
ast,
s
o
m
e
s
h
ar
ed
ca
ch
e
tech
n
iq
u
es
[
2
1
]
h
a
v
e
lo
o
k
e
d
at
ca
ch
e
s
h
ar
in
g
o
f
way
s
b
ased
o
n
h
ash
m
ap
p
in
g
in
s
tead
o
f
tr
ad
itio
n
a
l
c
ac
h
e
s
et
s
h
ar
in
g
f
o
r
m
u
lti
-
co
r
e
p
latf
o
r
m
s
.
I
n
g
e
n
er
al,
it
is
k
n
o
wn
th
at
b
y
in
cr
ea
s
in
g
th
e
n
u
m
b
er
o
f
p
r
o
ce
s
s
o
r
co
r
es
ca
n
d
ir
ec
tl
y
in
cr
ea
s
e
L
L
C
(
last
lev
el
ca
ch
e)
h
it
a
n
d
m
is
s
co
n
cu
r
r
en
cy
g
iv
in
g
r
ed
u
ce
d
C
-
AM
AT
.
As
o
u
r
s
y
s
tem
u
s
es
b
u
s
es
eq
u
al
to
o
n
e
-
h
alf
th
e
n
u
m
b
er
o
f
co
r
es,
th
e
m
em
o
r
y
ac
ce
s
s
m
is
s
ed
in
p
er
co
r
e
ca
ch
e
is
s
ea
r
ch
ed
in
SB
C
.
Sin
ce
a
s
h
ar
ed
r
ed
u
ce
d
n
u
m
b
er
o
f
b
u
s
es
in
o
u
r
ap
p
r
o
ac
h
n
atu
r
ally
ca
p
t
u
r
es
all
co
r
e
ac
ce
s
s
es
v
ia
th
e
b
u
s
in
t
er
co
n
n
ec
tio
n
,
p
lacin
g
a
n
SB
C
at
ea
ch
b
u
s
lin
e
o
f
GB
I
r
ep
licates
clo
s
e
ly
to
a
t
r
ad
itio
n
al
L
3
s
h
ar
e
d
ca
c
h
e
n
o
r
m
ally
u
s
ed
in
cu
r
r
en
t
co
m
m
er
cial
p
r
o
ce
s
s
o
r
s
y
s
tem
s
.
As
we
u
s
ed
2
n
u
m
b
er
o
f
SB
C
at
lev
el
2
,
a
n
y
m
is
s
in
L
1
in
cr
ea
s
es
th
e
h
it
co
n
c
u
r
r
en
cy
in
SB
C
.
I
n
o
u
r
ap
p
r
o
ac
h
,
we
ac
co
u
n
ted
o
n
ly
a
p
u
r
e
m
is
s
co
n
cu
r
r
en
c
y
[
1
7
]
(
o
n
ly
if
n
o
n
e
o
f
th
e
b
u
s
ca
c
h
e
h
as
a
h
it
in
th
e
h
it
cy
cle,
a
m
is
s
is
ac
co
u
n
ted
)
.
2
.
4
.
Ca
che
a
s
s
o
cia
t
io
n
i
m
pa
ct
o
n
C
-
AM
AT
C
ac
h
e
ass
o
ciatio
n
ca
n
also
im
p
ac
t
o
u
r
s
o
l
u
tio
n
.
Au
th
o
r
s
in
[
2
2
]
,
[
2
3
]
attr
ib
u
te
d
t
o
th
e
f
ac
t
th
at
h
ig
h
er
ca
ch
e
ass
o
ciatio
n
n
o
r
m
ally
in
cr
ea
s
es
th
e
ca
ch
e
h
it
r
ate
b
u
t
at
th
e
ex
p
en
s
e
o
f
h
ar
d
w
ar
e
co
m
p
lex
ity
f
o
r
th
e
ca
ch
e
co
n
tr
o
ller
an
d
ad
d
iti
o
n
al
laten
cy
f
o
r
ca
ch
e
s
ea
r
ch
t
im
e
with
in
cr
ea
s
ed
ass
o
ciatio
n
.
Ho
wev
er
,
in
o
u
r
ap
p
r
o
ac
h
,
th
e
ass
o
cia
tio
n
was
s
elec
ted
to
en
s
u
r
e
th
at
cr
iter
ia
(
5
)
ar
e
s
atis
f
ied
.
T
h
u
s
,
s
elec
ti
n
g
a
d
ir
ec
t
m
a
p
p
ed
ca
ch
e
m
ay
b
en
e
f
it
to
ac
h
iev
e
r
ed
u
ce
d
C
-
AM
AT
.
I
n
g
e
n
er
al,
m
is
s
co
n
cu
r
r
en
cies
in
L
L
C
ca
n
n
o
r
m
ally
b
e
s
u
p
p
o
r
ted
b
y
u
s
e
o
f
m
u
lti
-
p
o
r
ted
m
em
o
r
y
,
o
r
m
u
lti
-
b
a
n
k
m
em
o
r
y
(
m
em
o
r
y
m
o
d
u
les)
with
a
s
in
g
le
b
u
s
.
Ho
wev
er
,
f
o
r
a
s
in
g
le
b
u
s
s
y
s
tem
,
b
u
s
c
o
n
ten
tio
n
im
p
ac
ts
th
e
th
r
o
u
g
h
p
u
t
p
er
f
o
r
m
a
n
ce
.
T
h
e
m
is
s
co
n
cu
r
r
e
n
cy
ca
n
b
e
f
ac
ilit
ated
b
y
u
s
in
g
a
m
u
lti
-
b
an
k
m
em
o
r
y
m
o
d
u
le
with
m
u
ltip
le
b
u
s
in
ter
co
n
n
ec
tio
n
b
etwe
en
s
h
ar
ed
ca
ch
e
an
d
m
e
m
o
r
y
m
o
d
u
les.
T
h
e
m
is
s
co
n
cu
r
r
en
c
y
ca
n
b
e
s
u
p
p
o
r
ted
b
y
m
u
ltip
le
b
u
s
es in
GB
I
y
ield
in
g
lo
wer
C
-
AM
AT
.
3.
CACH
E
SY
ST
E
M
SI
M
U
L
A
T
I
O
N
3
.
1
.
Sy
s
t
em
o
pera
t
io
n wit
h L
1
-
SB
C
Fig
u
r
e
2
s
h
o
ws
th
e
o
p
er
atio
n
f
lo
wch
ar
t
f
o
r
r
ea
d
an
d
wr
ite
cy
cles
f
o
r
L
1
s
y
s
tem
.
SB
C
is
u
s
ed
o
n
ly
d
u
r
in
g
“r
ea
d
cy
cle”
with
a
“
wr
ite
th
r
o
u
g
h
”
p
o
licy
to
u
p
d
ate
o
n
ca
c
h
e
m
is
s
.
I
n
“n
o
r
m
al
n
o
-
f
a
u
lt
m
o
d
e”
,
d
u
r
in
g
r
ea
d
cy
cle,
th
e
d
ata
is
f
ir
s
t sear
ch
ed
in
L
1
.
I
f
th
e
L
1
r
ea
d
is
a
“
m
i
s
s
,
”
i
t is
th
en
s
ea
r
c
h
ed
in
SB
C
.
I
f
i
t i
s
“
h
it
,
”
th
e
d
ata
is
ca
ch
e
d
.
O
n
r
ea
d
“
mis
s
”
in
SB
C
,
b
u
s
es
in
GB
I
ar
e
a
r
b
itra
ted
to
u
tili
ze
f
u
ll
m
em
o
r
y
b
an
d
wid
th
a
n
d
th
e
d
ata
is
r
ea
d
f
r
o
m
t
h
e
g
lo
b
al
m
e
m
o
r
y
m
o
d
u
le
an
d
is
wr
itten
to
SB
C
an
d
L
1
ca
ch
e
as
well.
I
f
th
e
cu
r
r
en
t
b
u
s
th
at
is
g
r
an
ted
f
ails
,
th
en
ca
ch
e
s
y
s
tem
s
witch
es
to
“b
u
s
f
au
lt
m
o
d
e”
an
d
th
e
in
ter
co
n
n
ec
tio
n
is
r
e
-
a
r
b
itra
te
d
to
u
s
e
o
th
er
b
-
1
c
o
n
n
ec
ted
b
u
s
es.
Af
ter
b
u
s
r
e
-
ar
b
itra
ti
o
n
,
th
e
d
ata
is
r
e
-
s
ea
r
ch
ed
f
ir
s
t
in
L
1
a
n
d
if
“
h
it
”,
th
e
d
ata
is
ca
ch
e
d
in
L
1
ca
ch
e,
o
th
e
r
wis
e
s
ea
r
ch
ed
in
SB
C
.
Du
r
in
g
wr
ite
cy
cle
,
if
th
e
L
1
ca
c
h
e
b
lo
ck
i
s
“
p
r
esen
t
”,
th
e
n
d
ata
is
wr
itten
in
to
L
1
ca
ch
e.
On
L
1
wr
ite
“
m
is
s
”,
th
e
L
1
ca
ch
e
b
lo
ck
is
r
ep
lace
d
an
d
th
e
d
ata
is
u
p
d
ated
to
L
1
an
d
co
n
s
eq
u
en
tly
th
e
d
ata
is
wr
itten
to
g
lo
b
al
m
em
o
r
y
u
s
in
g
ar
b
itra
ted
b
u
s
es in
GB
I
.
T
h
e
p
r
o
p
o
s
ed
ca
ch
e
s
y
s
tem
was
s
im
u
lated
u
s
in
g
p
u
b
licly
av
ailab
le
“
l
r
u
ca
ch
e
”
lib
r
ar
ie
s
in
p
yth
o
n
an
d
cr
ea
ted
m
u
ltip
le
o
b
jects
o
f
a
“
lr
u
ca
c
h
e
”
with
in
d
e
x
in
g
to
im
p
lem
en
t
L
1
,
L
2
an
d
SB
C
.
W
e
iter
at
ed
ca
ch
e
o
p
er
atio
n
f
o
r
o
v
er
1000
f
o
r
n
n
u
m
b
er
o
f
c
o
r
es.
T
a
b
le
2
s
h
o
ws
th
e
g
en
er
al
p
ar
a
m
eter
s
u
s
ed
f
o
r
th
e
s
im
u
latio
n
.
Usi
n
g
as
m
u
ch
o
f
in
s
ig
h
t
in
to
t
o
d
ay
’
s
m
em
o
r
y
t
ec
h
n
o
lo
g
ies,
we
ap
p
r
o
x
im
atel
y
u
s
ed
a
r
elativ
e
b
it
co
s
t f
o
r
L
1
,
L
2
a
n
d
S
B
C
as
g
iv
en
in
T
a
b
le
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
I
SS
N:
2089
-
4
8
6
4
A
n
efficien
t m
u
lti
-
leve
l c
a
ch
e
s
ystem
fo
r
g
eo
metrica
lly
in
te
r
co
n
n
ec
ted
m
a
n
y
-
co
r
e
… (
Tir
u
ma
le
R
a
mesh
)
97
Fig
u
r
e
2
.
L1
-
SB
C
Op
er
atio
n
T
ab
le
2
.
C
ac
h
e
s
y
s
tem
s
im
u
latio
n
p
ar
am
ete
r
s
C
l
o
c
k
c
y
c
l
e
L1
a
c
c
e
ss
c
y
c
l
e
s
(
t
1
)
S
B
C
a
c
c
e
ss
c
y
c
l
e
s
(
t
2
)
L2
a
c
c
e
ss
t
i
m
e
(
c
y
c
l
e
s)
G
l
o
b
a
l
me
mo
r
y
a
c
c
e
ss c
y
c
l
e
s
(
t
m)
B
u
s
d
a
t
a
w
i
d
t
h
L1
r
e
l
a
t
i
v
e
c
o
s
t
L2
r
e
l
a
t
i
v
e
c
o
s
t
S
B
C
r
e
l
a
t
i
v
e
c
o
s
t
0
.
5
n
s
(2
G
H
z
)
5
25
10
1
0
0
2
b
y
t
e
s
10
6
3
3
.
2
.
Rela
t
iv
e
no
rm
a
lized
s
y
s
t
em
co
s
t
T
ab
le
3
s
h
o
ws
th
e
n
o
r
m
alize
d
s
y
s
tem
co
s
t
as
th
e
to
tal
s
y
s
tem
co
s
t
th
at
i
n
clu
d
es
th
e
n
o
r
m
alize
d
in
ter
co
n
n
ec
tio
n
co
s
t
f
r
o
m
T
ab
le
1
an
d
r
elativ
e
ca
ch
e
m
em
o
r
y
co
s
t
f
r
o
m
T
ab
le
2
.
As we
n
o
ticed
f
r
o
m
T
ab
le
3
,
L
2
ca
ch
e
ad
d
s
2
%
ad
d
itio
n
a
l
s
y
s
tem
co
s
t
an
d
SB
C
ad
d
s
0
.
5
%
ad
d
itio
n
al
co
s
t.
W
e
r
an
s
im
u
latio
n
s
u
s
in
g
m
in
im
u
m
n
u
m
b
er
o
f
L
1
,
L
2
,
an
d
SB
C
ca
ch
e
b
l
o
ck
s
s
elec
ted
to
m
ee
t
th
e
cr
iter
ia
g
iv
en
in
(
5
)
.
T
o
r
ed
u
ce
t
h
e
ca
ch
e
“h
it”
tim
e,
we
u
s
ed
an
o
p
tim
al
ca
ch
e
ass
o
ciati
on,
b
u
t
at
th
e
s
am
e
tim
e
en
s
u
r
ed
co
n
cu
r
r
en
cy
cr
iter
ia
g
iv
en
b
y
(
5
)
.
T
ab
le
3
.
No
r
m
alize
d
s
y
s
tem
c
o
s
t
C
a
c
h
e
S
y
s
t
e
m
N
o
.
o
f
L
1
b
l
o
c
k
s
L1
a
sso
c
i
a
t
i
o
n
N
o
.
o
f
L
2
set
s
L2
a
sso
c
i
a
t
i
o
n
N
o
.
o
f
S
B
C
set
s
S
B
C
a
sso
c
i
a
t
i
o
n
N
o
r
mal
i
z
e
d
sy
s
t
e
m
c
o
s
t
L1
1
2
8
1
1
L1
2
1
2
8
1
4
2
1
.
0
2
L1
-
S
B
C
1
2
8
1
4
2
1
.
0
0
5
3
.
3
.
Ca
che
re
a
d a
nd
writ
e
m
is
s
es
cr
it
ica
lity
im
pa
c
t
I
t
is
wel
l
k
n
o
wn
th
at
c
ac
h
e
r
ea
d
m
is
s
es
ar
e
m
o
r
e
cr
itical
an
d
in
cu
r
s
m
o
r
e
p
en
alty
in
r
ea
d
th
an
wr
ite
cy
cles
.
T
o
allev
iate
th
is
p
r
o
b
l
em
,
s
o
m
e
r
ea
d
-
wr
ite
p
ar
titi
o
n
in
g
p
o
licy
was
s
u
g
g
ested
in
[
2
4
]
th
at
m
i
n
i
m
izes
th
e
r
ea
d
m
is
s
es
u
s
in
g
d
y
n
a
m
ic
ca
ch
e
m
an
a
g
em
en
t.
T
o
p
r
o
v
id
e
m
o
r
e
r
ea
d
m
is
s
s
u
p
p
o
r
t,
i
n
o
u
r
ap
p
r
o
ac
h
,
we
in
clu
d
ed
SB
C
d
u
r
in
g
r
ea
d
o
n
ly
.
I
n
g
en
er
al,
as
th
e
r
ea
d
is
in
cr
ea
s
ed
f
r
o
m
5
0
to
8
0
%
o
f
th
e
p
r
o
ce
s
s
o
r
d
ata
r
eq
u
ests
,
we
f
o
u
n
d
d
r
asti
c
im
p
r
o
v
em
e
n
t
in
SB
C
h
it
co
n
cu
r
r
en
cy
as
a
r
esu
lt
o
f
its
ex
clu
s
iv
e
s
u
p
p
o
r
t
d
u
r
in
g
r
ea
d
cy
cle
.
H
o
wev
er
,
as
n
o
t
a
ll
ap
p
licatio
n
s
en
s
u
r
e
a
less
d
ata
r
ea
d
s
th
an
d
ata
wr
ites
,
we
m
ay
tr
ea
t
th
e
5
0
%
r
ea
d
d
ata
r
eq
u
ests
as
a
g
o
o
d
co
m
p
ar
is
o
n
f
o
r
n
o
w
an
d
lo
o
k
f
o
r
ap
p
licatio
n
ce
n
tr
ic
r
ea
d
/wr
ite
tr
ad
e
-
o
f
f
s
in
f
u
tu
r
e
u
s
in
g
n
o
v
el
ca
ch
e
p
r
o
t
o
co
ls
.
So
m
e
r
ec
en
t
n
o
v
el
r
ea
d
/wr
ite
co
s
t
tr
ad
eo
f
f
s
f
o
r
DNA
b
ased
d
ata
s
to
r
ag
e
[
2
5
]
h
as b
ee
n
s
u
g
g
ested
.
3
.
4
.
L
1
a
nd
SB
C
hit
co
ncurre
nci
es a
nd
m
is
s
co
ncurre
nc
ies
T
ab
les
4
a
n
d
5
s
h
o
w
t
h
e
L
1
ca
ch
e
h
it
co
n
c
u
r
r
en
c
y
(
ℎ
1
)
,
SB
C
h
it
co
n
c
u
r
r
e
n
cy
(
ℎ
2
)
,
a
n
d
m
is
s
co
n
cu
r
r
en
cy
in
SB
C
(
)
f
o
r
v
ar
io
u
s
s
y
s
tem
s
ize
s
f
o
r
L
1
,
L
1
2
an
d
L
1
-
SB
C
s
y
s
tem
s
f
o
r
5
0
%
an
d
8
0
%
r
ea
d
r
eq
u
ests
r
esp
ec
tiv
ely
.
C
o
r
e
1
ca
ch
e
co
n
tr
o
ller
Fig
u
r
e
2
.
L1
-
S
BC
Op
er
atio
n
Sear
ch
b
lo
ck
in
L
R
U
L
1
L
1
r
ea
d
Ar
b
itra
te
I
n
ter
co
n
n
ec
tio
n
an
d
u
p
d
ate
g
lo
b
al
m
em
o
r
y
a
n
d
ca
ch
e
(
wr
i
te
th
r
o
u
g
h
)
L
1
wr
ite
C
ac
h
e
C
o
n
tr
o
ller
s
[
2
…
]
SB
C
r
ea
d
U
p
d
a
t
e
L1
o
n
S
B
C
mi
ss
L1
r
e
a
d
mi
ss
L
R
U
SB
C
r
ea
d
s
ea
r
ch
[
1
…
]
hit
U
p
d
a
t
e
L1
o
n
S
B
C
mi
ss
L1
w
r
i
t
e
mi
ss
L1
r
e
a
d
mi
ss
C
o
r
es 2
to
n
s
en
d
t
h
eir
r
eq
u
est
to
ca
ch
e
co
n
tr
o
ller
s
C
o
r
e
1
r
eq
u
est
h
i
t
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
9
-
4
8
6
4
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
,
Vo
l.
11
,
No
.
1
,
M
ar
c
h
20
22
:
93
-
1
0
2
98
T
ab
le
4
.
C
ac
h
e
h
it a
n
d
m
is
s
co
n
cu
r
r
en
cy
with
5
0
%
r
ea
d
r
eq
u
ests
S
y
st
e
m
s
i
z
e
16
32
64
ℎ1
ℎ2
ℎ1
ℎ2
ℎ1
ℎ2
L1
0
.
4
8
.
3
0
.
4
1
6
.
1
0
.
5
3
1
.
9
L1
2
4
.
5
4
.
2
8
8
.
5
8
.
2
1
5
.
7
1
6
.
8
1
6
.
1
3
1
.
2
L1
-
S
B
C
4
.
4
5
.
2
7
.
4
8
.
5
1
1
.
9
1
2
.
8
1
6
.
7
3
1
.
5
1
6
.
1
T
ab
le
5
.
C
ac
h
e
h
it a
n
d
m
is
s
co
n
cu
r
r
en
cy
with
8
0
%
r
ea
d
r
eq
u
ests
S
y
st
e
m
s
i
z
e
16
32
64
ℎ1
ℎ2
ℎ1
ℎ2
ℎ1
ℎ2
L1
0
.
4
8
.
3
0
.
4
1
6
.
1
0
.
5
3
1
.
9
L1
2
1
.
9
6
.
8
7
.
1
3
.
7
1
3
.
1
1
4
.
3
7
.
3
2
5
.
9
2
9
.
2
L1
-
S
B
C
2
.
0
8
.
2
6
.
4
3
.
7
1
9
.
3
1
0
.
6
6
.
9
5
0
.
9
1
6
.
1
Fig
u
r
es
5
an
d
6
s
h
o
w
th
e
h
it
an
d
m
is
s
co
n
cu
r
r
e
n
cy
f
o
r
L
1
-
SB
C
f
o
r
5
0
%
an
d
8
0
%
r
ea
d
r
eq
u
ests
r
esp
ec
tiv
ely
.
Fo
r
th
e
s
am
e
n
u
m
b
er
o
f
co
r
es,
th
e
m
is
s
co
n
cu
r
r
en
cy
d
ec
r
ea
s
es
f
o
r
L
1
-
SB
C
as
co
m
p
ar
ed
t
o
L
1
d
u
e
to
h
ig
h
e
r
h
it
c
o
n
cu
r
r
en
c
y
in
SB
C
.
T
h
e
m
is
s
co
n
c
u
r
r
en
cy
u
tili
za
tio
n
in
L
1
-
SB
C
is
ab
o
u
t
5
0
%
f
o
r
lar
g
er
n
u
m
b
er
o
f
co
r
es.
T
h
is
is
attr
ib
u
ted
to
th
e
f
ac
t
th
at
SB
C
o
f
f
er
s
h
i
g
h
er
h
it
c
o
n
cu
r
r
en
cy
y
ield
in
g
r
ed
u
ce
d
m
em
o
r
y
tr
a
f
f
ic
o
v
e
r
th
e
in
ter
c
o
n
n
ec
tio
n
.
E
v
en
t
h
o
u
g
h
th
e
lo
w
m
is
s
cu
r
r
en
cy
u
tili
za
tio
n
m
ay
s
u
g
g
es
t
th
at
th
e
n
u
m
b
er
o
f
b
u
s
es
f
o
r
h
ig
h
er
n
u
m
b
er
o
f
c
o
r
es
m
ay
b
e
r
ed
u
c
ed
f
u
r
t
h
er
,
it
m
ay
in
v
ar
iab
ly
d
ec
r
ea
s
e
th
e
h
it
r
ate
f
o
r
SB
C
d
u
e
to
lo
wer
b
an
d
wi
d
th
av
ailab
lity
th
u
s
n
u
ll
i
f
y
in
g
an
y
o
v
er
all
ad
v
an
tag
e.
As
th
e
d
ata
r
ea
d
ar
e
m
o
r
e
th
an
d
ata
wr
ites
,
SB
C
h
it
co
n
cu
r
r
en
c
y
in
cr
ea
s
es b
y
a
p
p
r
o
x
i
m
a
t
ely
1
.
5
f
o
r
th
e
s
am
e
s
y
s
tem
s
ize.
Fig
u
r
e
5
.
C
ac
h
e
h
it a
n
d
m
is
s
c
o
n
cu
r
r
en
cy
f
o
r
L
1
with
5
0
% r
ea
d
r
eq
u
ests
Fig
u
r
e
6
.
C
ac
h
e
h
it a
n
d
m
is
s
c
o
n
cu
r
r
en
cy
f
o
r
L1
with
8
0
% r
ea
d
r
eq
u
ests
1,97
3,7
6,9
8,2
19,
3
50,9
6,4
10,6
16,1
0
10
20
30
40
50
60
16
32
64
H
i
t an
d
m
i
s
s
c
o
n
c
u
rr
e
n
c
y
N
u
mb
e
r
o
f
Co
re
s
ch
1
ch
2
cm
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
I
SS
N:
2089
-
4
8
6
4
A
n
efficien
t m
u
lti
-
leve
l c
a
ch
e
s
ystem
fo
r
g
eo
metrica
lly
in
te
r
co
n
n
ec
ted
m
a
n
y
-
co
r
e
… (
Tir
u
ma
le
R
a
mesh
)
99
3
.
5
.
Co
ncurre
nt
a
v
er
a
g
e
m
emo
r
y
a
cc
ess
t
im
e
(C
-
AM
AT
)
c
y
cles
W
e
ev
alu
ated
th
e
co
n
cu
r
r
en
t
av
er
ag
e
m
em
o
r
y
ac
ce
s
s
tim
e
(C
-
AM
AT
)
cy
cles
f
r
o
m
(
3
)
an
d
(
4
)
.
T
ab
les 6
an
d
7
s
h
o
w
th
e
C
-
A
MA
T
f
o
r
5
0
% a
n
d
8
0
% r
ea
d
r
eq
u
ests
r
esp
ec
tiv
ely
.
T
ab
le
6
.
C
-
AM
AT
with
5
0
%
r
ea
d
r
eq
u
ests
C
a
c
h
e
s
y
st
e
m
16
32
64
L1
1
2
.
4
6
.
4
3
.
2
L1
2
4
.
1
2
.
2
1
.
1
L1
-
S
B
C
3
.
7
1
.
6
0
.
3
T
ab
le
7
.
C
-
AM
AT
with
8
0
%
r
ea
d
r
eq
u
ests
C
a
c
h
e
s
y
st
e
m
16
32
64
L1
1
2
.
4
6
.
4
3
.
2
L1
2
6.
5
3
.
3
1
.
6
L1
-
S
B
C
5
.
7
2
.
4
0
.
3
As
a
r
esu
lt
o
f
in
cr
ea
s
ed
SB
C
h
it
co
n
cu
r
r
en
cy
,
t
h
e
C
-
AM
AT
d
ec
r
ea
s
es
with
th
e
n
u
m
b
er
o
f
co
r
es.
Fig
u
r
e
7
s
h
o
ws
th
e
C
-
AM
AT
f
o
r
5
0
%
an
d
8
0
%
r
ea
d
r
eq
u
ests
r
esp
ec
tiv
ely
.
Fu
r
th
er
r
e
d
u
ct
io
n
in
C
-
AM
AT
is
s
ee
n
f
o
r
8
0
%
r
ea
d
r
eq
u
ests
d
u
e
to
in
cr
ea
s
e
in
SB
C
h
it c
o
n
cu
r
r
en
cy
.
Fig
u
r
e
7
.
C
-
AM
AT
with
5
0
% a
n
d
8
0
% r
ea
d
d
ata
r
eq
u
ests
3
.
6
.
Ca
che
s
y
s
t
em
t
hro
ug
ht
T
h
e
th
r
o
u
g
h
p
u
t in
GB
/s
ec
(
g
)
g
iv
en
as
(
6
)
.
=
2
.
−
+
(
6
)
W
h
er
e
b
is
th
e
n
u
m
b
er
o
f
b
u
s
es
with
2
b
y
tes
b
u
s
d
ata
wid
t
h
.
W
e
ass
u
m
ed
GB
I
b
u
s
ar
b
it
r
atio
n
an
d
b
u
s
all
o
ca
tio
n
r
ec
o
n
f
ig
u
r
atio
n
tim
e
(
tr
)
o
f
1
cy
cle
an
d
a
clo
ck
cy
cle
tim
e
o
f
0
.
5
n
s
.
T
a
b
le
8
s
u
m
m
a
r
izes
o
u
r
r
esu
lts
f
o
r
th
r
o
u
g
h
p
u
t
in
GB
p
er
s
ec
.
W
e
u
s
ed
n
o
r
m
alize
d
u
n
it
co
s
t
f
r
o
m
T
ab
le
3
an
d
C
-
AM
AT
u
s
in
g
(
3
)
an
d
(
4
)
.
As
s
h
o
wn
in
T
ab
le
8
,
th
e
th
r
o
u
g
h
p
u
t
in
cr
ea
s
es
with
t
h
e
n
u
m
b
e
r
o
f
co
r
es
an
d
r
ea
d
r
eq
u
est
p
er
ce
n
tag
e
s
u
g
g
esti
n
g
a
g
o
o
d
a
d
v
an
ta
g
e.
T
ab
le
8
.
T
h
r
o
u
g
h
p
u
t in
GB
/s
ec
f
o
r
L
1
-
SB
C
f
o
r
5
0
% a
n
d
8
0
% r
ea
d
r
eq
u
ests
N
o
.
o
f
C
o
r
e
s
16
32
64
5
0
%
8
0
%
5
0
%
8
0
%
5
0
%
8
0
%
6
.
8
4
.
8
2
4
.
6
1
8
.
8
9
8
.
5
9
5
.
5
3,7
1,6
0,3
5,7
2,4
0,34
0
2
4
6
16
32
64
C
-
AM
AT
cy
cles
N
u
m
b
e
r o
f
Core
s
50 %
re
ad
80
%
re
a
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
9
-
4
8
6
4
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
,
Vo
l.
11
,
No
.
1
,
M
ar
c
h
20
22
:
93
-
1
0
2
100
Fig
u
r
e
8
s
h
o
ws
th
e
th
r
o
u
g
h
p
u
t
f
o
r
5
0
%
an
d
8
0
%
r
ea
d
r
e
q
u
ests
r
esp
ec
tiv
ely
.
Fig
u
r
e
9
s
h
o
ws
th
e
av
er
ag
e
t
h
r
o
u
g
h
p
u
t
im
p
r
o
v
e
m
en
t
f
ac
to
r
f
o
r
L
1
2
a
n
d
L
1
-
SB
C
ca
ch
e
s
y
s
tem
s
o
v
er
L
1
ca
ch
e
s
y
s
tem
.
W
e
f
o
u
n
d
th
at
th
e
av
er
ag
e
th
r
o
u
g
h
p
u
t
im
p
r
o
v
e
m
en
t
f
ac
to
r
o
f
L
1
2
ca
c
h
e
s
y
s
tem
ac
r
o
s
s
all
s
y
s
tem
s
i
ze
s
is
1
.
5
f
o
r
5
0
%
r
ea
d
r
eq
u
ests
an
d
1
.
8
f
o
r
8
0
%
r
ea
d
r
e
q
u
ests
co
m
p
a
r
ed
to
L
1
.
W
e
d
eter
m
in
ed
th
at
th
e
av
er
ag
e
th
r
o
u
g
h
p
u
t
im
p
r
o
v
em
e
n
t
f
o
r
L
1
-
SB
C
m
em
o
r
y
s
y
s
tem
is
2
.
5
f
o
r
5
0
%
r
ea
d
r
eq
u
ests
an
d
2
.
4
with
8
0
%
r
ea
d
r
eq
u
ests
co
m
p
ar
ed
to
L
1
s
y
s
tem
.
As
t
h
er
e
is
v
er
y
n
eg
lig
ib
le
c
o
s
t
in
cr
ea
s
e
f
o
r
L
1
-
SB
C
(
0
.
5
%)
o
v
er
L
1
,
we
co
n
clu
d
e
th
at
L
1
-
SB
C
ca
ch
e
is
b
o
th
c
o
s
t
an
d
p
er
f
o
r
m
an
ce
ef
f
icien
t
co
m
p
a
r
ed
to
L
1
o
r
L
1
2
ca
c
h
e
s
y
s
tem
,
L
1
-
SB
C
o
f
f
er
s
3
0
to
6
0
% in
cr
ea
s
e
in
th
r
o
u
g
h
p
u
t im
p
r
o
v
em
en
t f
ac
to
r
co
m
p
ar
ed
to
L
1
2
im
p
r
o
v
em
e
n
t f
ac
to
r
o
v
er
L
1
.
Fig
u
r
e
8
.
L1
-
SB
C
th
r
o
u
g
h
p
u
t
in
GB
/s
ec
f
o
r
5
0
% a
n
d
8
0
%
r
ea
d
d
ata
r
e
q
u
ests
Fig
u
r
e
9.
Av
e
r
ag
e
im
p
r
o
v
e
m
e
n
t in
th
r
o
u
g
h
p
u
t f
o
r
L
1
2
an
d
L
1
-
SB
C
co
m
p
ar
ed
to
L
1
3
.
7
.
Ca
che
s
y
s
t
em
t
hro
ug
hp
ut
wit
h si
ng
le
bu
s
f
a
ult
W
e
also
r
an
s
im
u
latio
n
f
o
r
L
1
-
SB
C
with
a
s
in
g
le
b
u
s
f
au
lt
i
n
th
e
s
y
s
tem
.
W
e
u
s
ed
b
o
th
c
r
itical
an
d
non
-
cr
itical
b
u
s
f
o
r
ass
ig
n
in
g
f
au
lty
b
u
s
.
A
b
u
s
is
a
“c
r
itical
b
u
s
”
if
a
m
e
m
o
r
y
is
o
n
ly
c
o
n
n
ec
ted
to
th
at
b
u
s
.
T
y
p
ically
,
r
h
o
m
b
ic
in
ter
c
o
n
n
ec
tio
n
[
1
1
]
h
as
a
s
in
g
le
“c
r
i
ti
ca
l
b
u
s
”.
Ho
we
v
er
,
with
GB
I
[
1
2
]
,
we
p
r
o
v
id
ed
r
ed
u
n
d
an
t
b
u
s
p
ath
s
y
ield
in
g
all
b
u
s
es
“n
o
n
-
cr
itical”.
Fig
u
r
e
1
0
s
h
o
ws
th
e
p
er
ce
n
tag
e
d
e
g
r
ad
atio
n
o
f
a
s
in
g
le
b
u
s
f
au
lted
s
y
s
tem
co
m
p
a
r
e
d
to
n
o
r
m
al
L
1
-
SB
C
s
y
s
tem
with
5
0
%
r
ea
d
re
q
u
ests
.
W
e
n
o
ticed
th
at
th
e
p
er
ce
n
tag
e
d
eg
r
ad
atio
n
in
th
r
o
u
g
h
p
u
t
f
o
r
a
s
in
g
le
b
u
s
f
a
u
lt
is
less
th
an
5
%
ac
r
o
s
s
all
s
y
s
tem
s
izes
an
d
d
ec
r
ea
s
es
with
h
ig
h
er
n
u
m
b
e
r
o
f
c
o
r
es.
T
h
is
s
u
g
g
ests
a
g
o
o
d
f
au
lt
to
ler
a
n
ce
f
o
r
L
1
-
SB
C
f
o
r
in
cr
ea
s
ed
n
u
m
b
er
o
f
c
o
r
es.
F
ig
u
r
e
10
.
T
h
r
o
u
g
h
p
u
t
d
eg
r
ad
atio
n
with
a
s
in
g
le
b
u
s
f
a
u
lt
f
o
r
5
0
% r
ea
d
r
eq
u
ests
6,81
24
,62
98,46
4,78
18,82
95
,52
0
50
100
150
16
32
64
Th
ro
u
gh
p
u
t
N
u
m
b
e
r
o
f
Co
re
s
50
%
re
a
d
80 %
re
ad
1,8
1,5
2,5
2,4
0
1
2
3
L
12
a
t
50
%
r
e
a
d
L
12
a
t
80
%
r
e
a
d
L
1-S
BC at 5
0 %
r
e
ad
L
1-S
BC at 8
0 %
r
e
ad
T
h
ro
u
g
h
p
u
t
I
m
pro
v
ement
4,1
3,7
1,5
1,45
0
1
2
3
4
5
16
32
64
A
V
G
T
H
R
O
UGH
P
UT
DE
GRA
DA
T
IO
N
PE
CE
N
T
A
G
E
N
UM
BE
R
O
F
CO
R
E
S
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
I
SS
N:
2089
-
4
8
6
4
A
n
efficien
t m
u
lti
-
leve
l c
a
ch
e
s
ystem
fo
r
g
eo
metrica
lly
in
te
r
co
n
n
ec
ted
m
a
n
y
-
co
r
e
… (
Tir
u
ma
le
R
a
mesh
)
101
4.
CO
NCLU
SI
O
N
AND
F
U
T
U
RE
RE
SE
ARCH
Ma
n
y
-
co
r
e
b
ased
h
eter
o
g
en
e
o
u
s
s
y
s
tem
d
em
an
d
s
h
ig
h
s
y
s
t
em
th
r
o
u
g
h
p
u
t
f
o
r
b
ig
d
ata
a
p
p
licatio
n
s
an
d
o
t
h
er
c
o
m
p
u
te
in
ten
s
iv
e
em
b
ed
d
e
d
ap
p
licatio
n
s
.
B
y
a
d
d
in
g
a
less
ex
p
en
s
iv
e
SB
C
in
ass
o
ciatio
n
with
ex
p
en
s
iv
e
p
er
co
r
e
L
1
p
r
iv
at
e
ca
ch
e
with
in
a
m
u
lti
-
lev
el
ca
c
h
e
h
ier
ar
ch
y
,
we
ca
n
ac
h
i
ev
e
h
ig
h
er
s
y
s
tem
th
r
o
u
g
h
p
u
t.
Fo
r
b
etter
ac
cu
r
a
cy
,
we
ex
tr
ac
ted
ca
ch
e
h
it
a
n
d
m
is
s
co
n
cu
r
r
en
cies
at
ea
c
h
lev
el
an
d
ap
p
lied
co
n
cu
r
r
en
t
av
er
a
g
e
m
em
o
r
y
a
cc
ess
tim
e
f
o
r
L
1
,
L
1
2
an
d
L
1
-
SB
C
s
y
s
tem
s
.
W
e
co
n
d
u
cted
s
im
u
latio
n
o
f
L
1
,
L
1
2
an
d
L1
-
SB
C
ca
ch
e
s
y
s
tem
s
.
Ou
r
s
im
u
latio
n
r
esu
lts
in
d
icate
th
at
b
y
u
s
in
g
L
1
-
SB
C
,
we
ca
n
ac
h
iev
e
2
.
5
th
r
o
u
g
h
p
u
t
im
p
r
o
v
e
m
en
t
co
m
p
ar
ed
to
u
s
in
g
o
n
ly
L
1
p
r
iv
a
te
ca
ch
e
an
d
we
s
ee
th
at
L
1
-
SB
C
o
f
f
er
s
h
ig
h
er
in
cr
ea
s
e
in
th
r
o
u
g
h
p
u
t
im
p
r
o
v
em
en
t
f
ac
to
r
co
m
p
ar
ed
to
L
1
2
im
p
r
o
v
em
en
t f
ac
to
r
at
a
v
er
y
n
eg
lig
ib
le
in
c
r
ea
s
e
in
SB
C
co
s
t
o
v
er
L
1
.
W
e
als
o
d
eter
m
in
e
d
th
at
th
e
th
r
o
u
g
h
p
u
t
d
eg
r
ad
atio
n
u
s
in
g
L
1
-
SB
C
with
a
s
in
g
le
b
u
s
f
au
lt
is
less
th
an
5
%
ac
r
o
s
s
all
s
y
s
tem
s
izes
an
d
th
is
d
eg
r
ad
atio
n
r
ed
u
ce
s
as
th
e
s
y
s
t
em
s
ize
in
cr
ea
s
es
s
u
g
g
esti
n
g
a
g
o
o
d
a
d
v
an
tag
e
f
o
r
h
ig
h
er
n
u
m
b
er
o
f
co
r
es.
As
we
u
s
ed
th
e
SB
C
o
n
ly
d
u
r
in
g
r
ea
d
r
eq
u
est,
in
th
e
f
u
tu
r
e
,
we
h
o
p
e
to
d
e
v
elo
p
s
o
m
e
ad
d
itio
n
al
n
o
v
el
SB
C
ca
ch
e
p
r
o
t
o
co
ls
u
s
in
g
ex
clu
s
ive
an
d
s
h
a
r
ed
m
o
d
es
an
d
in
cl
u
d
e
SB
C
in
b
o
th
r
ea
d
an
d
wr
ite
cy
cles.
W
e
also
h
o
p
e
to
p
er
f
o
r
m
s
o
m
e
h
eter
o
g
e
n
o
u
s
co
m
p
u
tin
g
b
ig
d
ata
ap
p
licatio
n
b
en
ch
m
a
r
k
s
with
L
R
U
L
1
-
SB
C
s
y
s
tem
an
d
ass
ess
th
e
o
v
er
all
s
y
s
tem
p
er
f
o
r
m
an
ce
.
ACK
NO
WL
E
DG
E
M
E
NT
S
T
h
is
wo
r
k
was
s
u
p
p
o
r
ted
in
p
ar
t
b
y
Ar
m
y
R
esear
ch
Of
f
ice
HB
C
U/MSI
co
n
tr
ac
t
n
u
m
b
er
W
9
1
1
NF
-
13
-
1
-
0
1
3
3
en
titl
ed
:
“E
x
p
lo
r
in
g
Hig
h
Per
f
o
r
m
a
n
ce
Hete
r
o
g
en
eo
u
s
C
o
m
p
u
tin
g
v
ia
Har
d
war
e/So
f
twar
e
C
o
-
Desig
n
”.
RE
F
E
R
E
NC
E
S
[
1
]
S
.
Le
B
e
u
x
,
P
.
V
.
G
r
a
t
z
,
a
n
d
I
.
O
’
C
o
n
n
o
r
,
“
G
u
e
s
t
e
d
i
t
o
r
i
a
l
:
e
m
e
r
g
i
n
g
t
e
c
h
n
o
l
o
g
i
e
s
a
n
d
a
r
c
h
i
t
e
c
t
u
r
e
s
f
o
r
ma
n
y
c
o
r
e
c
o
m
p
u
t
i
n
g
p
a
r
t
1
:
h
a
r
d
w
a
r
e
t
e
c
h
n
i
q
u
e
s
,
”
I
EEE
T
r
a
n
sa
c
t
i
o
n
s
o
n
M
u
l
t
i
-
S
c
a
l
e
C
o
m
p
u
t
i
n
g
S
y
st
e
m
s
,
v
o
l
.
4
,
n
o
.
2
,
p
p
.
9
7
–
9
8
,
A
p
r
.
2
0
1
8
,
d
o
i
:
1
0
.
1
1
0
9
/
T
M
S
C
S
.
2
0
1
8
.
2
8
2
6
7
5
8
.
[
2
]
S
.
S
a
v
a
s
,
Z.
U
l
-
A
b
d
i
n
,
a
n
d
T
.
N
o
r
d
st
r
ö
m,
“
A
f
r
a
mew
o
r
k
t
o
g
e
n
e
r
a
t
e
d
o
ma
i
n
-
s
p
e
c
i
f
i
c
ma
n
y
c
o
r
e
a
r
c
h
i
t
e
c
t
u
r
e
s
f
r
o
m
d
a
t
a
f
l
o
w
p
r
o
g
r
a
ms
,
”
Mi
c
ro
p
r
o
c
e
sso
rs
a
n
d
Mi
c
ro
sys
t
e
m
s
,
v
o
l
.
7
2
,
p
.
1
0
2
9
0
8
,
F
e
b
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
m
i
c
p
r
o
.
2
0
1
9
.
1
0
2
9
0
8
.
[
3
]
J.
A
x
e
t
a
l
.
,
“
C
o
r
e
V
A
-
M
P
S
o
C
:
a
man
y
-
c
o
r
e
a
r
c
h
i
t
e
c
t
u
r
e
w
i
t
h
t
i
g
h
t
l
y
c
o
u
p
l
e
d
s
h
a
r
e
d
a
n
d
l
o
c
a
l
d
a
t
a
memo
r
i
e
s,
”
I
EE
E
T
ra
n
s
a
c
t
i
o
n
s
o
n
P
a
ra
l
l
e
l
a
n
d
D
i
s
t
ri
b
u
t
e
d
S
y
st
e
m
s
,
v
o
l
.
2
9
,
n
o
.
5
,
p
p
.
1
0
3
0
–
1
0
4
3
,
M
a
y
2
0
1
8
,
d
o
i
:
1
0
.
1
1
0
9
/
TPD
S
.
2
0
1
7
.
2
7
8
5
7
9
9
.
[
4
]
H
.
H
o
m
a
y
o
u
n
,
“
H
e
t
e
r
o
g
e
n
e
o
u
s
c
h
i
p
m
u
l
t
i
p
r
o
c
e
s
so
r
a
r
c
h
i
t
e
c
t
u
r
e
s
f
o
r
b
i
g
d
a
t
a
a
p
p
l
i
c
a
t
i
o
n
s,”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
A
C
M
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
i
n
g
Fr
o
n
t
i
e
rs
,
M
a
y
2
0
1
6
,
p
p
.
4
0
0
–
4
0
5
,
d
o
i
:
1
0
.
1
1
4
5
/
2
9
0
3
1
5
0
.
2
9
0
8
0
7
8
.
[
5
]
A
.
P
a
r
a
s
h
a
r
,
A
.
A
b
r
a
h
a
m,
D
.
C
h
a
u
d
h
a
r
y
,
a
n
d
V
.
N
.
R
a
j
e
n
d
i
r
a
n
,
“
P
r
o
c
e
s
so
r
p
i
p
e
l
i
n
i
n
g
me
t
h
o
d
f
o
r
e
f
f
i
c
i
e
n
t
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
i
n
f
e
r
e
n
c
e
o
n
e
m
b
e
d
d
e
d
d
e
v
i
c
e
s
,
”
i
n
Pro
c
e
e
d
i
n
g
s
-
2
0
2
0
I
EE
E
2
7
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
H
i
g
h
P
e
rf
o
rm
a
n
c
e
C
o
m
p
u
t
i
n
g
,
D
a
t
a
,
a
n
d
A
n
a
l
y
t
i
c
s,
H
i
PC
2
0
2
0
,
D
e
c
.
2
0
2
0
,
p
p
.
8
2
–
9
0
,
d
o
i
:
1
0
.
1
1
0
9
/
H
i
P
C
5
0
6
0
9
.
2
0
2
0
.
0
0
0
2
2
.
[
6
]
L.
C
h
e
n
g
e
t
a
l
.
,
“
A
t
e
n
s
o
r
p
r
o
c
e
ss
i
n
g
f
r
a
mew
o
r
k
f
o
r
C
P
U
-
man
y
c
o
r
e
h
e
t
e
r
o
g
e
n
e
o
u
s
s
y
st
e
ms,
”
I
E
EE
T
ra
n
s
a
c
t
i
o
n
s
o
n
C
o
m
p
u
t
e
r
-
Ai
d
e
d
D
e
s
i
g
n
o
f
I
n
t
e
g
ra
t
e
d
C
i
r
c
u
i
t
s
a
n
d
S
y
st
e
m
s
,
p
p
.
1
–
1
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
t
c
a
d
.
2
0
2
1
.
3
1
0
3
8
2
5
.
[
7
]
M
.
G
o
u
d
a
r
z
i
,
“
H
e
t
e
r
o
g
e
n
e
o
u
s
a
r
c
h
i
t
e
c
t
u
r
e
s
f
o
r
B
i
g
D
a
t
a
b
a
t
c
h
p
r
o
c
e
ssi
n
g
i
n
M
a
p
R
e
d
u
c
e
p
a
r
a
d
i
g
m,”
I
EE
E
T
r
a
n
sa
c
t
i
o
n
s
o
n
Bi
g
D
a
t
a
,
v
o
l
.
5
,
n
o
.
1
,
p
p
.
1
8
–
3
3
,
M
a
r
.
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
T
B
D
A
TA
.
2
0
1
7
.
2
7
3
6
5
5
7
.
[
8
]
E.
A
l
a
r
e
q
i
,
T.
R
a
mes
h
,
a
n
d
K
.
A
b
e
d
,
“
F
u
n
c
t
i
o
n
a
l
h
e
t
e
r
o
g
e
n
e
o
u
s
p
r
o
c
e
sso
r
a
f
f
i
n
i
t
y
c
h
a
r
a
c
t
e
r
i
z
a
t
i
o
n
t
o
B
i
g
D
a
t
a
:
t
o
w
a
r
d
s
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
,
”
i
n
2
0
1
7
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
a
t
i
o
n
a
l
S
c
i
e
n
c
e
a
n
d
C
o
m
p
u
t
a
t
i
o
n
a
l
I
n
t
e
l
l
i
g
e
n
c
e
(
C
S
C
I
)
,
D
e
c
.
2
0
1
7
,
p
p
.
1
4
3
2
–
1
4
3
6
,
d
o
i
:
1
0
.
1
1
0
9
/
C
S
C
I
.
2
0
1
7
.
2
5
0
.
[
9
]
C
.
La
i
,
X
.
S
h
i
,
a
n
d
M
.
H
u
a
n
g
,
“
Ef
f
i
c
i
e
n
t
u
t
i
l
i
z
a
t
i
o
n
o
f
mu
l
t
i
-
c
o
r
e
p
r
o
c
e
s
so
r
s
a
n
d
ma
n
y
-
c
o
r
e
c
o
-
p
r
o
c
e
ss
o
r
s
o
n
s
u
p
e
r
c
o
m
p
u
t
e
r
b
e
a
c
o
n
f
o
r
sca
l
a
b
l
e
g
e
o
c
o
mp
u
t
a
t
i
o
n
a
n
d
g
e
o
-
si
m
u
l
a
t
i
o
n
o
v
e
r
b
i
g
e
a
r
t
h
d
a
t
a
,
”
Bi
g
E
a
rt
h
D
a
t
a
,
v
o
l
.
2
,
n
o
.
1
,
p
p
.
6
5
–
8
5
,
Ja
n
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
8
0
/
2
0
9
6
4
4
7
1
.
2
0
1
8
.
1
4
3
4
2
6
5
.
[
1
0
]
T.
N
.
M
u
d
g
e
,
J.
P
.
H
a
y
e
s,
G
.
D
.
B
u
z
z
a
r
d
,
a
n
d
D
.
C
.
W
i
n
so
r
,
“
A
n
a
l
y
s
i
s
o
f
mu
l
t
i
p
l
e
-
b
u
s
i
n
t
e
r
c
o
n
n
e
c
t
i
o
n
n
e
t
w
o
r
k
s
,
”
J
o
u
r
n
a
l
o
f
Pa
r
a
l
l
e
l
a
n
d
D
i
st
r
i
b
u
t
e
d
C
o
m
p
u
t
i
n
g
,
v
o
l
.
3
,
n
o
.
3
,
p
p
.
3
2
8
–
3
4
3
,
1
9
8
6
,
d
o
i
:
1
0
.
1
0
1
6
/
0
7
4
3
-
7
3
1
5
(
8
6
)
9
0
0
1
9
-
5.
[
1
1
]
T.
R
a
mes
h
a
n
d
K
.
A
b
e
d
,
“
R
e
c
o
n
f
i
g
u
r
a
b
l
e
ma
n
y
-
c
o
r
e
e
m
b
e
d
d
e
d
c
o
m
p
u
t
i
n
g
p
l
a
t
f
o
r
m
w
i
t
h
G
e
o
met
r
i
c
a
l
b
u
s
i
n
t
e
r
c
o
n
n
e
c
t
i
o
n
,
”
i
n
Pro
c
e
e
d
i
n
g
s
-
2
0
2
0
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
a
t
i
o
n
a
l
S
c
i
e
n
c
e
a
n
d
C
o
m
p
u
t
a
t
i
o
n
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
C
S
C
I
2
0
2
0
,
D
e
c
.
2
0
2
0
,
p
p
.
1
2
5
6
–
1
2
5
9
,
d
o
i
:
1
0
.
1
1
0
9
/
C
S
C
I
5
1
8
0
0
.
2
0
2
0
.
0
0
2
3
4
.
[
1
2
]
T.
R
a
m
e
sh
a
n
d
K
.
A
b
e
d
,
“
C
o
st
-
e
f
f
i
c
i
e
n
t
r
e
c
o
n
f
i
g
u
r
a
b
l
e
g
e
o
m
e
t
r
i
c
a
l
b
u
s
i
n
t
e
r
c
o
n
n
e
c
t
i
o
n
sy
s
t
e
m
f
o
r
m
a
n
y
-
c
o
r
e
p
l
a
t
f
o
r
ms
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
R
e
c
o
n
f
i
g
u
ra
b
l
e
a
n
d
Em
b
e
d
d
e
d
S
y
s
t
e
m
s
(
I
J
RES
)
,
v
o
l
.
1
0
,
n
o
.
2
,
p
p
.
7
7
–
8
9
,
Ju
l
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
r
e
s.
v
1
0
.
i
2
.
p
p
7
7
-
8
9
.
[
1
3
]
Y
.
Li
u
,
S
.
K
a
t
o
,
a
n
d
M
.
E
d
a
h
i
r
o
,
“
A
n
a
l
y
si
s o
f
M
e
mo
r
y
S
y
st
e
m o
f
T
i
l
e
d
M
a
n
y
-
C
o
r
e
P
r
o
c
e
sso
r
s,
”
I
E
EE
A
c
c
e
ss
,
v
o
l
.
7
,
p
p
.
1
8
9
6
4
–
1
8
9
7
4
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
1
9
.
2
8
9
5
7
0
1
.
[
1
4
]
Y
.
Te
Li
n
,
Y
.
H
.
H
si
a
o
,
F
.
P
.
L
i
n
,
a
n
d
C
.
M
.
W
a
n
g
,
“
A
h
y
b
r
i
d
c
a
c
h
e
a
r
c
h
i
t
e
c
t
u
r
e
o
f
sh
a
r
e
d
mem
o
r
y
a
n
d
me
t
a
-
t
a
b
l
e
u
se
d
i
n
b
i
g
mu
l
t
i
m
e
d
i
a
q
u
e
r
y
,
”
i
n
2
0
1
6
I
EEE
/
A
C
I
S
1
5
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
c
i
e
n
c
e
,
I
C
I
S
2
0
1
6
-
Pro
c
e
e
d
i
n
g
s
,
J
u
n
.
2
0
1
6
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
I
S
.
2
0
1
6
.
7
5
5
0
8
0
9
.
[
1
5
]
S
.
C
h
a
r
l
e
s,
A
.
A
h
me
d
,
U
.
Y
.
O
g
r
a
s,
a
n
d
P
.
M
i
sh
r
a
,
“
Ef
f
i
c
i
e
n
t
c
a
c
h
e
r
e
c
o
n
f
i
g
u
r
a
t
i
o
n
u
si
n
g
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
i
n
N
o
C
-
b
a
s
e
d
ma
n
y
-
c
o
r
e
C
M
P
s
,
”
AC
M
T
ra
n
s
a
c
t
i
o
n
s
o
n
D
e
si
g
n
Au
t
o
m
a
t
i
o
n
o
f
E
l
e
c
t
r
o
n
i
c
S
y
s
t
e
m
s
,
2
0
1
9
,
d
o
i
:
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
1
4
5
/
1
1
2
2
4
4
5
.
1
1
2
2
4
5
6
.
[
1
6
]
P
.
S
a
f
a
y
e
n
i
k
o
o
,
A
.
A
s
a
d
,
a
n
d
F
.
M
o
h
a
mm
a
d
i
,
“
A
n
En
e
r
g
y
-
Ef
f
i
c
i
e
n
t
C
a
c
h
e
A
r
c
h
i
t
e
c
t
u
r
e
f
o
r
C
h
i
p
-
M
u
l
t
i
p
r
o
c
e
ss
o
r
s
B
a
se
d
o
n
N
o
n
-
U
n
i
f
o
r
m
i
t
y
A
c
c
e
ss
e
s,
”
i
n
2
0
1
8
I
E
EE
C
a
n
a
d
i
a
n
C
o
n
f
e
re
n
c
e
o
n
El
e
c
t
ri
c
a
l
&
C
o
m
p
u
t
e
r E
n
g
i
n
e
e
r
i
n
g
(
C
C
EC
E)
,
M
a
y
2
0
1
8
,
p
p
.
1
–
4
,
d
o
i
:
1
0
.
1
1
0
9
/
C
C
EC
E.
2
0
1
8
.
8
4
4
7
7
3
6
.
[
1
7
]
X
.
H
.
S
u
n
a
n
d
D
.
W
a
n
g
,
“
C
o
n
c
u
r
r
e
n
t
a
v
e
r
a
g
e
m
e
m
o
r
y
a
c
c
e
ss
t
i
m
e
,
”
C
o
m
p
u
t
e
r
,
v
o
l
.
4
7
,
n
o
.
5
,
p
p
.
7
4
–
8
0
,
M
a
y
2
0
1
4
,
d
o
i
:
1
0
.
1
1
0
9
/
M
C
.
2
0
1
3
.
2
2
7
.
[
1
8
]
J.
H
.
Le
e
,
S
.
W
.
J
e
o
n
g
,
S
.
D
.
K
i
m,
a
n
d
C
.
C
.
W
e
e
ms
,
“
A
n
i
n
t
e
l
l
i
g
e
n
t
c
a
c
h
e
s
y
st
e
m
w
i
t
h
h
a
r
d
w
a
r
e
p
r
e
f
e
t
c
h
i
n
g
f
o
r
h
i
g
h
p
e
r
f
o
r
m
a
n
c
e
,
”
I
E
EE
T
r
a
n
sa
c
t
i
o
n
s
o
n
C
o
m
p
u
t
e
rs
,
v
o
l
.
5
2
,
n
o
.
5
,
p
p
.
6
0
7
–
6
1
6
,
M
a
y
2
0
0
3
,
d
o
i
:
1
0
.
1
1
0
9
/
T
C
.
2
0
0
3
.
1
1
9
7
1
2
7
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
9
-
4
8
6
4
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
,
Vo
l.
11
,
No
.
1
,
M
ar
c
h
20
22
:
93
-
1
0
2
102
[
1
9
]
Y
.
N
i
r
a
n
j
a
n
,
S
.
T
i
w
a
r
i
,
a
n
d
R
.
G
u
p
t
a
,
“
A
v
e
r
a
g
e
mem
o
r
y
a
c
c
e
ss
t
i
m
e
r
e
d
u
c
t
i
o
n
i
n
m
u
l
t
i
l
e
v
e
l
c
a
c
h
e
o
f
p
r
o
x
y
s
e
r
v
e
r
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
3
3
rd
I
EEE
I
n
t
e
rn
a
t
i
o
n
a
l
A
d
v
a
n
c
e
C
o
m
p
u
t
i
n
g
C
o
n
f
e
re
n
c
e
,
I
A
C
C
2
0
1
3
,
F
e
b
.
2
0
1
3
,
v
o
l
.
2
0
1
3
-
F
e
b
r
u
,
p
p
.
44
–
4
7
,
d
o
i
:
1
0
.
1
1
0
9
/
I
A
d
C
C
.
2
0
1
3
.
6
5
0
6
8
1
3
.
[
2
0
]
D
.
C
h
e
n
,
H
.
Ji
n
,
X
.
Li
a
o
,
H
.
Li
u
,
R
.
G
u
o
,
a
n
d
D
.
L
i
u
,
“
M
A
LR
U
:
M
i
ss
-
p
e
n
a
l
t
y
a
w
a
r
e
LR
U
-
b
a
s
e
d
c
a
c
h
e
r
e
p
l
a
c
e
me
n
t
f
o
r
h
y
b
r
i
d
memo
r
y
sy
s
t
e
ms
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
7
D
e
si
g
n
,
A
u
t
o
m
a
t
i
o
n
a
n
d
T
e
st
i
n
E
u
r
o
p
e
,
D
ATE
2
0
1
7
,
M
a
r
.
2
0
1
7
,
p
p
.
1
0
8
6
–
1
0
9
1
,
d
o
i
:
1
0
.
2
3
9
1
9
/
D
A
TE
.
2
0
1
7
.
7
9
2
7
1
5
1
.
[
2
1
]
A
.
K
.
S
i
n
g
h
,
K
.
G
e
e
t
h
a
,
S
.
V
o
l
l
a
l
a
,
a
n
d
N
.
R
a
mas
u
b
r
a
ma
n
i
a
n
,
“
E
f
f
i
c
i
e
n
t
U
t
i
l
i
z
a
t
i
o
n
o
f
S
h
a
r
e
d
C
a
c
h
e
s
i
n
M
u
l
t
i
c
o
r
e
A
r
c
h
i
t
e
c
t
u
r
e
s,
”
Ar
a
b
i
a
n
J
o
u
rn
a
l
f
o
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
ri
n
g
,
v
o
l
.
4
1
,
n
o
.
1
2
,
p
p
.
5
1
6
9
–
5
1
7
9
,
D
e
c
.
2
0
1
6
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
3
3
6
9
-
0
1
6
-
2
1
9
7
-
0.
[
2
2
]
M
.
D
.
H
i
l
l
a
n
d
A
.
J.
S
m
i
t
h
,
“
E
v
a
l
u
a
t
i
n
g
A
sso
c
i
a
t
i
v
i
t
y
i
n
C
P
U
C
a
c
h
e
s,”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
C
o
m
p
u
t
e
rs
,
v
o
l
.
3
8
,
n
o
.
1
2
,
p
p
.
1
6
1
2
–
1
6
3
0
,
1
9
8
9
,
d
o
i
:
1
0
.
1
1
0
9
/
1
2
.
4
0
8
4
2
.
[
2
3
]
D
.
R
a
m
t
a
k
e
,
N
.
S
i
n
g
h
,
S
.
K
u
m
a
r
,
a
n
d
V
.
K
.
P
a
t
l
e
,
“
C
a
c
h
e
A
ss
o
c
i
a
t
i
v
i
t
y
A
n
a
l
y
s
i
s
o
f
M
u
l
t
i
c
o
r
e
S
y
st
e
ms,”
i
n
2
0
2
0
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
E
n
g
i
n
e
e
r
i
n
g
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
I
C
C
S
EA
2
0
2
0
,
M
a
r
.
2
0
2
0
,
p
p
.
1
–
4
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
S
EA
4
9
1
4
3
.
2
0
2
0
.
9
1
3
2
8
8
4
.
[
2
4
]
S
.
K
h
a
n
,
A
.
R
.
A
l
a
m
e
l
d
e
e
n
,
C
.
W
i
l
k
e
r
s
o
n
,
O
.
M
u
t
l
u
y
,
a
n
d
D
.
A
.
Ji
m
e
n
e
z
z
,
“
I
mp
r
o
v
i
n
g
c
a
c
h
e
p
e
r
f
o
r
m
a
n
c
e
u
si
n
g
r
e
a
d
-
w
r
i
t
e
p
a
r
t
i
t
i
o
n
i
n
g
,
”
i
n
2
0
1
4
I
E
EE
2
0
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
S
y
m
p
o
si
u
m
o
n
H
i
g
h
Pe
r
f
o
r
m
a
n
c
e
C
o
m
p
u
t
e
r
Arc
h
i
t
e
c
t
u
re
(
H
PC
A)
,
F
e
b
.
2
0
1
4
,
p
p
.
4
5
2
–
4
6
3
,
d
o
i
:
1
0
.
1
1
0
9
/
H
P
C
A
.
2
0
1
4
.
6
8
3
5
9
5
4
.
[
2
5
]
S
.
C
h
a
n
d
a
k
e
t
a
l
.
,
“
I
mp
r
o
v
e
d
r
e
a
d
/
w
r
i
t
e
c
o
s
t
t
r
a
d
e
o
f
f
i
n
D
N
A
-
b
a
s
e
d
d
a
t
a
st
o
r
a
g
e
u
si
n
g
LD
P
C
c
o
d
e
s,
”
i
n
2
0
1
9
5
7
t
h
A
n
n
u
a
l
Al
l
e
rt
o
n
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
m
u
n
i
c
a
t
i
o
n
,
C
o
n
t
r
o
l
,
a
n
d
C
o
m
p
u
t
i
n
g
,
Al
l
e
rt
o
n
2
0
1
9
,
S
e
p
.
2
0
1
9
,
p
p
.
1
4
7
–
1
5
6
,
d
o
i
:
1
0
.
1
1
0
9
/
A
LLE
R
TO
N
.
2
0
1
9
.
8
9
1
9
8
9
0
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Tirum
a
le
R
a
m
e
sh
is
c
u
rr
e
n
tl
y
s
u
p
p
o
rt
in
g
Ja
c
k
so
n
S
tate
Un
iv
e
rsity
a
s
a
n
a
d
v
a
n
c
e
d
c
o
m
p
u
ti
n
g
re
se
a
rc
h
c
o
n
su
l
tan
t
w
h
e
re
h
e
p
re
v
io
u
sly
se
rv
e
d
a
s
a
S
e
n
i
o
r
Re
se
a
rc
h
As
so
c
iate
.
His
c
u
rre
n
t
re
se
a
rc
h
i
n
tere
sts
in
c
l
u
d
e
h
e
tero
g
e
n
e
o
u
s
c
o
m
p
u
ti
n
g
,
n
e
two
r
k
-
on
-
c
h
ip
,
c
a
c
h
e
sy
ste
m
s,
a
rti
fi
c
ial
in
telli
g
e
n
c
e
(AI)/m
a
c
h
in
e
lea
rn
in
g
.
He
re
c
e
iv
e
d
h
is
BE
d
e
g
re
e
i
n
e
lec
tri
c
a
l
e
n
g
in
e
e
rin
g
fr
o
m
Ba
n
g
a
lo
re
Un
iv
e
rsity
,
In
d
ia
in
1
9
7
5
,
a
n
M
S
EE
in
VLS
I
a
re
a
fro
m
M
issis
sip
p
i
S
tate
Un
iv
e
rsit
y
in
1
9
8
3
a
n
d
th
e
P
h
.
D
d
e
g
re
e
in
c
o
m
p
u
ter
e
n
g
in
e
e
ri
n
g
fr
o
m
Oa
k
lan
d
U
n
iv
e
rsit
y
,
M
ich
i
g
a
n
i
n
1
9
9
3
.
Ra
m
e
sh
h
a
s
a
lo
n
g
-
sta
n
d
i
n
g
c
a
re
e
r.
P
re
v
io
u
sly
h
e
se
rv
e
d
a
s
a
ten
u
re
d
p
ro
fe
ss
o
r
o
f
c
o
m
p
u
ter
e
n
g
in
e
e
ri
n
g
a
t
S
a
g
i
n
a
w
Va
ll
e
y
S
tate
Un
iv
e
rsity
in
M
ich
ig
a
n
.
He
wa
s
a
c
o
rp
o
ra
te
fe
ll
o
w
f
o
r
a
d
v
a
n
c
e
d
c
o
m
p
u
ti
n
g
a
t
B
o
e
in
g
a
n
d
p
ro
v
i
d
e
d
tec
h
n
ica
l
lea
d
e
rsh
ip
f
o
r
se
v
e
ra
l
re
se
a
rc
h
p
ro
jec
ts
fu
n
d
e
d
b
y
Bo
e
i
n
g
Co
rp
o
ra
te
Re
se
a
rc
h
.
He
wa
s
a
se
n
io
r
e
n
g
in
e
e
r
a
t
IBM
.
H
e
a
lso
se
rv
e
d
a
s
a
p
r
o
fe
ss
o
rial
le
c
tu
re
r
in
th
e
d
e
p
a
rtme
n
t
o
f
e
lec
tri
c
a
l
a
n
d
c
o
m
p
u
ter
e
n
g
i
n
e
e
rin
g
a
t
G
e
o
rg
e
Was
h
i
n
g
t
o
n
Un
i
v
e
rsity
in
DC.
R
a
m
e
sh
h
a
s
n
u
m
e
ro
u
s
US
a
n
d
fo
re
i
g
n
p
a
ten
t
s
a
n
d
p
u
b
l
ish
e
d
wid
e
l
y
.
He
is
a
se
n
io
r
m
e
m
b
e
r
o
f
I
EE
E
a
n
d
h
a
s
se
rv
e
d
i
n
lea
d
e
rsh
i
p
ro
les
fo
r
IEE
E
c
o
n
fe
re
n
c
e
s
a
n
d
I
EE
E
c
o
m
p
u
ter
so
c
iety
a
n
d
re
c
e
iv
e
d
se
v
e
ra
l
p
ro
fe
ss
io
n
a
l
a
wa
rd
s.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
rjfeb
3
5
@
g
m
a
il
.
c
o
m
.
Kh
a
li
d
Ab
e
d
is
a
Ten
u
re
d
P
ro
fe
ss
o
r
in
th
e
De
p
a
rtme
n
t
o
f
El
e
c
t
rica
l
&
Co
m
p
u
ter
En
g
i
n
e
e
rin
g
a
n
d
C
o
m
p
u
ter
S
c
ie
n
c
e
a
t
Ja
c
k
so
n
S
tate
Un
i
v
e
rsity
(JSU).
His
re
se
a
rc
h
i
n
tere
sts
in
c
lu
d
e
h
i
g
h
p
e
rfo
rm
a
n
c
e
h
e
tero
g
e
n
e
o
u
s/re
c
o
n
f
ig
u
ra
b
le
c
o
m
p
u
ti
n
g
(H
P
RC/HP
HC),
e
d
g
e
c
o
m
p
u
ti
n
g
,
a
rti
f
icia
l
in
telli
g
e
n
c
e
(AI),
m
a
c
h
in
e
lea
rn
in
g
(M
L),
a
n
d
d
e
e
p
lea
rn
in
g
(DL).
He
re
c
e
iv
e
d
h
is
B.
S
.
,
M
.
S
.
,
a
n
d
P
h
.
D.
in
El
e
c
tri
c
a
l
En
g
i
n
e
e
rin
g
fr
o
m
Wr
ig
h
t
S
tate
Un
iv
e
rsit
y
i
n
1
9
9
5
,
1
9
9
6
,
2
0
0
0
,
re
sp
e
c
ti
v
e
ly
.
H
e
h
a
s p
u
b
li
s
h
e
d
e
x
t
e
n
si
v
e
ly
i
n
IE
EE
jo
u
r
n
a
ls an
d
c
o
n
fe
re
n
c
e
s
a
n
d
is
a
tec
h
n
ica
l
re
v
iew
e
r
fo
r
s
e
v
e
ra
l
IEE
E
jo
u
rn
a
ls
a
n
d
c
o
n
fe
r
e
n
c
e
s.
Dr.
Ab
e
d
is
a
S
e
n
i
o
r
M
e
m
b
e
r
o
f
IEE
E
,
I
EE
E
Co
m
p
u
t
e
r
S
o
c
iety
.
He
c
o
-
a
u
t
h
o
re
d
se
v
e
r
a
l
p
a
ten
t
su
b
m
issio
n
s
i
n
t
h
e
HPHC/HP
RC,
a
re
a
s.
He
h
a
s
re
c
e
iv
e
d
fu
n
d
i
n
g
fro
m
so
u
rc
e
s
in
c
l
u
d
in
g
t
h
e
NSF
,
t
h
e
D
o
D,
a
n
d
th
e
Arm
y
Re
se
a
rc
h
Offic
e
.
H
e
h
a
s
re
c
e
iv
e
d
a
b
o
u
t
$
3
M
in
g
ra
n
ts
fo
r
HPHC/HP
RC
e
d
u
c
a
ti
o
n
a
n
d
re
se
a
rc
h
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
k
h
a
li
d
.
a
b
e
d
@js
u
m
s.e
d
u
.
Evaluation Warning : The document was created with Spire.PDF for Python.