<s>
MDGRAPE-3	B-Algorithm
is	O
an	O
ultra-high	O
performance	O
petascale	B-General_Concept
supercomputer	B-Architecture
system	O
developed	O
by	O
the	O
Riken	O
research	O
institute	O
in	O
Japan	O
.	O
</s>
<s>
MDGRAPE-3	B-Algorithm
consists	O
of	O
201	O
units	O
of	O
24	O
custom	O
MDGRAPE-3	B-Algorithm
chips	O
(	O
4,824	O
total	O
)	O
,	O
plus	O
additional	O
dual-core	O
Intel	B-Device
Xeon	I-Device
processors	O
(	O
codename	O
"	O
Dempsey	O
"	O
)	O
which	O
serve	O
as	O
host	O
machines	O
.	O
</s>
<s>
In	O
June	O
2006	O
Riken	O
announced	O
its	O
completion	O
,	O
achieving	O
the	O
petaFLOPS	O
level	O
of	O
floating	B-Algorithm
point	I-Algorithm
arithmetic	I-Algorithm
performance	O
.	O
</s>
<s>
This	O
was	O
more	O
than	O
three	O
times	O
faster	O
than	O
the	O
2006	O
version	O
of	O
the	O
IBM	B-Operating_System
Blue	I-Operating_System
Gene/L	I-Operating_System
system	O
,	O
which	O
then	O
led	O
the	O
TOP500	B-Operating_System
list	B-Operating_System
of	I-Operating_System
supercomputers	I-Operating_System
at	O
0.28petaFLOPS	O
.	O
</s>
<s>
Because	O
it	O
's	O
not	O
a	O
general-purpose	O
machine	O
capable	O
of	O
running	O
the	O
LINPACK	B-Device
benchmarks	I-Device
,	O
MDGRAPE-3	B-Algorithm
does	O
not	O
qualify	O
for	O
the	O
TOP500	B-Operating_System
list	O
.	O
</s>
