<s>
Mira	B-Device
is	O
a	O
retired	O
petascale	B-General_Concept
Blue	B-General_Concept
Gene/Q	I-General_Concept
supercomputer	B-Architecture
.	O
</s>
<s>
As	O
of	O
November	O
2017	O
,	O
it	O
is	O
listed	O
on	O
TOP500	B-Operating_System
as	O
the	O
11th	O
fastest	B-Operating_System
supercomputer	I-Operating_System
in	O
the	O
world	O
,	O
while	O
it	O
debuted	O
June	O
2012	O
in	O
3rd	O
place	O
.	O
</s>
<s>
It	O
has	O
a	O
performance	O
of	O
8.59	O
petaflops	O
(	O
LINPACK	B-Application
)	O
and	O
consumes	O
3.9	O
MW	O
.	O
</s>
<s>
The	O
supercomputer	B-Architecture
was	O
constructed	O
by	O
IBM	O
for	O
Argonne	O
National	O
Laboratory	O
's	O
Argonne	O
Leadership	O
Computing	O
Facility	O
with	O
the	O
support	O
of	O
the	O
United	O
States	O
Department	O
of	O
Energy	O
,	O
and	O
partially	O
funded	O
by	O
the	O
National	O
Science	O
Foundation	O
.	O
</s>
<s>
Mira	B-Device
was	O
used	O
for	O
scientific	O
research	O
,	O
including	O
studies	O
in	O
the	O
fields	O
of	O
material	O
science	O
,	O
climatology	O
,	O
seismology	O
,	O
and	O
computational	O
chemistry	O
.	O
</s>
<s>
The	O
supercomputer	B-Architecture
was	O
used	O
initially	O
for	O
sixteen	O
projects	O
selected	O
by	O
the	O
Department	O
of	O
Energy	O
.	O
</s>
<s>
The	O
Argonne	O
Leadership	O
Computing	O
Facility	O
,	O
which	O
commissioned	O
the	O
supercomputer	B-Architecture
,	O
was	O
established	O
by	O
the	O
America	O
COMPETES	O
Act	O
,	O
signed	O
by	O
President	O
Bush	O
in	O
2007	O
,	O
and	O
President	O
Obama	O
in	O
2011	O
.	O
</s>
<s>
The	O
United	O
States	O
 '	O
emphasis	O
on	O
supercomputing	B-Architecture
was	O
seen	O
as	O
a	O
response	O
to	O
China	O
's	O
progress	O
in	O
the	O
field	O
.	O
</s>
<s>
China	O
's	O
Tianhe-1A	O
,	O
located	O
at	O
the	O
Tianjin	O
National	O
Supercomputer	B-Architecture
Center	O
,	O
was	O
ranked	O
the	O
most	O
powerful	O
supercomputer	B-Architecture
in	O
the	O
world	O
from	O
October	O
2010	O
to	O
June	O
2011	O
.	O
</s>
<s>
Mira	B-Device
is	O
,	O
along	O
with	O
IBM	B-Device
Sequoia	I-Device
and	O
Blue	B-Device
Waters	I-Device
,	O
one	O
of	O
three	O
American	O
petascale	B-General_Concept
supercomputers	B-Architecture
deployed	O
in	O
2012	O
.	O
</s>
<s>
The	O
cost	O
for	O
building	O
Mira	B-Device
has	O
not	O
been	O
released	O
by	O
IBM	O
.	O
</s>
<s>
Early	O
reports	O
estimated	O
that	O
construction	O
would	O
cost	O
US$50	O
million	O
,	O
and	O
Argonne	O
National	O
Laboratory	O
announced	O
that	O
Mira	B-Device
was	O
bought	O
using	O
money	O
from	O
a	O
grant	O
of	O
US$180	O
million	O
.	O
</s>
<s>
In	O
a	O
press	O
release	O
,	O
IBM	O
marketed	O
the	O
supercomputer	B-Architecture
's	O
speed	O
,	O
claiming	O
that	O
"	O
if	O
every	O
man	O
,	O
woman	O
and	O
child	O
in	O
the	O
United	O
States	O
performed	O
one	O
calculation	O
each	O
second	O
,	O
it	O
would	O
take	O
them	O
almost	O
a	O
year	O
to	O
do	O
as	O
many	O
calculations	O
as	O
Mira	B-Device
will	O
do	O
in	O
one	O
second	O
"	O
.	O
</s>
<s>
"	O
Argonne	O
scientists	O
used	O
Mira	B-Device
to	O
identify	O
and	O
improve	O
a	O
new	O
mechanism	O
for	O
eliminating	O
friction	O
,	O
which	O
fed	O
into	O
the	O
development	O
of	O
a	O
hybrid	O
material	O
that	O
exhibited	O
superlubricity	O
at	O
the	O
macroscale	O
for	O
the	O
first	O
time	O
 [ .. ] 	O
simulating	O
up	O
to	O
1.2	O
million	O
atoms	O
for	O
dry	O
environments	O
and	O
up	O
to	O
10	O
million	O
atoms	O
for	O
humid	O
environments	O
 [ .. ] 	O
The	O
researchers	O
used	O
the	O
LAMMPS	B-Application
(	O
Large-scale	B-Application
Atomic/Molecular	I-Application
Massively	I-Application
Parallel	I-Application
Simulator	I-Application
)	O
code	O
to	O
carry	O
out	O
the	O
computationally	O
demanding	O
reactive	O
molecular	O
dynamics	O
simulations	O
.	O
</s>
<s>
[	O
..	O
A ]	O
team	O
of	O
computational	O
scientists	O
 [ .. ] 	O
were	O
able	O
to	O
overcome	O
a	O
performance	O
bottleneck	O
with	O
the	O
code	O
's	O
ReaxFF	B-Algorithm
module	O
,	O
an	O
add-on	O
package	O
that	O
was	O
needed	O
to	O
model	O
the	O
chemical	O
reactions	O
occurring	O
in	O
the	O
system	O
.	O
</s>
<s>
The	O
team ]	O
optimized	O
LAMMPS	B-Application
and	O
its	O
implementation	O
of	O
ReaxFF	B-Algorithm
by	O
adding	O
OpenMP	B-Application
threading	O
,	O
replacing	O
MPI	B-Application
point-to-point	O
communication	O
with	O
MPI	B-Application
collectives	O
in	O
key	O
algorithms	O
,	O
and	O
leveraging	O
MPI	B-Application
I/O	O
.	O
</s>
