<s>
Aurora	B-Device
is	O
a	O
planned	O
supercomputer	B-Architecture
to	O
be	O
completed	O
in	O
2023	O
.	O
</s>
<s>
It	O
will	O
be	O
the	O
United	O
States	O
 '	O
second	O
exascale	B-General_Concept
computer	I-General_Concept
after	O
the	O
AMD-powered	O
Frontier-supercomputer	O
.	O
</s>
<s>
It	O
will	O
follow	O
Frontier	B-Device
,	O
which	O
was	O
the	O
world	O
's	O
first	O
exascale	B-General_Concept
supercomputer	B-Architecture
in	O
2022	O
.	O
</s>
<s>
In	O
2013	O
DOE	O
presented	O
their	O
exascale	B-General_Concept
vision	O
of	O
one	O
exaFLOP	O
at	O
20MW	O
by	O
2020	O
.	O
</s>
<s>
Aurora	B-Device
was	O
first	O
announced	O
in	O
2015	O
and	O
to	O
be	O
finished	O
in	O
2018	O
.	O
</s>
<s>
It	O
was	O
expected	O
to	O
have	O
a	O
speed	O
of	O
180	O
petaFLOPS	O
which	O
would	O
be	O
around	O
the	O
speed	O
of	O
Summit	B-Device
.	O
</s>
<s>
Aurora	B-Device
was	O
meant	O
to	O
be	O
the	O
most	O
powerful	O
supercomputer	B-Architecture
at	O
the	O
time	O
of	O
its	O
launch	O
and	O
to	O
be	O
built	O
by	O
Cray	O
with	O
Intel	O
processors	O
.	O
</s>
<s>
Later	O
,	O
in	O
2017	O
,	O
Intel	O
announced	O
that	O
Aurora	B-Device
would	O
be	O
delayed	O
to	O
2021	O
but	O
scaled	O
up	O
to	O
1	O
exaFLOP	O
.	O
</s>
<s>
In	O
March	O
2019	O
,	O
DOE	O
said	O
that	O
it	O
would	O
build	O
the	O
first	O
supercomputer	B-Architecture
with	O
a	O
performance	O
of	O
one	O
exaFLOP	O
in	O
the	O
United	O
States	O
in	O
2021	O
.	O
</s>
<s>
In	O
October	O
2020	O
,	O
DOE	O
said	O
that	O
Aurora	B-Device
would	O
be	O
delayed	O
again	O
for	O
a	O
further	O
6	O
months	O
and	O
would	O
no	O
longer	O
be	O
the	O
first	O
exascale	B-General_Concept
computer	I-General_Concept
in	O
the	O
US	O
.	O
</s>
<s>
In	O
late	O
October	O
2021	O
Intel	O
announced	O
Aurora	B-Device
would	O
now	O
exceed	O
2	O
exaFLOPS	O
in	O
peak	O
double-precision	O
compute	O
.	O
</s>
<s>
Aurora	B-Device
will	O
have	O
over	O
nine	O
thousand	O
nodes	O
,	O
with	O
each	O
node	O
being	O
composed	O
of	O
first	O
Xeon	B-Device
Max	I-Device
two	O
processors	O
,	O
six	O
Intel	O
Max	O
series	O
GPU	B-Architecture
's	O
and	O
a	O
unified	O
memory	O
architecture	O
which	O
will	O
make	O
a	O
single	O
node	O
have	O
a	O
maximum	O
computing	O
power	O
of	O
130	O
teraFLOPS	O
.	O
</s>
<s>
For	O
comparison	O
,	O
the	O
fastest	B-Operating_System
computer	I-Operating_System
in	O
the	O
world	O
today	O
,	O
Frontier	B-Device
uses	O
21MW	O
while	O
Summit	B-Device
uses	O
13MW	O
.	O
</s>
