<s>
adapt	O
to	O
any	O
given	O
architecture	O
and	O
program	O
automatically	O
and	O
predict	O
profitable	O
optimizations	O
to	O
improve	O
program	O
execution	O
time	O
,	O
code	O
size	O
and	O
compilation	B-Language
time	O
.	O
</s>
<s>
It	O
is	O
currently	O
used	O
and	O
supported	O
by	O
academia	O
and	O
industry	O
and	O
is	O
intended	O
to	O
open	O
up	O
research	O
opportunities	O
to	O
automate	O
compiler	B-Language
and	O
architecture	O
design	O
and	O
optimization	O
.	O
</s>
<s>
MILEPOST	B-Application
GCC	I-Application
is	O
currently	O
a	O
part	O
of	O
the	O
community-driven	O
Collective	O
Tuning	O
Initiative	O
(	O
cTuning	O
)	O
to	O
enable	O
self-tuning	O
computing	O
systems	O
based	O
on	O
collaborative	O
open-source	O
R&D	O
infrastructure	O
with	O
unified	O
interfaces	O
and	O
to	O
improve	O
the	O
quality	O
and	O
reproducibility	O
of	O
research	O
on	O
code	O
and	O
architecture	O
optimization	O
.	O
</s>
<s>
MILEPOST	B-Application
GCC	I-Application
is	O
connected	O
with	O
the	O
Collective	O
Optimization	O
Database	O
to	O
collect	O
and	O
reuse	O
profitable	O
optimization	O
cases	O
from	O
the	O
community	O
and	O
predict	O
high-quality	O
optimizations	O
based	O
on	O
statistical	O
analysis	O
of	O
past	O
optimization	O
data	O
.	O
</s>
<s>
In	O
January	O
2018	O
,	O
the	O
cTuning	O
foundation	O
and	O
the	O
Raspberry	O
Pi	O
Foundation	O
published	O
an	O
interactive	O
article	O
featuring	O
MILEPOST	B-Application
GCC	I-Application
and	O
Collective	O
Knowledge	O
framework	O
"	O
for	O
collaborative	O
research	O
into	O
multi-objective	O
autotuning	O
and	O
machine	O
learning	O
techniques.	O
"	O
</s>
