<s>
Process	B-Operating_System
Contention	I-Operating_System
Scope	I-Operating_System
is	O
one	O
of	O
the	O
two	O
basic	O
ways	O
of	O
scheduling	O
threads	O
.	O
</s>
<s>
Many-to-Many	O
model	O
)	O
and	O
system	O
global	O
scheduling	O
(	O
known	O
as	O
System	B-Operating_System
Contention	I-Operating_System
Scope	I-Operating_System
,	O
or	O
Bound	O
Threads	O
—	O
the	O
One-to-One	O
model	O
)	O
.	O
</s>
<s>
thread	O
will	O
be	O
scheduled	O
on	O
an	O
LWP	B-Operating_System
.	O
</s>
<s>
The	O
library	O
chooses	O
which	O
unbound	O
thread	O
will	O
be	O
put	O
on	O
which	O
LWP	B-Operating_System
.	O
</s>
<s>
The	O
scheduling	O
of	O
the	O
LWP	B-Operating_System
is	O
(	O
of	O
course	O
)	O
still	O
global	O
and	O
independent	O
of	O
the	O
local	O
scheduling	O
.	O
</s>
<s>
While	O
this	O
does	O
mean	O
that	O
unbound	O
threads	O
are	O
subject	O
to	O
a	O
sort	O
of	O
funny	O
,	O
two-tiered	O
scheduling	O
architecture	O
,	O
in	O
practice	O
,	O
you	O
can	O
ignore	O
the	O
scheduling	O
of	O
the	O
LWP	B-Operating_System
and	O
deal	O
solely	O
with	O
the	O
local	O
scheduling	O
algorithm	O
.	O
</s>
<s>
In	O
that	O
case	O
,	O
the	O
lowest	O
priority	O
active	O
thread	O
(	O
T1	O
)	O
will	O
be	O
preempted	O
,	O
and	O
T2	O
will	O
take	O
its	O
place	O
on	O
the	O
LWP	B-Operating_System
.	O
</s>
<s>
Active	O
:	O
Meaning	O
that	O
it	O
is	O
on	O
an	O
LWP	B-Operating_System
5	O
.	O
</s>
<s>
It	O
will	O
remain	O
here	O
until	O
an	O
active	O
thread	O
loses	O
its	O
LWP	B-Operating_System
or	O
until	O
a	O
new	O
LWP	B-Operating_System
is	O
created	O
.	O
</s>
