<s>
The	O
Intel	B-Device
80387SX	I-Device
(	O
387SX	O
or	O
i387SX	O
)	O
is	O
the	O
math	B-General_Concept
coprocessor	I-General_Concept
,	O
also	O
called	O
an	O
FPU	B-General_Concept
,	O
for	O
the	O
Intel	O
80386SX	B-General_Concept
microprocessor	I-General_Concept
.	O
</s>
<s>
Introduced	O
in	O
1987	O
,	O
it	O
was	O
used	O
to	O
perform	O
floating-point	B-Algorithm
arithmetic	I-Algorithm
operations	O
directly	O
in	O
hardware	O
.	O
</s>
<s>
The	O
coprocessor	B-General_Concept
was	O
designed	O
only	O
to	O
work	O
with	O
the	O
386SX	O
,	O
rather	O
than	O
the	O
standard	O
386DX	B-General_Concept
.	O
</s>
<s>
This	O
was	O
because	O
the	O
original	O
80387	O
could	O
not	O
communicate	O
with	O
the	O
altered	O
16	O
bit	O
data	B-General_Concept
bus	I-General_Concept
of	O
the	O
386SX	O
,	O
which	O
was	O
modified	O
from	O
the	O
original	O
386DX	B-General_Concept
's	O
32	O
bit	O
data	B-General_Concept
bus	I-General_Concept
.	O
</s>
<s>
The	O
387SX	O
uses	O
a	O
68-pin	O
PLCC	B-Algorithm
socket	O
,	O
just	O
like	O
some	O
variants	O
of	O
the	O
80286	B-General_Concept
and	O
the	O
less	O
common	O
80186	B-Device
CPU	O
,	O
and	O
was	O
made	O
in	O
speeds	O
ranging	O
from	O
16	O
MHz	O
to	O
33	O
MHz	O
,	O
matching	O
the	O
clock	O
speed	O
range	O
of	O
the	O
Intel	O
manufactured	O
386SX	O
.	O
</s>
<s>
With	O
the	O
386SX/387SX	O
combination	O
,	O
it	O
provides	O
three	O
to	O
five	O
times	O
the	O
floating-point	B-Algorithm
performance	O
compared	O
to	O
these	O
80286/80287	O
combination	O
.	O
</s>
