<s>
The	O
Finite	B-Device
Element	I-Device
Machine	I-Device
(	O
FEM	O
)	O
was	O
a	O
late	O
1970s-early	O
1980s	O
NASA	O
project	O
to	O
build	O
and	O
evaluate	O
the	O
performance	O
of	O
a	O
parallel	B-Operating_System
computer	I-Operating_System
for	O
structural	B-Algorithm
analysis	I-Algorithm
.	O
</s>
<s>
The	O
motivation	O
for	O
FEM	O
arose	O
from	O
the	O
merger	O
of	O
two	O
concepts	O
:	O
the	O
finite	B-Application
element	I-Application
method	I-Application
of	O
structural	B-Algorithm
analysis	I-Algorithm
and	O
the	O
introduction	O
of	O
relatively	O
low-cost	O
microprocessors	B-Architecture
.	O
</s>
<s>
In	O
the	O
finite	B-Application
element	I-Application
method	I-Application
,	O
the	O
behavior	O
(	O
stresses	O
,	O
strains	O
and	O
displacements	O
resulting	O
from	O
load	O
conditions	O
)	O
of	O
large-scale	O
structures	O
is	O
approximated	O
by	O
a	O
FE	O
model	O
consisting	O
of	O
structural	O
elements	O
(	O
members	O
)	O
connected	O
at	O
structural	O
node	O
points	O
.	O
</s>
<s>
On	O
the	O
Finite	B-Device
Element	I-Device
Machine	I-Device
,	O
microprocessors	B-Architecture
located	O
at	O
each	O
node	O
point	O
perform	O
these	O
nodal	O
computations	O
in	O
parallel	O
.	O
</s>
<s>
If	O
there	O
are	O
more	O
node	O
points	O
(	O
N	O
)	O
than	O
microprocessors	B-Architecture
(	O
P	O
)	O
,	O
then	O
each	O
microprocessor	B-Architecture
performs	O
N/P	O
computations	O
.	O
</s>
<s>
The	O
Finite	B-Device
Element	I-Device
Machine	I-Device
contained	O
32	O
processor	O
boards	O
each	O
with	O
a	O
Texas	B-General_Concept
Instruments	I-General_Concept
TMS9900	I-General_Concept
processor	O
,	O
32	O
Input/Output	B-General_Concept
(	O
IO	O
)	O
boards	O
and	O
a	O
TMS99/4	O
controller	O
.	O
</s>
<s>
The	O
FEM	O
concept	O
was	O
first	O
successfully	O
tested	O
to	O
solve	O
beam	O
bending	O
equations	O
on	O
a	O
Langley	O
FEM	O
prototype	B-Application
(	O
4	B-Device
IMSAI	I-Device
8080s	I-Device
)	O
.	O
</s>
<s>
The	O
first	O
significant	O
Finite	B-Device
Element	I-Device
Machine	I-Device
results	O
are	O
documented	O
in	O
:	O
The	O
Finite	B-Device
Element	I-Device
Machine	I-Device
:	O
An	O
experiment	O
in	O
parallel	B-Operating_System
processing	I-Operating_System
(	O
NASA	O
TM	O
84514	O
)	O
.	O
</s>
<s>
Based	O
on	O
the	O
Finite	B-Device
Element	I-Device
Machine	I-Device
's	O
success	O
in	O
demonstrating	O
Parallel	B-Operating_System
Computing	I-Operating_System
viability	O
,	O
(	O
alongside	O
ILLIAC	B-Device
IV	I-Device
and	O
Goodyear	B-Device
MPP	I-Device
)	O
,	O
commercial	O
parallel	B-Operating_System
computers	I-Operating_System
soon	O
were	O
sold	O
.	O
</s>
<s>
NASA	O
Langley	O
subsequently	O
purchased	O
a	O
Flex/32	O
Multicomputer	B-Operating_System
(	O
and	O
later	O
Intel	B-Device
iPSC	I-Device
and	O
Intel	B-Device
Paragon	I-Device
)	O
to	O
continue	O
parallel	O
finite	B-Application
element	I-Application
algorithm	O
R&D	O
.	O
</s>
<s>
In	O
1989	O
,	O
the	O
parallel	O
equation	O
solver	O
code	O
,	O
first	O
prototyped	O
on	O
FEM	O
,	O
and	O
tested	O
on	O
FLEX	O
was	O
ported	O
to	O
NASA	O
's	O
first	O
Cray	O
YMP	O
via	O
Force	O
(	O
Fortran	O
for	O
Concurrent	B-Operating_System
Execution	O
)	O
to	O
reduce	O
the	O
structural	B-Algorithm
analysis	I-Algorithm
computation	O
time	O
for	O
the	O
space	O
shuttle	O
Challenger	O
Solid	O
Rocket	O
Booster	O
resdesign	O
with	O
54,870	O
equations	O
from	O
14	O
hours	O
to	O
6	O
seconds	O
.	O
</s>
<s>
This	O
code	O
evolved	O
into	O
NASA	O
's	O
General-Purpose	O
Solver	O
(	O
GPS	O
)	O
for	O
Matrix	O
Equations	O
used	O
in	O
numerous	O
finite	B-Application
element	I-Application
codes	O
to	O
speed	O
solution	O
time	O
.	O
</s>
