<s>
The	O
R520	B-Device
(	O
codenamed	O
Fudo	O
)	O
is	O
a	O
graphics	B-Architecture
processing	I-Architecture
unit	I-Architecture
(	O
GPU	B-Architecture
)	O
developed	O
by	O
ATI	O
Technologies	O
and	O
produced	O
by	O
TSMC	O
.	O
</s>
<s>
It	O
was	O
the	O
first	O
GPU	B-Architecture
produced	O
using	O
a	O
90	O
nm	O
photolithography	B-Algorithm
process	O
.	O
</s>
<s>
The	O
R520	B-Device
is	O
the	O
foundation	O
for	O
a	O
line	O
of	O
DirectX	B-Application
9.0c	I-Application
and	O
OpenGL	B-Application
2.0	O
3D	B-Architecture
accelerator	I-Architecture
X1000	O
video	B-Device
cards	I-Device
.	O
</s>
<s>
It	O
is	O
ATI	O
's	O
first	O
major	O
architectural	O
overhaul	O
since	O
the	O
R300	B-Device
and	O
is	O
highly	O
optimized	O
for	O
Shader	B-Language
Model	I-Language
3.0	I-Language
.	O
</s>
<s>
The	O
Radeon	B-Device
X1000	I-Device
series	I-Device
using	O
the	O
core	O
was	O
introduced	O
on	O
October	O
5	O
,	O
2005	O
,	O
and	O
competed	O
primarily	O
against	O
Nvidia	O
's	O
GeForce	O
7000	O
series	O
.	O
</s>
<s>
ATI	O
released	O
the	O
successor	O
to	O
the	O
R500	B-Device
series	I-Device
with	O
the	O
R600	O
series	O
on	O
May	O
14	O
,	O
2007	O
.	O
</s>
<s>
ATI	O
does	O
not	O
provide	O
official	O
support	O
for	O
any	O
X1000	O
series	O
cards	O
for	O
Windows	B-Application
8	I-Application
or	O
Windows	B-Operating_System
10	I-Operating_System
;	O
the	O
last	O
AMD	B-Application
Catalyst	I-Application
for	O
this	O
generation	O
is	O
the	O
10.2	O
from	O
2010	O
up	O
to	O
Windows	B-Device
7	I-Device
.	O
</s>
<s>
AMD	O
stopped	O
providing	O
drivers	O
for	O
Windows	B-Device
7	I-Device
for	O
this	O
series	O
in	O
2015	O
.	O
</s>
<s>
A	O
series	O
of	O
open	O
source	O
Radeon	B-Device
drivers	O
are	O
available	O
when	O
using	O
a	O
Linux	B-Application
distribution	I-Application
.	O
</s>
<s>
The	O
same	O
GPUs	B-Architecture
are	O
also	O
found	O
in	O
some	O
AMD	B-Device
FireMV	I-Device
products	O
targeting	O
multi-monitor	B-General_Concept
set-ups	O
.	O
</s>
<s>
The	O
Radeon	B-Device
X1800	I-Device
video	B-Device
cards	I-Device
that	O
included	O
an	O
R520	B-Device
were	O
released	O
with	O
a	O
delay	O
of	O
several	O
months	O
because	O
ATI	O
engineers	O
discovered	O
a	O
bug	O
within	O
the	O
GPU	B-Architecture
in	O
a	O
very	O
late	O
stage	O
of	O
development	O
.	O
</s>
<s>
The	O
R520	B-Device
architecture	O
is	O
referred	O
to	O
by	O
ATI	O
as	O
an	O
"	O
Ultra	O
Threaded	B-Operating_System
Dispatch	O
Processor	O
"	O
,	O
which	O
refers	O
to	O
ATI	O
's	O
plan	O
to	O
boost	O
the	O
efficiency	O
of	O
their	O
GPU	B-Architecture
,	O
instead	O
of	O
going	O
with	O
a	O
brute	O
force	O
increase	O
in	O
the	O
number	O
of	O
processing	O
units	O
.	O
</s>
<s>
A	O
central	O
pixel	O
shader	O
"	O
dispatch	O
unit	O
"	O
breaks	O
shaders	O
down	O
into	O
threads	B-Operating_System
(	O
batches	O
)	O
of	O
16	O
pixels	O
(	O
4×4	O
)	O
and	O
can	O
track	O
and	O
distribute	O
up	O
to	O
128	O
threads	B-Operating_System
per	O
pixel	O
"	O
quad	O
"	O
(	O
4	O
pipelines	O
each	O
)	O
.	O
</s>
<s>
With	O
a	O
large	O
number	O
of	O
threads	B-Operating_System
per	O
quad	O
,	O
ATI	O
created	O
a	O
very	O
large	O
processor	B-General_Concept
register	I-General_Concept
array	O
that	O
is	O
capable	O
of	O
multiple	O
concurrent	O
reads	O
and	O
writes	O
,	O
and	O
has	O
a	O
high-bandwidth	O
connection	O
to	O
each	O
shader	O
array	O
,	O
providing	O
the	O
temporary	O
storage	O
necessary	O
to	O
keep	O
the	O
pipelines	O
fed	O
by	O
having	O
work	O
available	O
as	O
much	O
as	O
possible	O
.	O
</s>
<s>
With	O
chips	O
such	O
as	O
RV530	O
and	O
R580	B-Device
,	O
where	O
the	O
number	O
of	O
shader	O
units	O
per	O
pipeline	O
triples	O
,	O
the	O
efficiency	O
of	O
pixel	O
shading	O
drops	O
off	O
slightly	O
because	O
these	O
shaders	O
still	O
have	O
the	O
same	O
level	O
of	O
threading	O
resources	O
as	O
the	O
less	O
endowed	O
RV515	O
and	O
R520	B-Device
.	O
</s>
<s>
R420	B-Device
and	O
R300	B-Device
had	O
nearly	O
identical	O
memory	O
controller	O
designs	O
,	O
with	O
the	O
former	O
being	O
a	O
bug	O
fixed	O
release	O
designed	O
for	O
higher	O
clock	O
speeds	O
.	O
</s>
<s>
R520	B-Device
's	O
memory	O
bus	O
differs	O
with	O
its	O
central	O
controller	O
(	O
arbiter	O
)	O
that	O
connects	O
to	O
the	O
"	O
memory	O
clients	O
"	O
.	O
</s>
<s>
This	O
design	O
allows	O
memory	O
accesses	O
to	O
be	O
quicker	O
though	O
lower	O
latency	O
from	O
the	O
smaller	O
distance	O
the	O
signals	O
need	O
to	O
move	O
through	O
the	O
GPU	B-Architecture
,	O
and	O
by	O
increasing	O
the	O
number	O
of	O
banks	O
per	O
DRAM	O
.	O
</s>
<s>
The	O
vertex	O
shader	O
engines	O
were	O
already	O
at	O
the	O
required	O
FP32	B-Algorithm
precision	O
in	O
ATI	O
's	O
older	O
products	O
.	O
</s>
<s>
The	O
pixel	O
shader	O
engines	O
are	O
actually	O
quite	O
similar	O
in	O
computational	O
layout	O
to	O
their	O
R420	B-Device
counterparts	O
,	O
although	O
they	O
were	O
heavily	O
optimized	O
and	O
tweaked	O
to	O
reach	O
high	O
clock	O
speeds	O
on	O
the	O
90nm	O
process	O
.	O
</s>
<s>
Other	O
improvements	O
include	O
4096x4096	O
texture	O
support	O
and	O
ATI	O
's	O
3Dc	O
normal	B-Application
map	I-Application
compression	O
saw	O
an	O
improvement	O
in	O
compression	O
ratio	O
for	O
more	O
specific	O
situations	O
.	O
</s>
<s>
Like	O
the	O
Radeon	B-Device
cards	O
since	O
the	O
R100	O
,	O
the	O
R5xx	O
can	O
offload	O
almost	O
the	O
entire	O
MPEG-1/2	O
video	O
pipeline	O
.	O
</s>
<s>
The	O
R5xx	O
can	O
also	O
assist	O
in	O
Microsoft	O
WMV9/VC	O
-1	O
and	O
MPEG	O
H.264/AVC	B-Application
decoding	O
,	O
by	O
a	O
combination	O
of	O
the	O
3D/pipeline	O
'	O
s	O
shader-units	O
and	O
the	O
motion-video	O
engine	O
.	O
</s>
<s>
Benchmarks	O
show	O
only	O
a	O
modest	O
decrease	O
in	O
CPU-utilization	O
for	O
VC-1	B-Operating_System
and	O
H.264	B-Application
playback	O
.	O
</s>
<s>
It	O
showcased	O
a	O
highly	O
complex	O
environment	O
,	O
with	O
high-dynamic-range	B-Device
lighting	I-Device
(	O
HDR	O
)	O
and	O
dynamic	O
soft	O
shadows	O
.	O
</s>
<s>
The	O
cards	O
support	O
dual-link	B-Protocol
DVI	I-Protocol
output	O
and	O
HDCP	B-Protocol
.	O
</s>
<s>
However	O
,	O
using	O
HDCP	B-Protocol
requires	O
external	O
ROM	O
to	O
be	O
installed	O
,	O
which	O
were	O
not	O
available	O
for	O
early	O
models	O
of	O
the	O
video	B-Device
cards	I-Device
.	O
</s>
<s>
RV515	O
,	O
RV530	O
,	O
and	O
RV535	O
cores	O
include	O
a	O
single	O
and	O
a	O
double	O
DVI	B-Protocol
link	O
;	O
R520	B-Device
,	O
RV560	O
,	O
RV570	O
,	O
R580	B-Device
,	O
R580+	B-Device
cores	O
include	O
two	O
double	O
DVI	B-Protocol
links	O
.	O
</s>
<s>
AMD	O
released	O
the	O
final	O
Radeon	B-Device
R5xx	O
Acceleration	O
document	O
.	O
</s>
<s>
The	O
last	O
AMD	B-Application
Catalyst	I-Application
version	O
that	O
officially	O
supports	O
this	O
series	O
,	O
is	O
10.2	O
,	O
display	O
driver	O
version	O
8.702	O
.	O
</s>
<s>
The	O
chips	O
have	O
four	O
texture	B-General_Concept
units	I-General_Concept
,	O
four	O
ROPs	B-General_Concept
,	O
four	O
pixel	O
shaders	O
,	O
and	O
2	O
vertex	O
shaders	O
,	O
similar	O
to	O
the	O
older	O
X300	B-Device
–	I-Device
X600	I-Device
cards	O
.	O
</s>
<s>
These	O
chips	O
use	O
one	O
quad	O
of	O
an	O
R520	B-Device
,	O
whereas	O
the	O
faster	O
boards	O
use	O
just	O
more	O
of	O
these	O
quads	O
;	O
for	O
example	O
,	O
the	O
X1800	B-Device
uses	O
four	O
quads	O
.	O
</s>
<s>
Eventually	O
,	O
ATI	O
created	O
the	O
X1550	O
and	O
discontinued	O
the	O
X1300	B-Device
.	O
</s>
<s>
The	O
X1050	O
was	O
based	O
on	O
the	O
R300	B-Device
core	O
and	O
was	O
sold	O
as	O
an	O
ultra-low-budget	O
part	O
.	O
</s>
<s>
Early	O
Mobility	B-Device
Radeon	I-Device
X1300	B-Device
to	O
X1450	O
are	O
based	O
around	O
the	O
RV515	O
core	O
as	O
well	O
.	O
</s>
<s>
Beginning	O
in	O
2006	O
,	O
Radeon	B-Device
X1300	I-Device
and	O
X1550	O
products	O
were	O
shifted	O
to	O
the	O
RV505	O
core	O
,	O
which	O
had	O
similar	O
capabilities	O
and	O
features	O
as	O
the	O
previous	O
RV515	O
core	O
,	O
but	O
was	O
manufactured	O
by	O
TSMC	O
using	O
an	O
80nm	O
process	O
(	O
reduced	O
from	O
the	O
90nm	O
process	O
of	O
the	O
RV515	O
)	O
.	O
</s>
<s>
X1600	B-Device
uses	O
the	O
M56	O
core	O
which	O
is	O
based	O
on	O
the	O
RV530	O
core	O
,	O
a	O
core	O
similar	O
but	O
distinct	O
from	O
RV515	O
.	O
</s>
<s>
The	O
RV530	O
has	O
a	O
3:1	O
ratio	O
of	O
pixel	O
shaders	O
to	O
texture	B-General_Concept
units	I-General_Concept
.	O
</s>
<s>
It	O
possesses	O
12	O
pixel	O
shaders	O
while	O
retaining	O
RV515	O
's	O
four	O
texture	B-General_Concept
units	I-General_Concept
and	O
four	O
ROPs	B-General_Concept
.	O
</s>
<s>
The	O
chip	O
's	O
single	O
"	O
quad	O
"	O
has	O
3	O
pixel	O
shader	O
processors	O
per	O
pipeline	O
,	O
similar	O
to	O
the	O
design	O
of	O
R580	B-Device
's	O
4	O
quads	O
.	O
</s>
<s>
This	O
means	O
that	O
RV530	O
has	O
the	O
same	O
texturing	O
ability	O
as	O
the	O
X1300	B-Device
at	O
the	O
same	O
clock	O
speed	O
,	O
but	O
with	O
its	O
12	O
pixel	O
shaders	O
it	O
is	O
on	O
par	O
with	O
the	O
X1800	B-Device
in	O
shader	O
computational	O
performance	O
.	O
</s>
<s>
Due	O
to	O
the	O
programming	O
content	O
of	O
available	O
games	O
,	O
the	O
X1600	B-Device
is	O
greatly	O
hampered	O
by	O
lack	O
of	O
texturing	O
power	O
.	O
</s>
<s>
The	O
X1600	B-Device
was	O
positioned	O
to	O
replace	O
Radeon	B-Device
X600	I-Device
and	O
Radeon	B-Device
X700	O
as	O
ATI	O
's	O
mid-range	O
GPU	B-Architecture
.	O
</s>
<s>
The	O
Mobility	B-Device
Radeon	I-Device
X1600	B-Device
and	O
X1700	B-Device
are	O
also	O
based	O
on	O
the	O
RV530	O
.	O
</s>
<s>
The	O
X1650	O
series	O
has	O
two	O
parts	O
:	O
the	O
X1650	O
Pro	O
uses	O
the	O
RV535	O
core	O
(	O
which	O
is	O
a	O
RV530	O
core	O
manufactured	O
on	O
the	O
newer	O
80nm	O
process	O
)	O
,	O
and	O
has	O
both	O
a	O
lower	O
power	O
consumption	O
and	O
heat	O
output	O
than	O
the	O
X1600	B-Device
.	O
</s>
<s>
Originally	O
the	O
flagship	O
of	O
the	O
X1000	O
series	O
,	O
the	O
X1800	B-Device
series	O
was	O
released	O
with	O
mild	O
reception	O
due	O
to	O
the	O
rolling	O
release	O
and	O
the	O
gain	O
by	O
its	O
competitor	O
at	O
that	O
time	O
,	O
NVIDIA	O
's	O
GeForce	O
7	O
Series	O
.	O
</s>
<s>
When	O
the	O
X1800	B-Device
entered	O
the	O
market	O
in	O
late	O
2005	O
,	O
it	O
was	O
the	O
first	O
high-end	O
video	B-Device
card	I-Device
with	O
a	O
90nm	O
GPU	B-Architecture
.	O
</s>
<s>
The	O
X1800	B-Device
replaced	O
the	O
R480-based	O
Radeon	B-Device
X850	O
as	O
ATI	O
's	O
premier	O
performance	O
GPU	B-Architecture
.	O
</s>
<s>
With	O
R520	B-Device
's	O
delayed	O
release	O
,	O
its	O
competition	O
was	O
far	O
more	O
impressive	O
than	O
if	O
the	O
chip	O
had	O
made	O
its	O
originally	O
scheduled	O
spring/summer	O
release	O
.	O
</s>
<s>
Like	O
its	O
predecessor	O
,	O
the	O
X850	O
,	O
the	O
R520	B-Device
chip	O
carries	O
4	O
"	O
quads	O
"	O
,	O
which	O
means	O
it	O
has	O
similar	O
texturing	O
capability	O
at	O
the	O
same	O
clock	O
speed	O
as	O
its	O
ancestor	O
and	O
the	O
NVIDIA	O
6800	O
series	O
.	O
</s>
<s>
Unlike	O
the	O
X850	O
,	O
the	O
R520	B-Device
's	O
shader	O
units	O
are	O
vastly	O
improved	O
:	O
they	O
are	O
Shader	B-Language
Model	I-Language
3	I-Language
capable	O
,	O
and	O
received	O
some	O
advancements	O
in	O
shader	O
threading	O
that	O
can	O
greatly	O
improve	O
the	O
efficiency	O
of	O
the	O
shader	O
units	O
.	O
</s>
<s>
Unlike	O
the	O
X1900	B-Device
,	O
the	O
X1800	B-Device
has	O
16	O
pixel	O
shader	O
processors	O
and	O
equal	O
ratio	O
of	O
texturing	O
to	O
pixel	O
shading	O
capability	O
.	O
</s>
<s>
With	O
the	O
90nm	O
low-K	B-Algorithm
fabrication	O
process	O
,	O
these	O
high-transistor	O
chips	O
could	O
still	O
be	O
clocked	O
at	O
very	O
high	O
frequencies	O
,	O
which	O
allows	O
the	O
X1800	B-Device
series	O
to	O
be	O
competitive	O
with	O
GPUs	B-Architecture
with	O
more	O
pipelines	O
but	O
lower	O
clock	O
speeds	O
,	O
such	O
as	O
the	O
NVIDIA	O
7800	O
and	O
7900	O
series	O
that	O
use	O
24	O
pipelines	O
.	O
</s>
<s>
The	O
X1800	B-Device
was	O
quickly	O
replaced	O
by	O
the	O
X1900	B-Device
because	O
of	O
its	O
delayed	O
release	O
.	O
</s>
<s>
The	O
X1900	B-Device
was	O
not	O
behind	O
schedule	O
,	O
and	O
was	O
always	O
planned	O
as	O
the	O
"	O
spring	O
refresh	O
"	O
chip	O
.	O
</s>
<s>
However	O
,	O
due	O
to	O
the	O
large	O
quantity	O
of	O
unused	O
X1800	B-Device
chips	O
,	O
ATI	O
decided	O
to	O
kill	O
one	O
quad	O
of	O
pixel	O
pipelines	O
and	O
sell	O
them	O
off	O
as	O
the	O
X1800GTO	O
.	O
</s>
<s>
The	O
X1900	B-Device
and	O
X1950	O
series	O
fixed	O
several	O
flaws	O
in	O
the	O
X1800	B-Device
design	O
and	O
added	O
a	O
significant	O
pixel	O
shading	O
performance	O
boost	O
.	O
</s>
<s>
The	O
R580	B-Device
core	O
is	O
pin-compatible	O
with	O
the	O
R520	B-Device
PCBs	O
,	O
which	O
meant	O
a	O
redesign	O
of	O
the	O
X1800	B-Device
PCB	O
was	O
not	O
needed	O
.	O
</s>
<s>
The	O
primary	O
change	O
between	O
the	O
R580	B-Device
and	O
the	O
R520	B-Device
is	O
that	O
ATI	O
changed	O
the	O
pixel	O
shader	O
processor-to-texture	O
processor	O
ratio	O
.	O
</s>
<s>
The	O
X1900	B-Device
cards	O
have	O
three	O
pixel	O
shaders	O
on	O
each	O
pipeline	O
instead	O
of	O
one	O
,	O
giving	O
a	O
total	O
of	O
48	O
pixel	O
shader	O
units	O
.	O
</s>
<s>
In	O
the	O
latter	O
half	O
of	O
2006	O
,	O
ATI	O
introduced	O
the	O
Radeon	B-Device
X1950	I-Device
XTX	I-Device
,	O
which	O
is	O
a	O
graphics	B-Device
board	I-Device
using	O
a	O
revised	O
R580	B-Device
GPU	B-Architecture
called	O
R580+	B-Device
.	O
</s>
<s>
R580+	B-Device
is	O
the	O
same	O
as	O
R580	B-Device
except	O
it	O
supports	O
GDDR4	O
memory	O
,	O
a	O
new	O
graphics	O
DRAM	O
technology	O
that	O
offers	O
lower	O
power	O
consumption	O
per	O
clock	O
and	O
offers	O
a	O
significantly	O
higher	O
clock	O
rate	O
ceiling	O
.	O
</s>
<s>
The	O
X1950	O
XTX	O
clocks	O
its	O
RAM	O
at	O
1GHz	O
(	O
2GHz	O
DDR	O
)	O
,	O
providing	O
64.0	O
GB/s	O
of	O
memory	O
bandwidth	O
,	O
a	O
29%	O
advantage	O
over	O
the	O
X1900	B-Device
XTX	O
.	O
</s>
<s>
The	O
X1950	O
Pro	O
GPU	B-Architecture
is	O
built	O
off	O
of	O
the	O
80nm	O
RV570	O
core	O
with	O
only	O
12	O
texture	B-General_Concept
units	I-General_Concept
and	O
36	O
pixel	O
shaders	O
,	O
and	O
is	O
the	O
first	O
ATI	O
card	O
that	O
supports	O
native	O
Crossfire	O
implementation	O
by	O
a	O
pair	O
of	O
internal	O
Crossfire	O
connectors	O
,	O
which	O
eliminates	O
the	O
need	O
for	O
the	O
unwieldy	O
external	O
dongle	O
found	O
in	O
older	O
Crossfire	O
systems	O
.	O
</s>
