|
Installation and Drivers
Installation of the card we were supplied
was a doddle in the case we use but if this was the
size of the final retail card I can imagine lots of
people shuffling their drives around to make room
or returning the card to the store for a refund because
they just kind make it fit. Fortunately as I said,
the retail product will be shorter.
Now for the flame bait....ATi's drivers!
I am the first to admit that ATi's drivers are so
much better than previously that they're almost unrecognisable,
in fact once they're installed properly and working
they are simply excellent, but it's getting them installed
and working properly that's the problem. Throw a Radeon
on a system with a freshly installed OS and things
are superb, though quite how you can claim to have
a unified driver architecture when the core drivers
and control panel come as separate installations is
beyond me. However if you're installing a Radeon on
a system that's had a non-Radeon installed before
it and things will either go swimmingly or you'll
face hours of frustration battling messages that insist
you install a VGA driver before you're allowed to
proceed or telling you the required hardware or software
is missing despite you knowing full well it isn't.
NVIDIA drivers on the other hand are bombproof for
the most part. Slot in the card, point your OS at
the latest Detonators and that's it, job done. Apart
from a minor issue with their 40.41s I've never faced
any issues installing NVIDIA drivers no matter what
state the system or the OS I'm using is in and this
is great testament to them. ATi need to take another
look at their driver installation routine and also
find some way to incorporate the control panel into
the main drivers for the sake of simplicity.
Test Setup:
Epox 4PDA2+ i865 (Springdale) motherboard
Intel 3.06GHz (533FSB) CPU
2x256MB Corsair TwinX 3200LL
Maxtor 80GB ATA-133 HDD
SB Audigy 2 Sound
Hercules Radeon 9700 Pro graphics
Why no 9800 Pro? Simple! All three of
our "trusted sources" failed to get a card
here in time for this review so I had to use a very
capable substitute, our loyal 9700 Pro.
Performance
 |
 |
|
Best Performance
|
Shadermark
v1.7
I wanted to kick off with a look at
comparative shader performance using Tommti-Systems'
ShaderMark v1.7. This uses code and textures from
ATi's Treasure Chest demo to examine performance under
both DirectX8.0 and DirectX9.0 and using Pixel Shader
1.0, 1.1, 1.4 and 2.0. The treasure chest code isn't
optimised for ATi graphics cards and applies similarly
for all cards that use it, or at least that's what
I'm told, the results however raise several questions
about driver specific "routines" that show
how damned difficult it's getting to find a fair test
these days..
ShaderMark v1.7 - DX9 2.0 Pixel Shader
Benchmark - ToMMTi-Systems (http://www.tommti-systems.com)
video mode / device info
(1024x768), X8R8G8B8 (D24X8)
HAL (pure hw vp):
benchmark info
mip filter reflections: DX9
The green box represents the highest performance
for each test and as you can clearly see the 5900
Ultra trails the 5800 Ultra in all but two tests.
The 5800 Ultra's PS2.0 performance was never really
questioned and was always considered (by us anyway)
one of its strongest features, so we weren't expecting
any massive improvements on the 5900 Ultra. While
the fixed function routines have been very nicely
boosted it seems the less specific shader functions
are slightly slower on the 5900, no doubt due to the
75MHz reduction in core frequency. The fact that the
5900 Ultra remains in such close contention does suggest
an impreovement in overall PS2.0 performance, something
that will become increasingly vital as we see market
penetration for true DirecX9.0 games. The Radeon 9700
Pro looks to be a little off the pace in this test.
PS2.0 (DirectX9.0) Performance
|
Function
|
FX 5800U
|
FX 5900U
|
Radeon 9700 Pro
|
shaders:
Fixed Function - Gouraud Shading
Fixed Function - Gouraud Shading |
548.28 fps
|
615.62 fps
|
408.78 fps
|
shaders:
Fixed Function - Diffuse Texture Mapping
Fixed Function - Diffuse Texture Mapping
|
544.13 fps
|
602.28 fps
|
409.57 fps
|
shaders:
Fixed Function - Diffuse Bump Mapping
Fixed Function - Diffuse Bump Mapping |
361.53 fps
|
357.11 fps
|
210.83 fps
|
shaders:
PS 2.0 - Diffuse Bump Mapping
PS 2.0 - Diffuse Bump Mapping |
353.64 fps
|
347.89 fps
|
202.48 fps
|
shaders:
PS 2.0 - Diffuse Bump Mapping
PS 2.0 - Bumped Diffuse Lighting with per pixel
intensity falloff
|
277.33 fps
|
263.39 fps
|
186.25 fps
|
shaders:
PS 2.0 - Bumped Diffuse + Specular
PS 2.0 - Diffuse Bump Mapping
|
367.05 fps
|
357.70 fps
|
219.40 fps
|
shaders:
PS 2.0 - Bumped Diffuse + Specular
PS 2.0 - Bumped Diffuse Lighting with per pixel
intensity falloff
|
296.11 fps
|
281.03 fps
|
207.90 fps
|
shaders:
PS 2.0 - Bumped Diffuse and Specular Lighting
with per pixel Specular Exponent
PS 2.0 - Diffuse Bump Mapping
|
350.94 fps
|
338.37 fps
|
233.41 fps
|
shaders:
PS 2.0 - Bumped Diffuse and Specular Lighting
with per pixel Specular Exponent
PS 2.0 - Bumped Diffuse Lighting with per pixel
intensity falloff
|
285.13 fps
|
268.83 fps
|
219.26 fps
|
shaders:
PS 2.0 - Per Pixel Anisotropic Lighting
PS 2.0 - Diffuse Bump Mapping
|
370.37 fps
|
359.50 fps
|
260.61 fps
|
shaders:
PS 2.0 - Per Pixel Anisotropic Lighting
PS 2.0 - Bumped Diffuse Lighting with per pixel
intensity falloff |
298.28 fps
|
282.46 fps
|
243.71 fps
|
shaders:
PS 2.0 - Per Pixel Bumped Anisotropic Lighting
plus Diffuse
PS 2.0 - Diffuse Bump Mapping
|
312.17 fps
|
299.80 fps
|
213.98 fps
|
shaders:
PS 2.0 - Per Pixel Bumped Anisotropic Lighting
plus Diffuse
PS 2.0 - Bumped Diffuse Lighting with per pixel
intensity falloff
|
258.66 fps
|
243.52 fps
|
202.73 fps
|
shaders:
PS 2.0 - Cubic Environment Bumped Reflections
PS 2.0 - Diffuse Bump Mapping
|
229.09 fps
|
226.41 fps
|
183.72 fps
|
shaders:
PS 2.0 - Cubic Environment Bumped Reflections
PS 2.0 - Bumped Diffuse Lighting with per pixel
intensity falloff
|
201.88 fps
|
195.46 fps
|
178.53 fps
|
shaders:
PS 2.0 - Cubic Environment Bumped Diffuse and
Independently Colored Reflections
PS 2.0 - Diffuse Bump Mapping
|
207.89 fps
|
204.85 fps
|
140.07 fps
|
shaders:
PS 2.0 - Cubic Environment Bumped Diffuse and
Independently Colored Reflections
PS 2.0 - Bumped Diffuse Lighting with per pixel
intensity falloff
|
182.57 fps
|
177.07 fps
|
134.10 fps
|
shaders:
PS 2.0 - Ghost Shader
PS 2.0 - Diffuse Bump Mapping
|
218.03 fps
|
207.09 fps
|
169.64 fps
|
shaders:
PS 2.0 - Ghost Shader
PS 2.0 - Bumped Diffuse Lighting with per pixel
intensity falloff
|
186.85 fps
|
174.20 fps
|
158.66 fps
|
shaders:
PS 2.0 - Cubic Environment Bumped Diffuse and
Tinted Reflections with per pixel Fresnel Term
PS 2.0 - Diffuse Bump Mapping
|
236.09 fps
|
229.26 fps
|
185.20 fps
|
shaders:
PS 2.0 - Cubic Environment Bumped Diffuse and
Tinted Reflections with per pixel Fresnel Term
PS 2.0 - Bumped Diffuse Lighting with per pixel
intensity falloff
|
202.89 fps
|
194.04 fps
|
178.13 fps
|
shaders:
PS 2.0 - Cubic Environment Bumped Diffuse and
Independently Colored Reflections
PS 2.0 - Diffuse Bump Mapping
|
248.52 fps
|
241.70 fps
|
187.34 fps
|
shaders:
PS 2.0 - Cubic Environment Bumped Diffuse and
Independently Colored Reflections
PS 2.0 - Bumped Diffuse Lighting with per pixel
intensity falloff
|
211.59 fps
|
202.46 fps
|
177.96 fps
|
shaders:
PS 2.0 - 4 Lights/Pass Diffuse Bump Mapping
PS 2.0 - 4 Lights/Pass Diffuse Bump Mapping
|
134.52 fps
|
125.69 fps
|
95.69 fps
|
shaders:
PS 2.0 - 2 Spot Lights
PS 2.0 - 2 Spot Lights
|
90.67 fps
|
82.62 fps
|
76.52 fps
|
shaders:
PS 2.0 - Cubic Environment Bumped Diffuse and
Independently Colored Reflections
PS 2.0 - 4 Lights/Pass Diffuse Bump Mapping
|
120.97 fps
|
114.20 fps
|
99.68 fps
|
shaders:
PS 2.0 - Cubic Environment Diffuse Light and Tinted
Refractions
PS 2.0 - 4 Lights/Pass Diffuse Bump Mapping
|
129.96 fps
|
121.08 fps
|
107.98 fps
|
ShaderMark also includes the same set
of tests run using PixelShaders 1.0 to 1.4 but for
now I've decided not to include them pending a little
investigation. Results were so far skewed in favour
of the Radeon that I have to question the statement
that no ATi specific shader optimisations were used.
If after further investigation the results prove to
be accurate then I'll reinstate them here with an
accompanying statement but for now it seems to me
that the results are so far out of balance as to make
them meaningless.
|
