 |
|
SPEC Viewperf |
|
A
Multi ISV Pro Graphics OpenGL Benchmark
|
|
Currently, there are six
standard SPECopc application viewsets: (courtesy
SPEC.org)
-
3dsmax-02, based on SPECapc for 3ds max 3.1 configured
with the Open GL driver, includes three models containing
an average of 1.5 million vertices each, and tests
performance of scenes with different levels of lighting.
-
dx-08, based on IBM's Data Explorer application, has
10 different tests.
-
drv-09, based on Intergraph's DesignReview model
review package, has five different tests.
-
light-06, based on Discreet's Lightscape radiosity
application, has four tests.
-
proe-02, based on SPECapc for Pro/ENGINEER 2001,
measures two models in three modes - shaded, wireframe and
hidden-line removal (HLR).
-
ugs-03, based on SPECapc for Unigraphics V17, tests
performance based on an engine model containing 2.1
million vertices.
These type
of applications typically render large data sets. All six
viewsets represent relatively high-end applications.
The end result would be the number of frames rendered/total
time which will equal frames/second.
Weighted Geometric Mean - Higher Scores Are Better
Here we see
the Quadro FX 2000 take the lead by a large margin in the
DesignReview-09 viewset, besting the ATi card by over 50%.
Functions such as triangle meshing, depth-buffering, and
smooth-shading with one light source, are where the QFX
2000 shines. The Data Explorer (dx-08) testing
proved to be more of a draw, where the graphics subsystems
are less stressed than is over all system bandwidth.
Finally, the Unigraphics Engine model (pictured above)
proved to be right within NVIDIA's element, where the QFX
2000 outpaced the FireGL X1 with a 45% margin lead.
Weighted Geometric Mean - Higher Scores Are Better
In this series
of tests, we see much of the same picture that we saw in the
other three previous viewsets. ProEngineer, with its
Sports Car model pictured above, in three different modes,
shaded wireframe and HLR (hidden line removal), proved to be
the Quadro FX 2000's strong suit by a long shot, besting the
ATi card by over 60%. Then Discreet's Lightscape
lighting radiosity performance was more of a horse-race,
with the Light-06 scores coming within 5% or so of each
other. And once again, as we have shown in our 3D
Studio Max 4.2 test in of earlier SPECapc numbers, the
Quadro FX 2000 aces out the ATi FireGL X1 by a significant
margin, in the 3D Studio Max 3.1 version test that
SPECviewperf runs in its benchmark suite.
What's
interesting to note in all of the SPECviewperf tests, is how
balanced the NVIDIA Quadro FX 2000 is across many different
applications. When it lost to the ATi card, it didn't
lose by much but when it decided to step out and take the
lead, as it did in the 3D Studio Max, Pro-E, UGS and DRV-09
tests, it did so in a big way. Again, since we're
fully aware of the hardware level capabilities of the ATi
card, with its robust R300/FGL9700 based core and strong
memory bandwidth, one can only point to drivers as the
bottleneck and ultimately what is holding the FireGL X1 back
from being all it can be.
|
|
|
Final Analysis |
|
Sizing It Up |
|
Conclusion Update - September 4, 2003
In our little
round-up we've shown you here today, there are definitely
some patterns that have developed and conclusions we can
draw from the performance measurements we took in each of
the benchmarks. First and foremost, it's obvious to
see that the NVIDIA Quadro FX 2000 is a currently a
considerably faster card than the ATi FireGL X1, as it was
able to beat out the ATi card in every single test, with the
exception of the Gaming Benchmarks and those are hardly all
that meaningful for a Pro Graphics solution. Then
their is the issue of driver optimizations for each of the
specific Pro Graphics applications we tested. You see,
unlike the Consumer Graphics space, application specific
driver optimization is a openly discussed strategy that all
major hardware OEMs practice. Additionally, since
there are far fewer of these applications to support, versus
the hundreds of thousands of PC Games on the market, it's
much easier to support and continually optimize drivers for
Pro Graphics applications and next generation graphics
hardware architectures. And here is where clearly
NVIDIA's product shines.
Although the ATi
FireGL X1 was able to show it is a very capable card, with
respect to rendering models and lighting, as it displayed
with its prowess in the gaming benchmarks, the Quadro FX
2000 muscled out the FireGL X1 in every Pro Graphics test,
due to it's stronger more configurable driver suite.
In the end, it all depends on which applications are most
important to you. If your an Engineer working with
SolidWorks 2003, then the FireGL X1 may be a decent
solution, since the ATi card is within striking distance
here. However a 3D Artist working with 3D Studio Max
should look to the Quadro FX 2000 for its strength in this
application. Additionally, the NVIDIA Quadro FX 2000
has proven self to be a slightly more balanced, well rounded
card in the Pro Graphics arena, versus the ATi FireGL X1.
On the other
hand, one aspect to consider, cost, is the same issue we
keep coming back to with Pro Graphics cards. At this
point in time, various price search engines have the ATi
FireGL X1 listed at or around $530. Conversely, the
NVIDIA Quadro FX 2000 is listed at no less that $1250.
So with this in mind, the FireGL X1 price/performance ratio
is rather compelling, at less than half the cost of the
competing NVIDIA product.
And finally, the
fight is not over yet (is it ever?). The ATi FireGL
X2-256 is right around the corner and the Quadro FX 3000 is
here now at HotHardware spinning models as we write this
article. Stick around, this should get even more
interesting in the weeks ahead.
Discuss This Article in Hardware's PC Hardware Forum!
|
