|
We used
SPECviewperf v7.1 for our next batch of tests.
SPECviewperf is a program that tests a workstation's OpenGL
rendering performance with a variety of different "viewsets",
taken from a number of popular CAD and DCC applications,
like 3D Studio Max, Pro Engineer and Unigraphics. This
is another benchmark that usually shows significant
performance gains directly correlating with increased
overall system bandwidth. It is also affected by the
graphics card's performance. However, because we used
the same card on all of our test systems, that factor was
removed from the equation.
The SPEC.Org
website explains what this benchmark does to test system
performance:
"SPECviewperf
measures the 3D rendering performance of systems running
under the OpenGL application programming interface (API).
The benchmark's viewsets represent a similar mix of graphics
rendering and manipulation found in actual applications.
There are currently six viewsets based on popular
applications for digital content creation, CAD/CAM and
visualization."
 |
|
Benchmarks With SPECviewperf v7.1 |
|
Details: www.spec.org/gpc/opc.static/opcview71.html |
|
Surprisingly,
the Athlon 64 FX-51 didn't fare very well in the
SPECviewperf v7.1 benchmark. The FX-51 managed to pull
ahead of the standard Pentium 4 3.2GHz and Athlon XP 3200+
in ugs-03 and light-06 viewsets, but in all of the other
tests it didn't perform very well at all. Either this
benchmark prefers the Athlon XP 3200+'s shorter pipeline, or
it hasn't been optimized for AMD's new architecture as of
yet. The P4 Extreme Edition's extra 2MB of L3 cache
help it sweep many of the SPECviewperf tests, except for the
dx-08 test.
Update: It
seems like the nForce 3 may be the reason for the lower than
expected scores in this test. Some other publications
tested the FX-51 with VIA's K8T800 chipset and saw much
better scores.
|
|
|
Benchmarks With KribiBench v1.1 |
|
Details: www.adeptdevelopment.com |
|
Next up, we used
the KribiBench benchmark produced by
Adept Development. KribiBench is an SSE aware
software renderer. A 3D model is rendered and animated
by the host CPU and the average frames per second are
reported. We used two of the included models with this
benchmark; an Office model consisting of 42K polygons and an
absolutely gargantuan "Ultra" model that is comprised of
over 16G (million) polygons!
With
KribiBench's "Office" model, the Athlon 64 FX-51 was
competitive with the standard 3.2GHz P4, and much faster
than the 3200+, but the P4 3.2GHz Extreme Edition was
clearly the victor in this test. The 3.2GHz P4 EE was
24.7% faster than the standard P4, and a full 33.9% faster
than the Athlon 64 FX-51. However, KribiBench's
"Ultra" model slowed every system to a crawl. None of
the four test machines were able to break the 1FPS barrier
with the Ultra model.
|
|
|
Benchmarks With Cinebench 2003 |
|
Details: www.cinebench.com |
|
The Cinebench
2003 benchmark is an OpenGL 3D rendering performance test,
based on the commercially available Cinema 4D application.
This is a multi-threaded, multi-processor aware benchmark
that renders a single 3D scene. The time it took each
test system to render the entire scene is represented in the
graph below (listed in seconds). Cinebench 2003 also
recognizes an HT enabled P4 as either one or two CPUs.
We tested the P4s with the single-CPU and dual-CPU tests.
The Athlons could only be benchmarked with the single-CPU
test.
The Pentium 4s
outperformed the Athlons in the Cinebench 2003 benchmark by
a decent margin. The P4 EE's L3 cache didn't give it
much of a boost over the standard P4, but it didn't matter.
In the single CPU tests, AMD's Athlon 64 FX-51 fell about
6.4% short of the 3.2GHz P4, but it was about 9.6% faster
than the 3200+. With Hyperthreading enabled, running
Cinebench 2003's Dual-CPU test, gives the Pentium 4 a
significant performance edge in this benchmark.
Gaming Scores
| 
