March 18, 2004
testing the Athlon 64 FX-53 with another video encoding test
using Windows Media Encoder 9. In this test, we took a
416MB Digital Video file and encoded to WMV9 format.
Once again, times were recorded in minutes : seconds, and
lower times indicate better performance.
Windows Media Encoder 9
Digital Video Encoding Test
The results in
the Windows Media Encoder 9 test are a stark contrast to the
XMPEG / DivX results. With WME 9, the Athlon 64 FX-53
performed quite well, but all of the Pentium 4 based systems
posted slightly better times. The FX-53 came in 1
second behind the 3.2GHz P4C, which is within the margin of
error in this test. The 3.4GHz P4EE and 3.4GHz P4C,
however, finished encoding the video 12 and 9 seconds
quicker than the FX-53, differences of 7.5% and 5.5%
Cinebench 2003 Performance Tests
Modeling and Rendering Tests
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 and tracks the length of the
entire process. The time it took each test system to
render the entire scene is represented in the graph below
(listed in seconds). We ran two sets of numbers, one
in single-thread mode, and one in the benchmark's
multi-threaded mode for our HyperThreading enabled P4 test
systems. The Athlons are only capable of running the
single thread test, hence the "N/A"s in the graphs below.
At 78.4 seconds,
the Athlon 64 FX-53 landed right about in the middle of the
pack in Cinebench 2003's single CPU test. The 3.4GHz
P4EE and "Northwood" were slightly faster, and the 3.2GHz
P4C was a bit slower. Interestingly enough, the new
3.2GHz "Prescott" wasn't even in the running with the single
threaded test, coming in almost 20 second behind the P4 EE.
The tables tilted in heavily in favor of the Pentium 4s when
we ran the Multi-CPU tests though. HyperThreading may
not positively affect performance in every situation, but in
a multi-threaded test like this one the benefits of Intel's
Hyper-Threading technology are blatantly obvious.
Modeling and Rendering Performance
gears and take a look now at something a little more diverse
and "industrial strength" with SPECViewperf v7.1.
SPECViewperf v7.1 draws performance metrics on many
data-points in several different OpenGL based applications
from various ISVs (Independent Software Vendors). The
SPECopc (SPEC OpenGL Performance Characterization) project
group is comprised of companies like 3DLabs, Intel,
AMD, NVIDIA, ATi, Dell, IBM, SGI and Sun
Micro, as well as others. They help define and endorse
what application viewsets are used in the SPECViewperf
benchmark. Currently, there are six standard SPECopc
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
on Discreet's Lightscape radiosity application, has four
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.
The Athlon 64
FX-53 nearly swept the competition in the SPECViewperf v7.1
tests. With two of the viewsets, namely light-06 and
3dsmax-02, the 3.2GHz "Prescott", or P4E as it's now known,
pulled slightly ahead of the FX-53. However, the rest
of the viewsets belonged to the FX-53. In the ugs-03
and drv-09 tests, nothing came close to the FX-53 - it was
over 10% faster than anything Intel has at the moment.
The proe-02 and dx-08 tests were more competitive, but FX-53
still led the pack by a couple of percentage points.
Gaming Benchmarks - Wolfenstein ET, 3DMark 2003, Comanche 4