 |
|
SPECapc For SolidWorks 2003 |
|
A
3D CAD Designers Heavy Duty Tool |
|
SPECapc for
SolidWorks 2003 is a fairly new benchmark from the folks at
SPEC. Here's some background with respect to "apc" or
Application Performance Characterization tests and the
SolidWorks 2003 module in particular. The following
description of the benchmark is courtesy of
SPEC.org.
"Within SPEC's
Graphics Performance Characterization (GPC) Group there
was a strong belief that it is important to benchmark
graphics performance based on actual applications.
Application-level benchmarks exist, but they are not
standardized and they do not cover a wide range of
application areas. The Application Performance
Characterization (SPECapc) group feels that end users will
benefit from a broad-ranging set of standardized
benchmarks for graphics-intensive applications."
"SPECapc
for SolidWorks 2003?
is designed to represent a day in the life of a typical
SolidWorks 2003 user. The benchmark was developed by
SolidWorks. It is written in Visual Basic and C, and runs
on Windows 2000 and Windows XP platforms. The benchmark
uses different-sized CAD/CAM solid models, the largest of
which is an engine model with 3.13 million vertices.
Eight tests
are included within the benchmark: I/O-intensive
operations, CPU-intensive operations, and six different
graphics tests. A single number is derived from a weighted
geometric mean of the normalized score for all eight
tests. Scores are also reported for each of the eight
individual tests and for the geometric mean of the six
graphics tests."
**
SolidWorks 2003 Benchmark Update - September 4, 2003
**
Since we
initially published this article, we discovered an error
in our SolidWorks 2003 benchmark numbers. We've made
corrections to the score in the graphs represented here as
a result. The SPECapc test that we utilized for the
SolidWorks 2003 scoring has proven itself to be a bit
finicky sometimes and often times difficult to produce
repeatable results. We have since modified our test
methodology with this benchmark, in an effort to avoid
future mistakes like this. We apologize for any
inconvenience the initial scoring may have caused.
We should point
out that we took the graphics specific times for this test,
rather than the normalized weighted mean that the benchmark
likes to calculate, which includes processor and other IO
function performance. Since we're focusing on
isolating the graphics subsystem in this test, we felt it
was more relevant to work with only the time to complete the
graphics test. As you can see the Quadro FX 2000
completed the test about 10 seconds faster than the FireGL
X1, which is about a 7% advantage.
|
|
|
SPECapc For 3D Studio Max 4.2.6 |
|
A
3D Graphics Artists Toolset |
|
3D Studio Max
has long been a favored DCC application for many 3D Artists.
Here's what
SPEC says about the SPECapc test for 3D Studio Max 4.2.
"SPECapc
for 3ds max? 4.2.6 contains four scenes reflecting typical
uses of the application in an animation studio environment:
The first scene
shows an architectural visualization containing more than a
million polygons. It contains many objects, glass walls for
refraction and opacity tests, and some textures. Multiple
light sources are used.
Scene 2 is a closeup of sparks flying from wheels. Two
alternative particle systems are investigated, one with a
low particle count and the another with a high count.
Multiple light sources are used.
Scene 3
pulls back to show a juggler pedaling the wheels while
juggling three balls and balancing four teapots on top of
his head. The juggler consists of NURBS surfaces attached to
a skeleton and animated over time using inverse kinematics
chains. The scene has multiple light sources. A camera moves
along a path around the character. The camera viewport is
maximized.
Scene 4 is
a landscape modeled with displacement maps and meshes. Many
textures and texture blends are applied to it. A camera
moves along a path through the landscape.
The
benchmark is divided into 35 tasks. Each task was assigned a
weight by CAT Production representing the relative time
animators spend performing the task. The same tasks are
further characterized by SPECapc as either graphics- or
CPU-intensive. Benchmark results are presented from 25
individual graphics tests and 10 CPU tests. The total number
of seconds to run each test is normalized based on a
reference machine. The normalization process ensures a
scoring system where a bigger score is better."
At default 3DS
Max Driver settings, the ATi FireGL X1 takes the lead
handily once again. However, when we enabled NVIDIA's
custom "MAXtreme" driver for this application, when
configured to the "Speed" setting, rendering times we
dramatically reduced and the overall Graphics Mean score
shot up three fold and at the "Quality" setting it more than
double the score. The MAXtreme driver configuration
tool essentially gives users control over various image
quality and rendering settings in 3D Studio Max and allows
for dramatically reduced workloads, if desired.
Here's a quick
shot of the MAXtreme Driver Control Panel settings we used
for the scores indicated in the green and purple bar graphs
in the above chart.
NVIDIA
MAXtreme
Driver Settings Used
|
Speed Settings
|
Quality Settings
|
As you can see
in the "Speed" control panel, texture filtering settings
have been optimized for "fastest" performance rather than
quality, as well as Wireframe Views and Texture Compression.
The kicker here is that once you've got the desired general
end result of your work, you can always go back and then
adjust settings for quality. Hats off to NVIDIA for
providing an extra level of driver configurability to 3DS
Max users, which will ultimately cut rendering workloads to
a fraction, when need be.
Gaming Performance? - UT2003 and Q3
| 
