|
 |
|
Gaming Benchmarks With The SB61G2 |
|
First the synthetics... |
|
For our
next test, we chose Futuremark's 3DMark 2001 SE (Build
330). It generates an overall score after
rendering scenes and measuring performance using the
MaxFX game engine, found in Remedy's popular game Max
Payne. We ran the tests on each configuration at
1024x768x32, what we would consider a "gamers" minimum
configuration, with all other program settings left at
their defaults.
The Intel
Extreme Graphics 2 engine doesn't hold a candle to the
Radeon 9500, but we really didn't expect it to.
While it does provide users with an easy alternative
for those who do not need a performance card, most if
not all gamers, should probably look elsewhere.
The two scores that we obtained when a Radeon 9500 Pro
was installed were just about equal.
Even with
all of the controversy that seems to follow it, we ran
FutureMark's follow-up to 3DMark 2001, 3DMark03, using
the latest build which also happens to be 330.
3DMark03 is similar to its older cousin, but it uses a
mixture of DX9 and DX8 pixel and vertex shaders to
assess a video card's performance. This test is
meant to be run only on "DX9 Class" hardware, but we
were able to complete at least one part of it using
the on-board graphics.
We won't
even mention the score we got with the Intel Extreme
Graphics; this was more for us to see what would
happen rather than a true reference point. As
for the rest of the test metrics, the Shuttle SB61G2
and Abit IS7 were again neck and neck, and no real
winner can be called. Early indications show the
Shuttle to be a good gaming system when paired with a
powerful video card.
|
|
|
More gaming benchmarks |
|
...and then some real-world performance |
|
Perhaps no
other gaming benchmark is as CPU dependant as
Novalogic's combat helicopter simulation, Comanche 4.
We ran Comanche 4's built-in time demo at a resolution
of 800x600x32 with audio disabled. With these
minimal settings, CPU performance and memory bandwidth
have the largest impact on performance.
Without
T&L support, the test could not be run using the Intel
Extreme Graphics 2 engine. Continuing the trend
we have seen so far, the SBG162 and IS7 are matched
frame for frame in DirectX testing. 50+ frames
in this benchmark is actually quite good, so don't be
put off if these seem low.
To get
some OpenGL benchmarks in place, we updated Quake 3
Arena with the most recent v1.32 Point Release, and
ran timedemo "Four". While it can be said that
Quake 3 is past its prime as a benchmark, we find it
is still useful for demonstrating the relative
performance of one product versus another. We
set the game to its "High Quality" mode, enabled
Trilinear filtering and maxed out the texture quality
and geometry sliders before running any tests
Hey, we at
least managed to get a score in this older benchmark
with the Intel Extreme Graphics 2 engine, pushing out
an almost playable 42.7 frames per second. Drop
off some of the texture settings or lower the
resolution and you should be able to play at a
respectable frame rate. Installing a Radeon 9500
Pro (or better) and you've got yourself a perfect LAN
party rig. We got 218.4 frames per second with
the Shuttle XPC, almost four more frames faster, than
when using the Abit IS7.
Final benchmarks
and our rating
| 
