T
o judge the respective merits of the
Magnifi cent Seven, we turned to four
games and three applications. Yes, this
is a gaming roundup, but few people use
their computer strictly for games—every-
one has to edit a photo or encode a video
once in a while, so application perfor-
mance does matter. We ran all the tests
at 1600x1200 resolution—a compromise
between the 1280x1024 native resolu-
tion of standard 19-inch panels and the
1920x1200 resolution of the increasingly
popular 24-inch screens.APPLICATIONS
The three applications we chose refl ect
some of the more demanding tasks that
users today face: video editing, digital
photo editing, and media encoding. For
video editing (and encoding, to an extent),
we used our standard Adobe Premiere
Pro 2.0 test, which takes an HDV-resolu-
tion fi le, adds a ton of transitions and
video overlays, and outputs the result to
Windows Media. Ahead’s new Nero 7.5 and
the included Recode 2.0 app are also goodmeasures of a system’s video-encoding
chops. The latest revision of Recode sup-
ports H.264 encoding, so we take a feature
fi lm’s VOB from the local hard drive and
encode it for iPod video resolutions.
Both Recode 2.0 and Premiere Pro 2.
lean heavily on the CPU, so that part has
a huge bearing on performance. The same
goes for our Photoshop CS2 test, which
takes a digital RAW fi le and performs dozens
of fi lters on it. Photoshop CS2, surprisingly, is
also very disk dependent, which explains why
the RAID 0 rigs enjoyed a clear advantage
over the single-drive machines in this test.GAMING
As gaming is the primary focus of this
roundup, these tests do hold more weight
in our fi nal assessment of the rigs. FEAR
continues to be an utter GPU pig in the
latest 1.07 version. We ran the game’s
built-in performance test with all the fea-
tures, except antialiasing, set to maximum
and soft shadows enabled.
Quake 4 is primarily a CPU test. It’s
especially interesting because its multi-
threaded, so effects rendering is dividedamong a CPU’s two cores. We used a cus-
tom timedemo for our tests.
You can bag on FutureMark’s 3DMark-
series of benchmarks for being synthetic,
but at one point, the company was the
only one to offer a next-generation test
and it accurately predicted which graphics
architecture was better before there were
any real-world tests. We used the Shader
Model 3.0 test called Deep Freeze to iso-
late each machine’s graphics prowess.
The new RTS Company of Heroes is fun
to play, nice to look at, and a total system
hog. We ran the game’s built-in perfor-
mance test with the maximum settings at
1600x1200 to see if we could drag these
machines down. Unfortunately, we haven’t
been able to pinpoint exactly what the game
tests. Last month in our midrange CPU
showdown, we didn’t fi nd the benchmark to
be particularly CPU-bound, so we assumed
it would lean on the GPU. But as you’ll see
below, the results we received are a little
odd and somewhat confusing. We’re report-
ing our CoH results for all the WW2 strategy
buffs, but further analysis is needed.Speeds and Feeds
We break it all down: what our benchmarks measure and how each system performed—
plus a handy comparative chart of all the hardware that makes each system tickTHE MAGNIFICENT SEVEN 777 7777 777
MAXIMUMPC DECEMBER 2006
POLYWELL VELOCITY ALIENWARE CYBERPOWER SYSTEMAX GATEWAY OVERDRIVE ZERO POINT*BENCHMARKS
*Our current desktop zero-point rig is a Windows XP SP2 machine, using a dual-core 2.6GHz Athlon 64 FX-60, 2GB of Corsair DDR400 RAM on an Asus A8N32-SLI motherboard, two GeForce 7900 GTX videocards in SLI mode, a
Western Digital 4000KD hard drive, a Sound Blaster X-Fi soundcard, and a PC Power and Cooling Turbo Cool 850 PSU.FEAR 1.07 (FPS) 79.0 78.0 43.0 74.0 74.0 70.0 83.0 80.
QUAKE 4 1.2 (FPS) 121.7 120.1 78.7 137.8 136.4 114.4 133.7 110.3DMARK06 DEEP FREEZE (FPS) 35.4 35.3 23.5 42.9 WNR 42.8 37.7 39.
COMPANY OF HEROES (FPS) 34.5 39.0 42.7 42.7 46.5 21.3 42.1 77.ADOBE PHOTOSHOP CS2 (SEC) 186 218 227 216 190 195 202 295
ADOBE PREMIERE PRO 2.0 (SEC) 2,030 2,185 2,025 1,943 2,030 2,034 1,564 3,AHEAD NERO 7.5 RECODE 2.0 (SEC) 2,058 2,185 2,089 1,993 2,140 2,187 1,669 2,