BESTGAMINGPCS
ALIENWARE
AURORA ALX
Although Alienware is one of the oldest
and most respected names in perfor-
mance-gaming PCs, the company has hada tough time shaking the “mainstream”
label slapped on it a couple of years ago.
To say the company is a little touchy
about the mainstream moniker is an under-
statement. As a countermeasure, Alienware
has introduced its new ALX series of
machines, which are intended to help it
reclaim any hardcore ground lost to Falcon
Northwest and Voodoo PC.
Although we hoped to see a rig with
Alienware’s upcoming dual-videocard solu-
tion, the company instead sent us an Athlon
64 FX-53 in a new proprietary water-cooled
ALX case. When we posted the machine and
saw that the Athlon 64 FX-53 had been over-
clocked from 2.4GHz to 2.6GHz, we knew the
Aurora ALX would be a formidable force.
And Alienware didn’t just overclock the CPU;
it also took the GeForce 6800 Ultra from a
stock clock of 425MHz to 450MHz.
In terms of specs alone, the Aurora out-
clocked every other Athlon 64 and GeForce
6800 Ultra in the roundup. Of course, the real
showdown occurs in our Lab’s benchmarking
tests, and in that respect the Aurora and
Athlon 64 FX didn’t disappoint. The Aurora
won no fewer than nine of our benchmark
rounds with its 2.6GHz FX. Although the
Mach V closely nips at its heels, the Aurora/
AMD combo simply outran the crowd.
The machine itself is built on an Asus
A8V board in Socket 939 trim and comes
with 1GB of DDR400 RAM, a Raptor RAID,Audigy 2 ZS, and a Plextor PX-712A and
Premium CD-RW drive.
The Aurora’s water-cooling system lets
you manually choose between three fan
speeds to meet your conditions. While it’s
very quiet on the lowest setting, the fan
churn of the highest setting made it the
loudest machine in our deathmatch.
The case itself is a slight modifica-
tion to the standard Alienware Predator
case, with a little extra headroom for the
water-cooling and blue LED (which tells
you what temperature the proc is running
at) embedded in the system’s “forehead.”
While we still love the unique look, we
weren’t too happy with our review unit. We
found the small side-cover difficult to close,
and the front-door latch broke the second
day of testing. Our ultimate chassis for
this rig would be the Predator case cast in
aluminum; while expensive, that would be
one hell of an enclosure.
Despite our kvetches, it was clear during
testing that this deathmatch would boil
down to a battle of two: Alienware vs.
Falcon Northwest. Read on to find out
which company wins.ALIENWARE AURORA ALXMA XIMUMPC verdict ......... 9
$5,000, http://www.alienware.comand most respected names in perfor-
mance-gaming PCs, the company has hadreclaim any hardcore ground lost to Falcon
Northwest and Voodoo PC.Alienware’s upcoming dual-videocard solu-
tion, the company instead sent us an Athlon
64 FX-53 in a new proprietary water-cooled
ALX case. When we posted the machine and
saw that the Athlon 64 FX-53 had been over-
clocked from 2.4GHz to 2.6GHz, we knew the
Aurora ALX would be a formidable force.
And Alienware didn’t just overclock the CPU;
it also took the GeForce 6800 Ultra from a
stock clock of 425MHz to 450MHz.clocked every other Athlon 64 and GeForce
6800 Ultra in the roundup. Of course, the real
showdown occurs in our Lab’s benchmarking
tests, and in that respect the Aurora and
Athlon 64 FX didn’t disappoint. The Aurora
won no fewer than nine of our benchmark
rounds with its 2.6GHz FX. Although the
Mach V closely nips at its heels, the Aurora/
AMD combo simply outran the crowd.A8V board in Socket 939 trim and comes
with 1GB of DDR400 RAM, a Raptor RAID,MA XIMUMPC OCTOBER 2004
Aurora’s Predator case is certainly
unique but a little under built.How We Test: Gaming Performance PCs
Because this PC shootout is all about gaming
performance, we threw out the usual suite
of benchmarks. SYSmark 2004 remained
in its jewel case and we gave both Adobe
Premiere and Photoshop the week off. We
focused our testing on games, games, and
more games.
For our tests we tapped the old standby
3DMark 2003. Although once the target of
considerable and quite controversial “opti-
mizing,” 3DMark 2003 remains a valid bench-
mark of a 3D card’s performance because,
while it leans on the CPU somewhat, it’s
mostly a GPU test. We ran it at the default
resolution of 1024x768 as well as 1600x1200
with 4x antialiasing on. We ran and reported
the CPU portion of the test as well. We also
used AquaMark 3 as a DX8.1 test; we ran
the default test, recorded the default score,
and also broke down the individual GPU and
CPU scores.
Jedi Academy continues to be a CPU-lim-
ited benchmark even at 1600x1200 resolu-
tion with 4x AA enabled. Halo is the oppositeof Jedi and mostly taps the GPU core. We
ran our standard custom demo at 1600x1200
without AA enabled. Unreal Tournament
2003 is an oldie but still quite telling as an
overall system benchmark. (You know these
boxes are fast when they can break 200fps at
1600x1200 with 4x AA on.)
We were excited to see how this batch of
PCs fared with Far Cry , the summer’s sleeper
hit, and a game that brought many a system
to tears with its heavy-duty GPU require-
ments. We ran Far Cry at 1280x1024 and at
1600x1200. For the most part, it runs fast if
you have a fat 16-pipe GPU.
Our final performance benchmark was
Doom 3 —the game people have waited
years for. As a technical tour de force, people
are building whole systems around this one
game’s system requirements and recom-
mendations. We ran the standard timedemo
at 1280x1024 and 1600x1200 with 4x AA
enabled and disabled. For the most part,
Doom 3 almost exclusively tests the GPU.
Last, we performed the first annualMaximum PC “Hot Box” test. Any high-per-
formance machine can function in the 68º F
environment of our Lab, we reasoned, but
how about in a typical house on a balmy
summer day? To find out if these systems
could handle extreme temperatures, we
locked all six in a tiny room heated to
90º F and ran CPU Burn-In , which constantly
cranks FPU intensive chores through the
chip. We thought we’d break at least one
of the overclocked systems, but after six
hours in the sweatbox, we broke the room
instead. Our demo room couldn’t handle
the power requirements of our machines
as they warmed up. Why did they last six
hours? We suspect that as the cumulative
heat increased inside the machines, the
power efficiency on the power supplies
dropped. While it’s disappointing that we
couldn’t bring any of the PCs to their knees,
it’s comforting to know that all of these rigs
can operate full-bore in 90 degree ambient
temperatures without any data corruption.ALIENWARE