|
 |
|
HotHardware Test Systems |
|
AthlonXP All the
Way!! |
|
|
Soyo KT333
Dragon Ultra Motherboard
AMD AthlonXP
1800+
256MB Crucial PC2700 (2-4-2-6-2)CAS 2.5
2 IBM ATA100 7200RPM 42GB HD
eVGA eGeForce4 Ti4600
Creative 52X CD-ROM
Standard Floppy Drive
Windows XP Professional
Detonator 29.42
VIA 4-in-1s 4.38(2)v(a) |
A Few
Words About The Benchmarks:
To help explain the scores we achieve when testing a
product, we felt it necessary to explain how we set up a
system before running the benchmarks. With the
Soyo KT333 Dragon Ultra motherboard, we started of by
setting the BIOS to Load Optimized Defaults. The
memory frequency was manually set to 166MHz. (333MHz. DDR) and the advanced settings were set for 2.5-4-2-6-2.
We then set up the Two IBM Deskstar hard drives in a
RAID-0 configuration with the High-Point controller set
for a 16KB Stripe. Next, we installed a fresh copy
of Windows XP Professional using FAT32 and followed that
up with a download of all Critical Updates in Windows
Update, except for Messenger. We then disabled and
removed Messenger from the system, set the visual
quality to "Best Performance", defragged the RAID array,
and rebooted before starting our tests. As a
comparison for the KT333 Dragon Ultra, we included
scores from the Abit KX7-333R |
 |
|
Overclocking
the
Soyo
KT333 Dragon Ultra Motherboard |
|
Get On The Bus! |
|
With such a fine
set of overclocking settings available in the Soyo Combo
portion of the BIOS, we were anxious to see what the Dragon
Ultra could do. Before we get started, here's an
explanation of my philosophy when it comes to overclocking.
Certainly, you've seen reviews where testers report a
motherboard exceeding 160MHz. bus speeds when they overclock
their test bed. Well, there are two reasons why you
will not see that here. For one, those scores were
achieved with an "unlocked" processor, where ours are run
with a standard, "locked" processor. The other
important thing to note is that these scores are often hit
with conservative memory settings. The question is,
does it make sense to restrict the memory's performance to
get some extra CPU cycles? We feel that it is far more
beneficial to reach a balance between CPU, Memory, and bus
speeds, in order to increase the system's overall
performance, than to maximize a single component like a CPU.
With that said, we set the memory settings to their most
aggressive and began to increase the bus speed until the
system would show signs of trouble. With the Soyo
KT333 Dragon Ultra we reached a stable 146MHz. bus speed,
increasing the CPU by 149MHz. to 1679MHz., a healthy
increase of 9.7%. The memory speed also increased
nicely, up 26MHz. from 266MHz. DDR, topping out at 292MHz.,
a gain of (you guessed it) 9.7%!
Let's start out
by running a few passes of SiSoft Sandra at both the default
and overclocked settings to give us an idea of this system's
potential in the real world.
 |
|
Benchmarking with Sisoft Sandra 2002 Pro SP-1 |
|
Starting with the
Synthetic... |
|
SiSoft Sandra
2002 Professional is a great synthetic benchmarking program
that provides a comparison of a system's performance with a
large internal database. Let's take a look and see how
this system compared in the big picture.
CPU@1.53GHz.
 |
CPU@1.68GHz.
 |
The
Dragon Ultra's performance running the Athlon XP 1800+ was
on-par with what we would expect. When we increased
the bus speed to 146MHz., pushing the CPU to 1.68GHz., the
processor output was right on the heals of the Athlon XP
2200+ reference system.
|
Multimedia@1.53GHz.
 |
Multimedia@1.68GHz.
 |
The results of
the Multimedia tests maintain the same trend as the CPU
Arithmetic test, post expected scores at default speeds,
while surging into Athlon XP 2200+ territory when
overclocked.
|
Memory@1.53GHz.
 |
Memory@1.68GHz.
 |
The memory
performance clocked in slightly behind the VIA KT333
reference system while the Dragon Ultra surged past it
fairly well when overclocked.
Hard Drive - RAID
 |
The RAID scores
with the Dragon Ultra were a bit low for this system,
however, this is only a synthetic score. Let's move on
to some real word tests and see what affect this score has
when we put this board to work.
Quake
3 and MadOnion's 3DMark2001SE |