Neoseeker : Articles : Motherboards : Socket A : MSI KT3 Ultra ARU Motherboard Review
Hardware Newsletter:
Email:

News Headlines
New Articles

Compare Prices

Motherboards
Abit
ASUS
Gigabyte
MSI
eVGA
Intel
Tyan
More...

Processors
AMD
Intel
More...

Memory
DDR
DDR2
DDR3
More...

Video Cards
ATI
eVGA
XFX
BFG
Sapphire
More...

search for lowest prices

send article   hardware newsletter   article comments (20)
MSI KT3 Ultra ARU Motherboard Review - PAGE 3
Martin Krohn - Wednesday, April 3rd, 2002

Testing

This is our first KT333 board to arrive for testing so we currently don't have anything to compare it to. We will have some other boards coming in very shortly to compare to so stay tuned. Since the only difference between the test runs are 25% in theoretical memory bandwidth, don't expect much of a difference.

I am posting the exact memory and AGP settings that I use for all of our testing. Driver versions are also posted.

RAM settings

Cas2
Precharge 3
RAS Pulse 6
Ras to Cas 3
4 way bank interleaving
2T Command rate

AGP settings

64Mb aperture
Fast writes enabled
Other AGP read/write related items disabled

System Specs

XP 1700+
256Mb PC2700
30Gb Maxtor 7200 RPM ATA100 drive
Liteon 32x12x40 CD writer ATI Radeon 8500 WinXP
VIA 4-in1 v4.38
Radeon 6.13.10.6052
Sound disabled during testing.

Sisofts Sandra

This is a purely synthetic benchmark that many people rely too heavily upon. Synthetic benchmarks tend to show theoretical improvements a bit better than real world applications. Under these circumstances I am using the memory bandwidth benchmark that is one of many benchmarks that are a part of this program. The memory segment is based upon STREAM which in the past has been used on super computers and home computers alike. It measures sustained memory bandwidth not burst or peak. Therefore, the results may be lower than those of other benchmarks.

With our 25% increase in theoretical memory bandwidth I though it would be appropriate to see what theoretical translates to in the synthetic world.

A 2.2% increase in the Integer side of things and a 2.6% increase on the Floating point side of things are really not impressive at all!

Content Creation Winstone 2000

Content Creation Winstone is a system-level, application-based benchmark that measures a PC's overall performance when running top, Windows-based, 32-bit, content creation applications on Windows 98, Windows 2000, Windows Me, or Windows XP. Content Creation Winstone 2000 uses the following applications:

  • Adobe Photoshop 5.0
  • Adobe Premiere 5.1
  • Macromedia Director 7.0
  • Macromedia DreamWeaver 2.0
  • Netscape Navigator 4.6
  • Sonic Foundry Sound Forge 4.5

    Content Creation is the only "real world" application benchmark that is currently in my benchmark suite for using with motherboards. For the most part application benchmarks tend to vary a little bit too much for me.

    A 5.5% increase in performance from the applications listed above is a very welcome boost. Content creation does enjoy lots of memory bandwidth but it of course is not the #1 bottlekneck for these apps. I believe the bus speed and hard drive access speed are considerably more important to these types of applications.

    Quake 3

    If you need an explanation of what Quake 3 is you need a good slap. It's a GAME! What stresses a system better than a game? My reasoning for running this benchmark at such a low resolution is to truly have the memory and CPU as the bottleneck. If I were to use say 1024 or higher the video card is starting to be a limitation. We of course turned off all eye candy and use 16bit color even. It looks horrible but approaching 300 frames per second is quite a sight!

    Good old 640x480 Quake 3 stresses the CPU and memory system about as well as anything that I know.

    Only a 3.7% boost was seen in Quake 3 which further Solidifies what Sandra showed.

    3DMark 2001 SE

    Probably the only benchmark with groupies. This is a graphical benchmark that relies heavily on your CPU, Memory, and video card. The last of which is the biggest factor. Using Remedy Entertainment's acclaimed MAX-FX Technology, 3DMark2001 SE takes advantage of DirectX 8.1 and demonstrates some of the sweetest looking graphics and scenery that is possible using today's hardware. The scenes will be CRAWLING unless you have a good 3d accelerator.

    3DMark 2001 SE is a synthetic gaming environment that is used by many to determine their Direct3d performance.

    With a 1.8% increase in 3DMark It really makes you wonder where this 25% increase in memory bandwidth is going.

    Results

    Basically what we expected. The 25% increase in bandwidth brings a very small performance boost but isn't such a big improvement that you need to run out and upgrade. With the biggest improvement being only 5.5% you can't justify an upgrade from the performance standpoint alone. The CPU simply can't get the data through it's bus fast enough to process it and show a real performance boost. If you were to run your system bus at 166 you would see a considerably bigger performance boost. Unfortunately this board just can't run that kind of bus speed stable enough for me to show you.

  • next: Bottom Line »

    Article Index

    1.Introduction, Features & Specs
    2.Layout, Features & Overclocking
    3.Testing, Benchmarks & Results
    4.Bottom Line

    Submit our article to: diggDigg this! de.le.ciousdel.icio.us

    Get updates when we publish new articles
    Email Address:
    (0.0347/d/ascension)