|
Page 3 of 4
So, how about the benchmarks? The assembled system used for testing was as follows:
|
BIOS
|
|
5th Bios revision 1005.001 |
|
CPU
|
|
AXMH2500FQQ4C @ 11x200 |
|
Memory
|
|
2x256mb Kingston HyperX PC3200 |
|
Graphics
|
|
ASUS A9800PRO 256/TVD |
|
Graphics driver
|
|
Beta Catalyst 8.07 AGP drivers |
|
Hard disk
|
|
60GB Western Digital SE 8MB |
I
happened upon a nice utility called Bench’em All. It nicely runs a lot
of games in benchmark mode such as Far Cry, UT2004, Doom3, and Halo
just to mention a few. It will also run 3DMark01SE, 3DMark03, Aquamark,
and SPECViewperf 6, 7 and 7.1. Unfortunately I don’t have access to all
those programs, but all I had to do was click one button, and it tried
them all automatically. It did not miss any either, but I have
everything installed to default locations. If you do not, however,
there is an option to map the executables. Bench’EmAll was great in
that all you do is one click and walk away….sweet lazy heaven!
So, how did it do? Well, not too bad, but not up to Nforce standards.
Memory bandwidth and CPU scores were measured with the latest SiSoft
Sandra 2004.8.9.131.

Sanrda CPU Arithmetic Benchmark

Sandra CPU Multimedia Benchmark
As
you can see from the scores above, they are below the standards listed
in Sandra. The exact same CPU and memory scores about 3% higher than
the standard on a GA-7n400 Pro2, which is an nForce2 Ultra 400 based.

Sandra Memory Bandwidth Benchmark
The
ASUS board performed about the same performance-wise, in the memory
bandwidth tests, which is not all too surprising. Seems like VIA has
got it real close, but no cigars to be smoked here today!

Sandra Filesystem Benchmark
But
what's this? 10MB per second faster than the standard for the same
drive? You better believe it! VIA has a double access IDE channel, that
they call VLINK. Works wonders, but it uses almost twice the CPU time.
Good for audio/video rendering, but not quite so for games. This
implementation actually leads to faster load times to Windows or
whatever OS you happen to choose, and programs load faster too.
Although it pushes the CPU more, I actually don't mind so much, as
those long waits between loading levels in some games are quite a bit
shorter. Unfortunately, this only works on the IDE channel, but gave me
better times than a 120GB SATA drive, by 4 MB/second. Not too shabby!
How about the games! Well, here are the scores and nothing had been
overclocked, besides the CPU to 3200+ speeds (all measurements taken @
1024x768. No other resolutions were tested.
As you can see,
at stock settings, pretty much all recent games will play well. I would
have included Doom3 in the benchmark run, but lack of a 3rd optical
drive ruined that idea. Needless to say, all in all, I'm quite
impressed.
|