Abstract: This is not going to be pretty folks, so I'll take this moment to recommend that anyone easily disgusted by low performance, or has a weak stomach to leave this article
immediately. 83% Rating:
Now most will say the synthetic memory benchmarks are mildly inconclusive and are just that, synthetic. But in this case they are a good measure between bandwidth perceived on each subsystem and how it relates to the end users "feel" of a system.
There are arguments on both sides of this, and I
am not getting into that for now, just enough to say that regardless of what
some may see as valid readings of actual real world performance, these are
interesting enough in that there are viable and noticeable differences in each
boards implementation of the chipset and it's abilities to transfer information
along the memory, chipset, and CPU subsystem. Also added into the mix is a real
world performance type benchmark of Winstone 99, with it's abusive tests of the
3 most popular suites of business software, the systems ability to handle
multiple tasks at blazing speed, and how it relates to the user and what they
will be able to do at any given moment.
Not to be left out is a dose of Quake 3 Arena scores at the max of 32 bit colors and textures, with a resolution of 1024x768. Any other scores would have been redundant, and also this is not a review of a video card. I used those settings to show that being the video card is mostly taxed at this high of a viewing state, and relies more heavily on the subsystem to achieve higher frame rates it becomes a good comparison of the implementation of the chipset, the new AGP 4x option, and how well the motherboard can handle intense gaming graphics of this nature where the subsystem must pick up where the video card left off.
Test System Configuration
Processor(s) Intel Pentium III 550E Slot-1 Memory 64 Megs PC-133 SDRAM Hard Drive Quantum Fireball 3.2 Gig Video Card Elsa Erazor X2 GeForce DDR Operating System Windows 98 SE
Take a close look at these numbers folks. I ran all tests 3
times, averaged the scores and came up with what is shown. I also showed scores
with everything running at spec'd speeds, as well as the highest stable
attainable speeds on each. My definition of stable is the systems ability to run
all benchmark suites with absolutely no errors or hangs completely through 3
times. As you can see, the Tyan falls short in all areas, and wouldn't OC
substantially enough for a real performance gain. AOpen took the show with it's
excellent ability to run at the higher bus speeds, showing the attention they
put into the design and implementation of Via's new chipset.
While it is still in it's infancy (IMHO), and doesn't
quite get the numbers of our old standby the BX, it is still by far the better
of the two tested Via boards and offers the hobbyist a largely more enjoyable