The VT6X4 offers a few features that the BF6 lacks. The VT6X4 uses the
VIA 133A chipset, which offers native UDMA/66 support, AGP 4X, ? AGP divisor
when using 133MHZ FSB and beyond. Additionally, the VIA 133A chipset allows
for asynchronous RAM timings that can either A: save you a few dollars on RAM,
or B: increase your speed. For example, if you want to overclock your 100MHZ
FSB Pentium III, to 133MHZ FSB, but your RAM is only PC-100 (and cannot run stably
at 133MHZ) then the -33MHZ RAM setting can help out a lot. Alternatively, you
have a 100MHZ FSB CPU, but have PC-133 memory, then the +33MHZ RAM setting will
increase your memory bandwidth.
For many users, the Intel 440BX chipset is the preferred chipset, and for
good reason, it's mature, stable, well supported, and there's no need to worry
about downloading "4 in 1" Driver sets. Most importantly, the Intel 440 BX
chipset is still the current standard bearer for SPEED. Well you've all been
waiting to get to the benchmarks. so here they are:
Sisoft Sandra Professional 2000:Intel PIII
600E @ 600MHZ /w ABIT BF6:
|
Here we see that the PIII 600E is a little
bit slower than the Athlon 600 reference
system |
|
Once again we see that the PIII-600E is a
little bit slower than the Athlon 600 reference
system. |
|
The Athlon reference system seems to have
incredibly high memory bandwidth scores (200MHZ FSB is probably the
reason.) I am also getting substandard memory bandwidth scores, but
they're not too bad here. |
Intel PIII 600E @ 600MHZ /w ABIT VT6X4
|
Now we can finally make some comparisons. The
VIA 133A platform is lagging by only a small margin (less than 10 points.)
So there is no real edge for either
platform. |
|
Once again, the BX
platform is only in the lead by a few meager points.
|
|
Now we see a HUGE difference, the VIA 133A
chipset falls far behind the BX chipset for memory bandwidth. For
CPU/Memory Bandwidth, there is a difference of 89 points. For FPU/Memory
Bandwidth, we see a difference of 91 points. So how does this difference
in memory bandwidth affect `real-world' benchmarks? Let's begin with the
perennial favourite. Quake III, followed by 3D Mark 2000, and a
temperature reading with the Intel heat sink and the Alpha FC-PAL
35. |