Archived from groups: alt.comp.periphs.mainboard.asus (
More info?)
In article <p2hq71ddpaii2l9funr6nfn3agbr7mhnv6@4ax.com>, Michael Meyers
<meyers@shaw.ca> wrote:
> On Sat, 07 May 2005 23:29:50 +0100, Ben Pope <benpope81@_REMOVE_gmail.com>
> wrote:
>
> >Michael Meyers wrote:
> >> Recently had 2 PCs built by a local shop. Win XP SP2, each with
identical Asus
> >> K8V SE Deluxe MOBOs and single 160 GB Seagate SATA hard drives in
each PC. I
> >> recently realized that on one PC the HD is connected to the VIA
controller, and
> >> on the other it is connected to the Promise controller.
> >>
> >> My question is, is one setup "better" than the other in terms of
performance,
> >> all other things being equal?
> >
> >You tell us, you seem to have a pretty good benchmarking environment
😛
> >
> >I would guess the VIA, marginally.
> >
> >Ben
>
> Sorry, I forgot to mention, my PC has an Athlon 64 3400+ and OCZ RAM, and my
> son's has an Athlon 64 2800+ and "generic" RAM, so they're not identical
> systems. But I want to maximize performance in each, hence the question about
> which controller is "better".
There is some P.R. material here. It compares the 8237 to
one of the Promise controllers. The claim is, the internal
bus in the 8237 is not constrained by the PCI bus.
http://www.via.com.tw/en/products/chipsets/southbridge/vt8237/drivestation.jsp
Note that accepting benchmarks from P.R. material is
not a real good thing to do, so perhaps if you look in
the private forums, you can get more realistic numbers.
For example, if the PCI latency timer is cranked up on a
motherboard, a disk controller can achieve much higher
disk benchmarks, but with the danger of unfairness amongst
all the PCI cards. If you are using a PCI sound card, it could
be starved for data by the disk controller, and suffer an
underrun. In the real world, system operation must be adjusted
so all peripherals can operate properly, and this condition would
not give as good a disk benchmark.
You could test your two machines with Sisoft Sandra or Hdtach,
to do your own benchmarking. That testing will be much more
meaningful, since it is done on the target hardware.
HTH,
Paul