On Sat, 01 Jul 2006 15:01:31 +0100, Warra
wrote:

On 22 Jun 2006, wrote:

In other words, your board is among the
worst to use a PCI SATA controller on.

Oh bull****.

Have you ever actually TRIED a PCI card on that chipset?
I have... benched it too. Don't recall the scores but did
recall the very significant difference in use of a PCI
controller on that and prior, next gen Via chipsets.
Google for the info if you don't believe,

If you Google,
http://www.google.com/search?hl=en&q...ATA+benchmarks

look at the very first hit, it happens to be KT266A...
http://www.tecchannel.de/ueberblick/...70/index3.html

... and this is even BEFORE one tries to use the PCI bus for
other concurrent things like audio or whatever.

In computing most things are typical, but occasionally some
things stand out as very good or bad. Via chipsets PCI
performance in that era were very bad.


I clicked your Google link and saw this at
http://www.quepublishing.com/articles/article.asp?p=339072&rl=1

----------------------------- QUOTE
Performance Driven Design - More than Just a Slogan

VIA Technologies replaced the unimpressive memory controller in the
KT266's North Bridge (the VT8366) with a greatly improved design in
the new North Bridge chip (VT8366A). The new memory controller is
referred to by VIA's block diagram (available at VIA's web site) as
"Performance Driven Design". Technically speaking, this name sums up
a lot of changes, including:

double the burst rate for memory transfers
deeper read-write internal memory buffers (also called data queues)
faster memory timings

This improvement in the memory controller (Tom's Hardware calls it
the "industry's fastest DDR memory controller") is the major change
in the KT266A over the KT266, and takes care of the major weakness in
the original design
----------------------------- UNQUOTE

Seems that the KT266A chipset is not as affected as the KT266.
Presumably (?) this also applied to PCI cards?


Yes I think I was mistaken about the bus speed issue, that
was going from KT133 to KT133A. I now wonder how many
boards actually surfaced with the KT266 (non-A) chip though
as it stirs up another memory that boards may have been
designed for KT266 but that many ended up shipping with
KT266A (which is pin-compatible, IIRC) instead.