IDE Promise RAID on K8V SE slows system down

Take a look at the following article - only 28 pages and 200 graphs of SATA
raid benchmarks:

<a style='text-decoration: underline;' href="http://www.techreport.com/reviews/2004q2/chipset-raid/index.x?pg=1" target="_blank">http://www.techreport.com/reviews/2004q2/chipset-raid/index.x?pg=1</a>

Basically, I read this to say that 32 bit drivers for AMD 64 based boards
show a lot of promise but need some attention. The K8V has the VT8237
chip...

They say "
VT8237 - The VT8237 is probably the most widely-used south bridge with
Serial ATA RAID support, but popularity doesn't guarantee great performance.
It doesn't guarantee poor performance, either, and that's sort of where the
VT8237 sits. The chip's performance is reasonably consistent between Intel
and AMD-based platforms, and it's generally unremarkable across the board.
Unremarkable isn't necessarily a bad thing, it just means that there's
nothing overly impressive-or depressing-about the VT8237's scores relative
to the rest of the field.

If the VT8237 has one weak spot, it's performance scaling in IOMeter, where
RAID 1 and 0 arrays struggle to perform better than single-drive
configurations. It's also unfortunate that mobo makers haven't taken
advantage of the chip's extra two Serial ATA ports, which would enable
support for four-drive arrays and RAID 0+1. "

I would bet the drivers will improve...

- Tim



"David Pipe" <dontsendme RemoveThis @spam.com> wrote in message
news:otqtc.13628$be.10038@newsread2.news.pas.earthlink.net...
 > K8V SE, Athlon 64 3200, 1GB RAM, Windows XP
 >
 > I connected one WD800JB (80GB, 8M cache) to the standard IDE controller,
 > and
 > it did WinBench 99 tests, then connected it to the Promise RAID
 > controller,
 > added another IDENTICAL drive, had the controller build a mirror, then
 > re-booted and re-ran the WinBench tests. The results were disappointing
 >
 > Standard IDE, one drive
 > Business = 7020
 > High End = 24500
 >
 > Promise RAID 1, two drives
 > Business = 9220
 > High End =19500
 >
 > There's a slight increase in the business benchmark, but a significant
 > decrease in performance for the High End benchmark. I attribute this to
 > the
 > fact that there is only one (parallel) IDE RAID connector, so you've got
 > to
 > have the two drives connected to the same cable, therefore in RAID 1 write
 > operations the controller has to write the data twice--once to each drive
 > (since the IDE interface can't handle two commands simultaneously).
 >
 > I've performed this experiment with two IDE drives on Promise controllers
 > in
 > the past, but each drive had it's own IDE channel, and I have seen
 > increases, rather than decreases in performance. But as I said, there is
 > no
 > second parallel IDE connector on this board.
 >
 > Does anyone have any comments?
 >
 > Dave in Colorado
 >
 >
 >
 ><!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: IDE Promise RAID on K8V SE slows system down