On Fri, 17 Apr 2009 10:03:50 +0100, darklight
<nglennglen@netscape.net> wrote:
>regarding the post sent by muzician21 on 09/04/2009 about
>Suggestions for Core 2 Duo systems that use PCI - not PCI express
>why don't people download the 'benchmark designed by Australian PC User
>magazine' web address http://darrenyates.com.au/?p=573
>then post their results just to see what the difference in performance is
>between hardware and os.
The problem is threefold.
1) People need PCI slots when they have cards they either
can't replace, or don't want to pay to replace, so PCIe
isn't an option for them.
2) Since the exact same cards are usually not available in
both PCI and PCIe formats, a fair comparison cannot be made
of what the performance degradation would be while using the
PCI card instead of the PCIe, and if using neither than
which slot type it has will make no difference but rather
the particulars of the chipset performance and bios settings
might make a small difference, especially since the memory
controller for Core 2 Duo is still in the northbridge
meaning newer chipsets will tend to be faster.
Ultimately, if someone wants/needs to use a PCI card, they
will still get great benefit from the newer platform even if
the PCI bus occasionally bottlenecks things. The other
remaining issue is if there really aren't PCIe alternatives
and if not, will that situation improve in the future or are
they just delaying the inevitable by a few years to still
end up in the same position with even fewer alternatives
since slowly but surely there will be fewer and fewer PCI
slots on modern motherboards, then people will have to
settle for defeatured legacy industrial boards at much
higher cost.
Then again, it is possible that some day someone will come
up with a bridge card similar to the riser cards used years
ago that converts one PCIe slot into several PCI, but using
such a card would require a specialized case since it would
put all the cards' slot mount brackets at 90' orientation to
the an ATX case's slots, and a different location.
3) The linked benchmark does not specifically stress the
operations on the PCI bus to the cards one might want to
use. That would be necessary in order to show the
difference in performance when there is one.
Oh, there is a fourth issue too:
4) Quoting Darren Yates - "UserBench Encode 2009 is
exclusive to Australian PC User magazine and you’ll only
find it in the February 2009 issue, on sale now." That
makes acquiring the software inconvenient at least, costly
if not reasonably impossible for those outside of Australia.
just a bit of fun 
Linear Mode
