Ron - are there replicators for PCI slots? Can I make two PCI slots
out of one?
If this does not work I will have to get a new motherboard with more
PCI slots.
There are PCI expansion chassis. An example is shown here.
http://www.magma.com/products/pci/4PCI/
How this works, is there is a PCI bridge implementation on two
cards. One card plugs into the original PCI bus. A cable extends
from this interface card, to the expansion chassis. At the expansion
chassis, the signals from the cable are converted into something suitable
for driving the second PCI bus segment.
The performance impact is - bus bandwidth is preserved. The bus still has
its "133MB/sec" theoretical burst bandwidth (closer to 100-110MB/sec in
practice). But the "first cycle latency" is higher. There are one or more
cycles of PCI bus delay, going from the originating computer, to a PCI card
in the expansion chassis. And what that means, is doing single word cycles
on the expansion chassis address space, is a lot slower. So as long as the
data is travelling in bursts, the first cycle latency is manageable (amortized
with respect to the size of the burst). But if the PCI card in the expansion
chassis, is accessed and controlled with a lot of single word operations,
performance plummets.
You would need to juggle the PCI cards, to get the best performance. If
the data acquisition cards had low data rates, they might be good candidates
for the expansion chassis.
If you look at the price of the expansion chassis, that might put you off.
A better selected new computer, might be cheaper. But for people with a
large number of PCI cards, a box like this may be an inviting option.
The DualHead2Go compatibility table is here. It appears the max you could
expect, is two screens at 1024x768. The limit is likely that the 865GV cannot
handle a line wider than 2048.
http://www.matrox.com/graphics/en/cadgis/support/dh2go/compatibility.php
The datasheet for the 865GV is here, and many important details are missing.
(No register map for video that I can see. No way to see whether the 2048 is
hardware enforced.)
http://download.intel.com/design/chipsets/datashts/25251405.pdf
Paul