two IDE PCI contoller cards in one PC?

[Timothy Daniels]

I have a use for (4) 133MB/s IDE channels in my Dell
(which runs WinXP Pro). I'm now using a single SIIG
PCI controller card, but SIIG says that they don't know
if my Dell Dimension BIOS or their drivers would be
able to differentiate between two identical PCI cards.

Has anyone here used two identical IDE PCI cards in
the same PC? How about two different brands of IDE
PCI cards in the same PC?

*TimDaniels*

 

[Mike Yetsko]

I have a use for (4) 133MB/s IDE channels in my Dell
(which runs WinXP Pro). I'm now using a single SIIG
PCI controller card, but SIIG says that they don't know
if my Dell Dimension BIOS or their drivers would be
able to differentiate between two identical PCI cards.

Has anyone here used two identical IDE PCI cards in
the same PC? How about two different brands of IDE
PCI cards in the same PC?

*TimDaniels*

Well, unless the BIOS actually interferes, you should be able to
plug in as many cards as you want, all identical, and they should
work fine. That's because PCI 'rolls' the address as part of the
physical socket, and rotates the interrupt line.

As long as the BIOS stays out of the way of what it doesn't know
about, WIN98 and later, with appropriate drivers, should find
the card, and install more drives than you should ever need.

 

[Timothy Daniels]

Well, unless the BIOS actually interferes, you should be
able to plug in as many cards as you want, all identical,
and they should work fine. That's because PCI 'rolls' the
address as part of the physical socket, and rotates the
interrupt line.

As long as the BIOS stays out of the way of what it doesn't
know about, WIN98 and later, with appropriate drivers,
should find the card, and install more drives than you
should ever need.

That sounds encouraging. Do you know if the same
currently-installed driver could handle both PCI cards,
or would another driver have to be installed for the 2nd
card?

*TimDaniels*

 

[Ben Myers]

If the cards are the same manufacturer and model, one copy of the required
driver will suffice. Windows XP is smart enough to create instances of data
tied to each channel and each device, executing the same driver code.

Same with network cards, as in a firewall or internal network controller type of
computer. Throw in several network cards, all same manufacturer and model, and
there is only one driver required.

.... Ben Myers

 

[Timothy Daniels]

Thanks, Ben. I guess I can now spring for another
SIIG card (which I trust, since it has worked for me
so far for 1½ years).

*TimDaniels*

 

[Guest]

One problem that you could possibly run into is one that we had with some
Promise cards (two different models) - they did not seem to want to share
IRQs, and there were not enough to go around. Ways around this may be to move
cards around to other PCI slots, or disable unused devices (the PS/2 mouse
port if you are using an USB mouse, for example).

|Thanks, Ben. I guess I can now spring for another
|SIIG card (which I trust, since it has worked for me
|so far for 1½ years).
|
|*TimDaniels*
|
|(Ben Myers) wrote:
|>
|> If the cards are the same manufacturer and model,
|> one copy of the required driver will suffice. Windows XP
|> is smart enough to create instances of data tied to each
|> channel and each device, executing the same driver code.
|>
|> Same with network cards, as in a firewall or internal
|> network controller type of computer. Throw in several
|> network cards, all same manufacturer and model, and
|> there is only one driver required.
|>
|>"Timothy Daniels" wrote:
|>
|>> That sounds encouraging. Do you know if the same
|>> currently-installed driver could handle both PCI cards,
|>> or would another driver have to be installed for the 2nd
|>> card?
|>>
|>>"Mike Yetsko" wrote:
|>>>
|>>> Well, unless the BIOS actually interferes, you should be
|>>> able to plug in as many cards as you want, all identical,
|>>> and they should work fine. That's because PCI 'rolls' the
|>>> address as part of the physical socket, and rotates the
|>>> interrupt line.
|>>>
|>>> As long as the BIOS stays out of the way of what it doesn't
|>>> know about, WIN98 and later, with appropriate drivers,
|>>> should find the card, and install more drives than you
|>>> should ever need.
|>>>
|>>> "Timothy Daniels" wrote:
|>>>
|>>>> I have a use for (4) 133MB/s IDE channels in my Dell
|>>>> (which runs WinXP Pro). I'm now using a single SIIG
|>>>> PCI controller card, but SIIG says that they don't know
|>>>> if my Dell Dimension BIOS or their drivers would be
|>>>> able to differentiate between two identical PCI cards.
|>>>>
|>>>> Has anyone here used two identical IDE PCI cards in
|>>>> the same PC? How about two different brands of IDE
|>>>> PCI cards in the same PC?
|
|

 

[Mike Yetsko]

One problem that you could possibly run into is one that we had with some
Promise cards (two different models) - they did not seem to want to share
IRQs, and there were not enough to go around. Ways around this may be to move
cards around to other PCI slots, or disable unused devices (the PS/2 mouse
port if you are using an USB mouse, for example).

Uh, that wasn't the problem. PCI cards don't share interrupts. They have
PCI BUS INTERRUPTS. And the PCI bus supports 4 of them. Each card
generally supports PCI_INTA, with some also supporting PCI_INTB. Som
even PCI_INTC and PCI_INTD.

However, in the first slot the Card A is connected to INTA on the bus, B to
B, and so on. In the second slot, the Card A is connected to bus B, then
B to C, and so on. Third slot is A to C, B to D, C to A, etc.

Since by normal convention, the PCI cards will try to use INTA for it's main
interrupt, this 'shares the load' of PCI devices on the interrupt bus
structure.

THERE IS NO SUCH THING AS AN IRQ5 for example on a PCI
device. Now, the HOST to PCI bridge device may in fact generate
a 'conventional' IRQ, but that is another story.

ALL PCI INTERRUPTS ARE SHAREABLE BY DEFINITION AND
DESIGN!

There really is no such thing any more as 'not enough interrupts to go
around'. If you think that's your problem, you either have a brain dead
design or a defective driver. But it SURE ain't the interrupts.