[ On Thu, October 8, 1998 at 21:24:47 (-0500), seebs@plethora.net wrote: ]
> Subject: Re: Another changer, another changer problem 
>
> In message <m0zRRqU-0009MWC@most.weird.com>, Greg A. Woods writes:
> >	4 32-bit PCI slots 
> >	1 Asus MediaBus (Revision 2.0) Slot. 
> >	3 16-Bit ISA Slots. 
> 
> A "MediaBus" slot is a PCI slot and an ISA slot in one position.
> 
> 5 PCI slots, 4 ISA slots, but one overlap.

OK, then if there's also a couple of PCI attached controllers on the
motherboard then I'll bet dollars to donuts (i.e. coffe and donuts at
the next improbable mutual gathering to anyone proving me wrong!) that
there are at least two pci buses inside that board.

Of course a "dmesg" output from the board would would clear this up.

Remember Justin's dmesg from his honkin' big FreeBSD server?  It would
has a structure something like this (I think):

CPU---->PCI-to-Host--+-->PCI-to-ISA--+-->IDE controller
                     |               +-->USB controller
                     |               +-->APM controller
                     |               |
                     |               +-->ISA slots
                     |
                     +-->VGA controller (d-1)
                     +-->NIC controller (d-2)
                     |
                     +-->PCI-to-PCI--+-->PCI-to-PCI--+-->pci2.s1
                     |               |               +-->pci2.s2
                     |               |               +-->pci2.s3
                     |               |               +-->pci2.s4
                     |               |               |
                     |               |               +-->Parity Engine 0
                     |               |
                     |               |
                     |               +-->PCI-to-PCI--+-->pci3.s1
                     |                               +-->pci3.s2
                     |                               +-->pci3.s3
                     |                               +-->pci3.s4
                     |                               |
                     |                               +-->Parity Engine 1
                     |
                     +-->PCI-to-PCI--+-->pci4.s1
                     |               +-->pci4.s2
                     |               +-->pci4.s3
                     |               +-->pci4.s4
                     |               |
                     |               +-->Parity Engine 2
                     |
                     +-->PCI-to-PCI--+-->pci5.s1
                                     +-->pci5.s2
                                     +-->pci5.s3
                                     +-->pci5.s4
                                     |
                                     +-->Parity Engine 3

where "pciN.sM" are PCI bus #N, slot #M.  I'm guessing about the actual
tree structure, but I think this is one possiblity.  There may not
actually be that many physical slots either -- I'm just guessing here...

Of course the documentation for the 21150 PCI-to-PCI bridge chip gives
all the possibilities.....

> I've found lots of boards with at least 5 PCI slots, and someone was
> mentioning having one with *8*.

CompactPCI backplanes have up to 8 slots per *bus*, one for the CPU.

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>