[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Maximum load of PCI-BUS



The maximum number is 21 devices because the host bridge can configure
only 21 devices. See figure 3-3 of PCI spec 2.2.

It is not clear if the 21 devices include the host bridge itself. I think
it should because the host bridge is a PCI device and needs to be configure.
In that case, there should be 20 PCI device + the host bridge.

Of course this is the absolute maximum. I believe capacitance and loading
limits the device number to a smaller number.

Simon Lau

"Ingraham, Andrew" wrote:

> > > In the PCI specification the input capasitance is specified with 10 pf.
> > > Which is the maximum output load (capasitance) ?
> > >
> > > In a PC hardware description book I read the maximum number of
> > > participants is 10 !? (Is that true - up to date?)
> > > On a ASUS motherboard CUBX are 6 pci slots and some other PCI chips
> > (maybe
> > > 3).
> > >
> > > Can anybody tell me the rules ?
>
> I believe the maximum number of devices you can have per bus segment is 32,
> limited by the 5-bit Device Number field in  Configuration cycles.
>
> You might find references to a PCI bus limit of 10 devices.  But there is no
> such limit.  Ten is a recommended or typical number of equivalent loads, for
> buses running at 33 MHz.  In this equivalent load count, an IC on the
> motherboard counts as one load, but a plug-in card counts as two.  Thus, you
> get the often-quoted suggested configurations of:
>         6 motherboard devices plus 2 expansion cards, or
>         2 motherboard devices plus 4 expansion cards.
> Again, these are suggestions, not rules.
>
> Elsewhere in the PCI Spec, it also suggests that 16 loads are "believed to
> be a worst case;" and that system designers can have a trade-off between
> number of loads and the bus clock rate.  The system designer just has to
> make it work.
>
> In addition to device (IC) input capacitance, don't forget about trace
> capacitance and connector capacitance.  You might find that there is more
> capacitance in the traces than in the IC pin.
>
> Regards,
> Andy

--
------------------------------------------------------------------------
Eureka Technology Inc.                        http://www.eurekatech.com
4962 El Camino Real, #108,                    Tel: 650 960 3800 ext 110
Los Altos, CA 94022, USA                      Fax: 650 960 3805
                                             email: simon@eurekatech.com
------------------------------------------------------------------------