Re: PC bios recognizing PCI card

[Ted Firlit <firlit@utmc.aeroflex.com>]

Austin Franklin wrote:
> 
> This has been an issue.  There is nothing in the PCI spec (2.1 at least) for
> how long a card has to become "configured", per se...and since your FPGA has
> to take some time to configure, it is probably taking longer than the PC is
> taking to go look for it.
> 


There is a spec in PCI 2.2 Table 4-6 for 2^25 clocks from RST# high to First configuration Access.  I've attached previous postings about 'quick-start' motherboards violating this spec. (earliest posting is last)

If anyone has an opportunity to measure the RST# high to First config on a quick-start motherboard, that would be useful information for the rest of us!


regards,

Ted Firlit

----------------------------------------

Subject:            Re: quick start in new motherboards
      Date:            Fri, 15 Feb 2002 12:19:52 +0100
      From:            "Marco Serafini - QR s.r.l" <r&d@qrverona.it>
        To:            "Peter Marek" <peter.marek@marekmicro.de>
        CC:            <ted.firlit@utmc.aeroflex.com>, <pci-sig@znyx.com>
 References:           1 , 2 , 3 , 4

Peter and Ted,

In a message posted elsewhere by a  hardware designer I read:

"On some TYAN motherboard with AMIBIOS and Intel i810/815 chipset
or on some newest VIA boards with AWARD
after PowerON I see in the BIOS table no cards exept my. So
I run pci test soft and see that all cards on PCI Bus
phisicaly present, but part of their PCI Configuration space is failed.
If we see in PCI standard - motherboard must wait about 2 seconds
for flashing PCI Config Space of PCI cards.
But newest motherboard after 2 second boots Windows."

In all my PCI boards, for bus interfacing, I use FPGAs with configuaration
devices, what have I to expect?
I haven't met this problem yet.

Thanks for your last feed-back!

Marco



Subject:            Re: quick start in new motherboards
      Date:            Thu, 14 Feb 2002 17:08:16 +0100
      From:            "Peter Marek" <peter.marek@marekmicro.de>
        To:            <ted.firlit@utmc.aeroflex.com>
        CC:            "Marco Serafini - QR s.r.l" <r&d@qrverona.it>, <pci-sig@znyx.com>
 References:           1, 2 , 3

Ted,

although there is a minimum time from the rising edge of RST to the first
config access in the PCI2.2 specification, most BIOS/mainboard vendors do
not adhere to this parameter in "QuickStart" mode (which is legal in my
opinion since "Quick Starts" should be quick, indeed.).

The actual time depends on processor type and speed, mainboard and BIOS.
There are no general guidelines how quick "Quick" is...

Regards,

Peter Marek
General Director
MarekMicro GmbH
tel.: 049-9621-9732-110
fax: 049-9621-9732-199
www.marekmicro.de

----- Original Message -----
From: "Ted Firlit" <firlit@utmc.aeroflex.com>
To: "Peter Marek" <peter.marek@marekmicro.de>
Cc: "Marco Serafini - QR s.r.l" <r&d@qrverona.it>; <pci-sig@znyx.com>
Sent: Thursday, February 14, 2002 4:11 PM
Subject: Re: quick start in new motherboards


> Peter,
>
> Do you have any info as to just how fast the "Quick Start" BIOS gets from
RST# removal to PCI config cycles?
>
> thanks,
>
> Ted
>
> Peter Marek wrote:
> >
> > Hi,
> >
> > this is true. The cause for this strange behaviour is that these chips
load
> > a part of their config space from an external serial EEPROM. If this
takes
> > too long (or the Quick Start is too fast), the chips will not be
configured
> > correctly.
> >
> > Non volatile solutions for custom backend busses like QuickPCI from
> > QuickLogic do not face these problems.
> >
> > regards,
> >
> > Peter Marek
> > General Director
> > MarekMicro GmbH
> > tel.: 049-9621-9732-110
> > fax: 049-9621-9732-199
> > www.marekmicro.de
> >
> > ----- Original Message -----
> > From: "Marco Serafini - QR s.r.l" <r&d@qrverona.it>
> > To: <pci-sig@znyx.com>
> > Sent: Thursday, February 14, 2002 12:17 PM
> > Subject: quick start in new motherboards
> >
> > > Dear PCI experts,
> > >
> > > I've read in a tech-forum that some PCI cards based on nonstandard
> > > PCI-to-local bus chips
> > > (for example, PLX or ALTERA FLEX/ACEX) lost their PCI configuration
space
> > on
> > > early BIOS start.
> > > Particularly on the new motherboards with "quick start" CMOS option.
> > >
> > > Have you ever heard this topic?
> > > Have you ever met any similar problem with modern motherboards?
> > >
> > > Thanks for comments.
> > >
> > > Marco