Motherboard does not detect any PCI devices

[.]

Abit AB9 board doesn't recognize my Audigy 2 card, nor an ASUS PCI TV
card. Intel 965 chipset, E6300 CPU. This board only has two PCI card
slots and it is as if they are dead.

Does anyone have any suggestions. I have cleared the CMOS to no avail.
The BIOS is up to date.

 

[Paul]

Abit AB9 board doesn't recognize my Audigy 2 card, nor an ASUS PCI TV
card. Intel 965 chipset, E6300 CPU. This board only has two PCI card
slots and it is as if they are dead.

Does anyone have any suggestions. I have cleared the CMOS to no avail.
The BIOS is up to date.

If the board has Firewire, and the chip is TI 43AB22A, check
your Device Manager. If the Firewire is showing up there,
then it means your PCI bus is not dead. The Firewire appears to be a
PCI bus chip, and in a sense, your Firewire is like it is
sitting in a third PCI slot. It is part of the same bus.

http://focus.ti.com/docs/prod/folders/print/tsb43ab22a.html

You could also use a program like Everest Free Edition, to
check DevicesCI and see what shows up there.

http://www.majorgeeks.com/download4181.html

Paul

 

[.]

Try PCItree:
http://www.pcitree.de/

It will show all your PCI devices, not just the ones in the slots.

- Franc Zabkar

All of the motherboard embedded PCI devices are present, it seems that
the motherboard is just ignoring the two PCI connector based slots.

This board (AB9) is a replacement to a failed MSI Neo 965 chipset board,
it was the only 965 chipset based board the shop still had, so I grabbed
it. The firmware in the 2006 built BIOS was flakey (didn't support DMA on
DVD IDE) so it got upgraded, so were all the motherboard drivers. All
works now but the damned PCI cards.

What motherboard driver module is responsible for managing the PCI slots?

 

[Paul]

All of the motherboard embedded PCI devices are present, it seems that
the motherboard is just ignoring the two PCI connector based slots.

This board (AB9) is a replacement to a failed MSI Neo 965 chipset board,
it was the only 965 chipset based board the shop still had, so I grabbed
it. The firmware in the 2006 built BIOS was flakey (didn't support DMA on
DVD IDE) so it got upgraded, so were all the motherboard drivers. All
works now but the damned PCI cards.

What motherboard driver module is responsible for managing the PCI slots?

In Device Manager, under "System Devices", you should have a "PCI Bridge - 244E".
That is the DMI (Northbridge to Southbridge bus) to PCI bus, bridge. The
PCI devices should sit on that.

There should not be any difference, between a motherboard PCI device, and
the two PCI slots. If the motherboard PCI works, then so should the
slots.

Paul

 

[.]

In Device Manager, under "System Devices", you should have a "PCI Bridge
- 244E". That is the DMI (Northbridge to Southbridge bus) to PCI bus,
bridge. The PCI devices should sit on that.

There should not be any difference, between a motherboard PCI device,
and the two PCI slots. If the motherboard PCI works, then so should the
slots.

There must be difference, otherwise why wouldn't the motherboard
recognize my PCI cards? The cards are both good. I can put them into
another PC and they are recognized.

 

[Paul]

On Thu, 03 Apr 2008 23:11:20 -0400, Paul wrote:

There must be difference, otherwise why wouldn't the motherboard
recognize my PCI cards? The cards are both good. I can put them into
another PC and they are recognized.

PCI is a shared bus. The Address/Data lines go past all the PCI devices
in parallel.

The PCI spec would normally be purchased from pcisig.com . But if you
have a look with your search engine, you can find a copy of pci22.pdf
for download. About 3,835,430 bytes or so.

If you want a sample motherboard schematic, there is one at Intel.
PDF page 44 is where the PCI slots start. IDSEL gets wired to a different
AD bcon on each slot, for configuration cycles, but many of the other
signals are shared by all slots. The interrupt lines are "swizzled" but
that isn't important either to this discussion. All you need, is for
some interrupt line to be wired up, the exact line is not important
because they can be shared.

http://www.intel.com/design/chipsets/schematics/252812.htm

Something like a PCI slot clock could be disabled, and that
might be a reason for a slot to be disabled. But then the
question would be, why would that happen ? The BIOS has the
option of reaching into the clockgen, and disabling individual
clocks (it can be done to reduce EMI for example), but I don't see
why that would happen.

The clock generator on my motherboard, is this one. The register
to disable PCI clocks is on page 9, at the top of the page.

http://web.archive.org/web/20050530125355/http://www.icst.com/pdf/ics952607.pdf

And looking at the BIOS, I don't see something in particular
that would disable a portion of the PCI bus. There is an
"Onboard PCI Device" section to the BIOS, and the very last
item "OnBoard 1394 Controller" could be a PCI bus device. I
suspect the rest might be PCI Express devices. Do you have
the 1394 device disabled or enabled in the BIOS ?

I'm assuming here, that the 1394 device is an actual PCI bus
device, and seeing it visible in Windows is my proof that
*some* PCI device works. I would boot Ubuntu or Knoppix CD for
a second opinion, and try "lspci" and see what shows up there.

Paul

 

[.]

I don't really understand how PCI device enumeration works, but I'm
wondering if the BIOS thinks that it is Windows' job to enumerate the
devices while the OS thinks that it is the BIOS's job ???

Would experimenting with the "PnP aware OS" setting in the BIOS setup be
a worthwhile thing to try?

In the AB9 BIOS there is the choice of 'Automatic' or 'Manual'
specification. I Have tried both.

Otherwise, while I'm not a Win XP user, I notice that under Win98SE
there is a "PCI bus" under "System Devices" in Device Manager.
Properties > Settings allows me to select "Hardware" or "BIOS" for
device enumeration. There is also an "Override Bridges" checkbox.

In XP the device mananger has a command to scan for hardware changes.
Choosing that command forces the OS to enumerate the PCI bus. It doesn't
find the cards. Interestingly in the PCI bridge properties/Power state
mappings display, two slots (S1 and S2) have 'unspecified' mappings where
the remainder are mapped either to 'D0' or 'D3'.

 

[.]

It's an old Phoenix BIOS utility but it still appears to work for modern
BIOSes of all flavours.

I tried that, but it doesn't run on my Phoenix AWARD BIOS, it returns an
error.

C:\>nvram120.exe -e

NVRAM, Version 1.20
Non-Volatile RAM System Configuration Data Utility.

Copyright(c) 1994-1996 by Phoenix Technologies, Ltd.
All rights reserved.

Display ESCD info on all nodes:

Error reading ESCD NVRAM, function 0x41, rc=0x82.

C:\>

I think I will resign myself to having a faulty motherboard. It is not as
if I need those devices in this computer and I can't afford the down time
to send the board away because I use this PC for software development
work.

Thanks for your help anyway.

 

[.]

There may be another way to retrieve this information but it is a little
more involved. You may be able to use your BIOS flash utility (or
Uniflash?) to save an image of your BIOS EEPROM and then use a hex
editor to manually extract the ESCD table from this image. Look for an
"ACFG" signature preceded by two size bytes. Write this ACFG data block
to a file and then submit this file to NVRAM120.EXE.

I read the BIOS image and then had a look at the .bin file. There is no
"ACFG" pattern in it, but I did see a text string with the word "bios
compression =" so I guess that some of the data in the file is compressed.

For what it is worth I submitted the .bin file to the nvram120 utility
but it told me the file was no good.

Thanks for your help.