Anyone help?

Not sure how the interrupt controllers work on that board, but ISA always uses edge-triggered interrupts, that is how the bus was designed;
one nice thing with the PCI bus is the level-triggered interrupts, which allowed sharing them. If you have multiple interrupt controllers,
each controller could have its own configuration in that respect, but maybe that chipset only allows one common configuration for all IRQ/INT lines?

Why IRQ11 is not working, I don't know...maybe the physical connection is bad, and the video card is not actually needing any IRQ/INT line?

Rwolf wrote on 2023-05-25, 09:56:

maybe that chipset only allows one common configuration for all IRQ/INT lines?

This is very informative, thanks!

Rwolf wrote on 2023-05-25, 09:56:

... and the video card is not actually needing any IRQ/INT line?

Yes, you are right. The video card doesn't really need an IRQ. There is option in BIOS called "Assign IRQ For VGA". When set to "disabled", the video card(the PCI one)'s IRQ value becomes NA under the "PCI device listing...." table during system boot up. I've managed to update the BIOS to a new version and the old one didn't show the correct information. It would show 10/11 instead of NA. However, the update broke the ps/2 mouse support. The system, both Windows and DOS, cannot detect my ps/2 mouse now. I think I have to flash the previous version back if no other version could be found.

I tried another version of the BIOS, it was an Evaluation version. Mouse can be used now, but the result is the same as my original one. So sad!

PS/2 mouse should have IRQ12, what I recall. Do you have a PS/2 keyboard as well?
If the test bios actually turned off the mouse port, a cable splitter from the keyboard port might work.

Rwolf wrote on 2023-05-25, 09:56:

...maybe the physical connection is bad

Do you know how can i check the physical connection by myself? Is it a chip on the mainboard?

Hmm...it would go from one of the board chips yes.

If you look at https://pinoutguide.com/Slots/PCI_pinout.shtml
you will find the pins for INT A/B/C/D which are the names for the interrupt lines on the connectors, and each interrupt should be connected in parallel to all the PCI connectors on the board - you had two PCI slots.

These INT lines are mapped in the interrupt controller to whatever IRQ number you want to actually use.

On the older mixed ISA/PCI motherboards these were sometimes mapped with jumpers, but your board is a newer one, unless I misread some info online.

Anyway, it should go somewhere to the chip containing the interrupt controller - which I'm not sure which one it is, we need a manual for the chipset for that; maybe it is possible to find somewhere.

Edit: The chipset is Cx5510 according to an archived polish test page I found: https://www.chip.pl/1998/08/maly-moze-niewiele

Block diagram: https://www.pctechguide.com/cyrix-cpus/cyrix-mediagx
More details: https://halfhill.com/byte/1998-2_cover_system.html
Less interesting perhaps, details from the release: https://www.ardent-tool.com/CPU/docs/MPR/110301.pdf

It looks like the pins are accessible for probing on the bridge chip, being a quad flat pack, so it should be possible to verify an electrical connection with a simple resistance tester.

Rwolf wrote on 2023-05-26, 13:28:

It looks like the pins are accessible for probing on the bridge chip, being a quad flat pack, so it should be possible to verify an electrical connection with a simple resistance tester.

This is beyond my knowledge for now, but I do have the plan to learn hardware in the future.
However, your suggestion remind me of something. The south bridge chip of my modern laptop did fail 2 years ago and the repair technician fixed the machine by replacing the chip. Physical connections doesn't break very often but the chip itself kind of does. Maybe I should replace the chip to fix the IRQ11 issue. That's not possible to do by myself.

It is not trivial to replace a QFP package like that bridge, and I would rather suggest to find another working motherboard if that was the problem.
Maybe someone else has an identical motherboard to compare with?

Rwolf wrote on 2023-05-27, 23:19:

It is not trivial to replace a QFP package like that bridge, and I would rather suggest to find another working motherboard if that was the problem.
Maybe someone else has an identical motherboard to compare with?

I think I have to live with it for now. I suspect the failure may have something to do with this NIC. I installed it before, the problem with it is that when plugged by an ethernet cable, my pc wouldn't boot. If the cable was plugged when Windows was running, it simply froze the system. By default, the BIOS always assigns IRQ11 to network adapters. Maybe the presence of this doggy NIC burnt out the chip, though I'm not sure about it.

Hard to tell...could be something is shorted on the NIC that affected that specific INT line. (Hard to see with this picture, but maybe check for near pins shorted on that QFP package; there could be something there. Most likely not...with a lockup on cable attachment, the problem is possibly closer to the connector side - did it release the lockup when disconnecting the cable, without a reboot?) (Looks like a clone version of the Realtek RTL8139D chip, so an equivalent card would be possible to find)
The board is partially populated at the transformer, but that is probably to save on cost with some unused RJ pins, nothing obvious, except perhaps the missing safety cap that is normally close by the RJ connector.

Even if the bridge INT port was damaged, the IRQ11 itself should be mappable in the BIOS to another physical INT line, and usable in any other slot, if just a single line/slot was affected.

Rwolf wrote on 2023-05-28, 13:15:

did it release the lockup when disconnecting the cable, without a reboot?

No.

Rwolf wrote on 2023-05-28, 13:15:

The board is partially populated at the transformer, but that is probably to save on cost with some unused RJ pins, nothing obvious, except perhaps the missing safety cap that is normally close by the RJ connector.

This thingy was literally a cheap one, probably less than $2 USD when bought. And yes, I found the cap you mentioned on a similar RTL3194D card.

Rwolf wrote on 2023-05-28, 13:15:

Even if the bridge INT port was damaged, the IRQ11 itself should be mappable in the BIOS to another physical INT line, and usable in any other slot, if just a single line/slot was affected.

I'm not very familiar with the INT# settings. The only related one in my BIOS settings is the one in the picture. It seems to be related only with PCI IDE controller or something like that, rather than a general PCI IRQ setting.

Well the IDE controllers have IRQ 14/15 normally - these controllers are usually both on your motherboard as seen in the block diagrams as the EIDE controller, so only in case of the use of a separate IDE controller on the PCI bus would you need to map it like the last image you posted as just an example, I understand.
(In such a case the INT line A will use IRQ14 and INT line B will use IRQ15 - they could be used in the same PCI slot if the controller card has two IDE buses.)

To manually adjust more detailed mapping would probably need another BIOS, as you have already tried.