soviet conscript wrote on 2021-12-25, 08:00:

I've always read about how 486 motherboards that used PCI had issues because they mostly used an early implementation of PCI. I've also relayed this information to others as I've heard it from multiple sources but after using many PCI 486 boards and not having any issues and reading comments from others that have never had issues I'm left wondering myself. what are these issues that early use of PCI had on 486 motherboards? is it compatibility issues with some cards or just slots not running as fast as they could/should? does anyone have any specific examples? I do recall may be having a few 486 PCI motherboards where some PCI video cards just refused to work but I guess I just blamed it on other possible factors at the time.

- 5V slots only - so, no support for 3.3V cards.
- no PnP PCI BIOS support, you should configure PCI interrupts, DMA or other resources manually, or semi-manually.
- PCI protocol versions - PCI ver. 1.x motherboard or cards can conflict with ver. 2.x mb/cards.
- IRQ sharing compatibility in early protocol version is not so nice, req’d such tricks as ISA paddle boards for guaranteed IRQ.

And firmware initializations on turn-on not perfectly supported too. Happened on a Micronics HX chipset motherboard. I had to move to AN430TX then more of my PCI video cards worked.

Cheers,

Another common issue is that most early PCI motherboards don't support PCI ide specification. Thus they must use compatibility hacks (such as "IDE controller installed in slot X" setting in BIOS if anything at all) to route legacy interrupts/ IO port ranges to the specific slot. Many newer PCI ide controllers, such as those SATA/RAID don't work at all.

On the other hand early PCI boards typically have IDE controller integrated already. However, often it is a buggy CMD640/RZ1000 or alike.