Did more research. The Promise SATA300 cards do not have Windows 98 drivers so not suitable for this build!
I found an 80 gig WD Caviar SE in my pile, it's SATA I. There is the Promise SATA150 series which will work properly with Windows 9x.
However I also saw one person using a Sil3114 card on an AMD K6-III+ system under Windows NT. Those cards are cheap but some people say they're not good for Socket 7 and below

luRaichu wrote on 2025-05-12, 21:02:

However I also saw one person using a Sil3114 card on an AMD K6-III+ system under Windows NT. Those cards are cheap but some people say they're not good for Socket 7 and below

I am not sure if this is the exact issue you are suggesting but here goes.

I have one of those in my SE440BX-2 (Slot 1, Phoenix BIOS) and it works perfectly.

When I tried in my P5HX-A (Socket 7, Award BIOS 4.51PG) I encountered the following issue. Everything worked with loading the option ROM except when it was time to boot, somehow the Number of Hard Disks Attached byte in the BIOS Data Area (40:75) was zero. That blocked the BIOS from attempting to boot.

Using a little MBR on a floppy disk that would set that byte to 1, wait for any key so as to eject the disk, then attempt booting Int 19h again, it worked great as a proof-of-concept, demonstrating that nothing else was wrong.

I did find that having no IDE devices at all attached slowed down booting of the Award BIOS, even with everything set to "none." I think this is because for El Torito/Zip drive booting purposes, it probes for ATAPI devices anyway.

In a direct conversation with mkarcher about this, we wondered if it was because the BIOS Boot Specification was phasing in at this time, and if the SIL311x BIOS did not have the right fallback for a system BIOS that doesn't have it. (BBS is what enables that fancy boot order listing in the BIOS rather than just things like A,C and C,A). I am not sure whether those later Award BIOSes (that say A,C,SCSI vs. A,C) have this support or not. Anyway, maybe this is where the idea they aren't good on Socket 7 comes from.

What about a VIA VT6421A card for my system? Seems popular with the Xbox modding people

My Sil3114 works well in certain systems, not at all in others. However, one of the systems it works well in has an Asus TXP4 which is also an AT motherboard with the TX chipset.
Here are the ATTO benchmarks with a K6-3+ @ 500 and an SSD with a Promise Ultra 100 and a Sil3114 flashed to the base (non RAID) BIOS.

Good to know. Does your Asus board use the same version of the Award BIOS as mine?
Just want to make sure it’ll be compatible before I go and buy a Sil3114

luRaichu wrote on 2025-05-13, 15:39:

Good to know. Does your Asus board use the same version of the Award BIOS as mine?
Just want to make sure it’ll be compatible before I go and buy a Sil3114

It's an Award 4.51, though I use the patched BIOS that Jan Steunebrink provided.

Whoops, I forgot I’ll be using a PATA CD-ROM drive!
Does the onboard IDE controller on 430TX chipset play nice with the Sil3114? If not I need a VT6421A

I’ve installed this motherboard in the desktop case, although a non-standard mounting standoff was shorting out the SIMM sockets from the underside and preventing a POST. I covered them with clear tape.
Unfortunately this case requires rails for mounting 5.25” drives. I only had one pair.

Then I found this thread where one member mentioned using thermoplastic to mold new rails. I bought 11oz of this stuff on Amazon for $6 and it turned out alright.

To create the mold I put one rail on a smooth surface (glass) with the bracket facing up, then covered it with the molten plastic to create a mound. It can’t be too thick. The metal rail should be removed from the mold right before it hardens completely.
The same plastic is used for creating new rails in the mold. I applied mineral oil to prevent the hot plastic from sticking to the mold. But other oils should work as well. A paint scraper (or equivalent) helps to get the plastic flush with the mold.

The resulting plastic rails fit fine and snug despite not having the bracket (too small to mold).
Since I’ll be using two 5.25” drives, I put a metal rail on one side and a plastic rail on the other. Reverse sides for other drive.

But you can’t win them all. The ATI Rage II card that was tested and functional here doesn’t want to output video anymore (both VGA and NTSC composite). Windows will recognize it, load the driver, etc. but no video. I’ve only rarely seen it working now.
Whatever. They say it wasn’t good at playing Quake anyways.

Before cleanup:

After:

New VT6421A card installed but the BIOS does not list any devices attached to it. The included driver disc came with a big scratch which makes it unreadable 😀

According to its datasheet the VIA VT6421A uses PCI 2.2 while the motherboard only supports PCI 2.1. So I don’t think it will work.
I guess I will have to buy a Sil3114 card and use an IDE to SATA adapter for the CD-ROM drive.