Hi fix_metal,

I had some spare time this evening so I patched your Rev. 1.4 BIOS right away.
Here is a copy and I’ve packed a PATCH.TXT file in the zip with some info.

While I was at it, fixing the Win98 UDMA bug, I fixed the 32GB HDD limit bug as well. 😀

Other patches, like for K6-2+/III+ support, take a lot more time and are not included in this patch J.1 BIOS. This extended CPU support is only useful if you can get the Vcore supply down to the required 2.1V/2.0V, either via a board mod, undocumented jumper settings, or an interposer.

Please let us know how the patched BIOS works.

@douglar, when we have a positive report on the patch J.1 BIOS, you can put it on TRW. Thanks.

Cheers, Jan

Chkcpu wrote on 2025-01-24, 16:14:

Hi fix_metal, […]
Show full quote

fix_metal wrote on 2025-01-23, 20:36:

Unless anyone has any better ideas I guess I'll just leave it be no DMA as it's always been. So sad.

Hi fix_metal,

Yes I have a better idea, patch the offending BIOS code by hand to fix this annoying bug. 😉

The link douglar posted earlier in this thread was spot-on and the issue you are facing is indeed a BIOS bug, but often called the “Win98 UDMA bug”. This bug is present in all Award BIOSes from before 03/1999 for chipsets that support UDMA. In this timeframe that would mainly affect socket 7 boards with the i430TX chipset, but I’ve seen affected boards with Ali, SiS, and VIA chipsets as well.

As Award fixed this bug in March 1999, most board manufacturers from that era published a BIOS update to fix this bug. But with the latest 03/03/1998 BIOS for your SYE-P55TX board you seem to be out of luck here.

Strange that the BIOS Patcher tool didn’t work on your BIOS, but I can patch this bug by hand and fix the Rev 1.4 BIOS for you. I expect to have time for that this weekend.

I will report back when the BIOS is ready.
Cheers, Jan

Wow. Yes, I've read about the bug at [0]

Thanks so much for the service!!!

[0] https://tldp.org/HOWTO/Large-Disk-HOWTO-4.html

Chkcpu wrote on 2025-01-24, 21:33:

Hi fix_metal, […]
Show full quote

Hi fix_metal,

I had some spare time this evening so I patched your Rev. 1.4 BIOS right away.
Here is a copy and I’ve packed a PATCH.TXT file in the zip with some info.

Filename

P55TX_J1.zip

File size

99.77 KiB

Downloads

10 downloads

File comment

Lucky Tech SYE-P55TX patch J.1 BIOS

File license

Public domain

While I was at it, fixing the Win98 UDMA bug, I fixed the 32GB HDD limit bug as well. 😀

Other patches, like for K6-2+/III+ support, take a lot more time and are not included in this patch J.1 BIOS. This extended CPU support is only useful if you can get the Vcore supply down to the required 2.1V/2.0V, either via a board mod, undocumented jumper settings, or an interposer.

Please let us know how the patched BIOS works.

@douglar, when we have a positive report on the patch J.1 BIOS, you can put it on TRW. Thanks.

Cheers, Jan

Dear Jan,

Again, thanks so much for the service. I will test and report back asap.

Hello Jan,
I've flashed the bios. Unfortunately, it doesn't work either. I can see the new bios version, but win98 still won't boot when enabling DMA. I will attach some photos soon. One thing worth mentioning is that after the POST I the original bios reported the disk as UDMA5, while after the upgrade is reported as UDMA2. Is this part of the bios mod?

fix_metal wrote on 2025-01-25, 16:14:

Hello Jan,
I've flashed the bios. Unfortunately, it doesn't work either. I can see the new bios version, but win98 still won't boot when enabling DMA. I will attach some photos soon. One thing worth mentioning is that after the POST I the original bios reported the disk as UDMA5, while after the upgrade is reported as UDMA2. Is this part of the bios mod?

Yes, the BIOS indicating UDMA 5 on an i430TX board is a tell tale sign of the Win98 UDMA bug, and seeing UDMA 2 now tells me that the BIOS mod is working correctly.

I used this mod successfully on numerous BIOSes and when I put this mod in the 1998 BIOS of my 430TX board (Abit PX5) it worked right away. I had a previously fresh Win98 install, so I was using Microsoft’s Standard busmaster IDE drivers that came with Win98. They are very stable and the ATTO Disk benchmark increased from 9.5MB/s to a nice 30MB/s after the mod and enabling DMA in Device Manager.

What drivers are you having installed now?

Jan

Chkcpu wrote on 2025-01-25, 16:45:

Yes, the BIOS indicating UDMA 5 on an i430TX board is a tell tale sign of the Win98 UDMA bug, and seeing UDMA 2 now tells me tha […]
Show full quote

fix_metal wrote on 2025-01-25, 16:14:

Hello Jan,
I've flashed the bios. Unfortunately, it doesn't work either. I can see the new bios version, but win98 still won't boot when enabling DMA. I will attach some photos soon. One thing worth mentioning is that after the POST I the original bios reported the disk as UDMA5, while after the upgrade is reported as UDMA2. Is this part of the bios mod?

Yes, the BIOS indicating UDMA 5 on an i430TX board is a tell tale sign of the Win98 UDMA bug, and seeing UDMA 2 now tells me that the BIOS mod is working correctly.

I used this mod successfully on numerous BIOSes and when I put this mod in the 1998 BIOS of my 430TX board (Abit PX5) it worked right away. I had a previously fresh Win98 install, so I was using Microsoft’s Standard busmaster IDE drivers that came with Win98. They are very stable and the ATTO Disk benchmark increased from 9.5MB/s to a nice 30MB/s after the mod and enabling DMA in Device Manager.

What drivers are you having installed now?

Jan

Thanks for the effort.

I've installed Intel's original chipset driver, version 3.4.0. They also install IDE Bus master driver for the south bridge.
I'm not entirely sure a whole fresh installation would change anything at this point, it might be worth a shot...

Thanks for the pictures, they look exactly as they should.

Did you try uninstalling the Intel drivers, so Win98 reverts to its MS drivers?
Maybe that is all you need to avoid a clean re-install of Win98SE.

Jan

Chkcpu wrote on 2025-01-25, 21:33:

Thanks for the pictures, they look exactly as they should. […]
Show full quote

Thanks for the pictures, they look exactly as they should.

Did you try uninstalling the Intel drivers, so Win98 reverts to its MS drivers?
Maybe that is all you need to avoid a clean re-install of Win98SE.

Jan

I've just tested after fresh reinstall:
When disabling DMA from the bios, win98 will boot and DMA is ticked.
When enabling in the bios, it hangs during startup as usual.

So yea, same asylum as before unfortunately.

fix_metal wrote on 2025-01-25, 22:05:

When disabling DMA from the bios, win98 will boot and DMA is ticked.

^^^ Isn't that the end state that you want? DMA ticked in Win98?

douglar wrote on 2025-01-26, 02:48:

fix_metal wrote on 2025-01-25, 22:05:

When disabling DMA from the bios, win98 will boot and DMA is ticked.

^^^ Isn't that the end state that you want? DMA ticked in Win98?

I assume that's a false positive as if the bios isn't providing the facility how can it work in the end? To be 100% sure I should benchmark.

fix_metal wrote on 2025-01-26, 08:02:

douglar wrote on 2025-01-26, 02:48:

fix_metal wrote on 2025-01-25, 22:05:

When disabling DMA from the bios, win98 will boot and DMA is ticked.

^^^ Isn't that the end state that you want? DMA ticked in Win98?

I assume that's a false positive as if the bios isn't providing the facility how can it work in the end? To be 100% sure I should benchmark.

Confirmed, ATTO benchmark shows the disk is benchmarking the same results when DMA is or ir not ticked in WIN98 (and disabled in BIOS).

There is more to this DMA issue than meets the eye. 😉

After a fresh install, Win98 will operate in PIO mode 4 and the DMA box in Device Manager is not-ticked, for maximum compatability.
When you now disable UDMA in the BIOS, tick the DMA box in Windows and reboot, Win98 will switch to Multiword DMA mode 2.
Although both PIO4 and MWDMA2 have a theoretical transfer rate of 16.6MB/s, MWDMA is actually faster because it has less overhead. MWDMA2 usually benchmarks at 11 to 14MB/s as opposed to the 9MB/s in PIO4.

MWDMA2 is what you want on non-UDMA capable chipsets like the i430VX and i430HX, but on the i430TX with its improved PIIX4 southbridge, you of course want the double 30MB/s transferspeed of Ultra DMA mode 2!

Reading back through the thread, I was again puzzled by the hang during Win98 loading when UDMA in the BIOS is Enabled or Auto. I never experienced that on i430TX boards. Even with the Win98 UDMA bug still present in the BIOS, Windows just ignores the ticked DMA box and starts normally in PIO mode 4. When you then check Device Manager, you see that the DMA box is “un-ticked” again. That’s why the Win98 UDMA bug often results in a “DMA box selection doesn’t stick” complaint.

@fix_metal, So there must be something else amiss here…
The ATTO Benchmark should also have show a difference between PIO4 and MWDMA2 modes.
UDMA2 doubles the datarate of the IDE signals with respect to PIO4 and MWDMA2, so apart from replacing the crucial components of the IDE signalpath, like flatcable or Harddisk, I’m presently out of ideas. Note that a 80-wire IDE cable is not required here, but is only needed for UDMA4 (66MB/s) or faster.

Regards, Jan

Chkcpu wrote on 2025-01-26, 18:41:

There is more to this DMA issue than meets the eye. ;) […]
Show full quote

There is more to this DMA issue than meets the eye. 😉

After a fresh install, Win98 will operate in PIO mode 4 and the DMA box in Device Manager is not-ticked, for maximum compatability.
When you now disable UDMA in the BIOS, tick the DMA box in Windows and reboot, Win98 will switch to Multiword DMA mode 2.
Although both PIO4 and MWDMA2 have a theoretical transfer rate of 16.6MB/s, MWDMA is actually faster because it has less overhead. MWDMA2 usually benchmarks at 11 to 14MB/s as opposed to the 9MB/s in PIO4.

MWDMA2 is what you want on non-UDMA capable chipsets like the i430VX and i430HX, but on the i430TX with its improved PIIX4 southbridge, you of course want the double 30MB/s transferspeed of Ultra DMA mode 2!

Reading back through the thread, I was again puzzled by the hang during Win98 loading when UDMA in the BIOS is Enabled or Auto. I never experienced that on i430TX boards. Even with the Win98 UDMA bug still present in the BIOS, Windows just ignores the ticked DMA box and starts normally in PIO mode 4. When you then check Device Manager, you see that the DMA box is “un-ticked” again. That’s why the Win98 UDMA bug often results in a “DMA box selection doesn’t stick” complaint.

@fix_metal, So there must be something else amiss here…
The ATTO Benchmark should also have show a difference between PIO4 and MWDMA2 modes.
UDMA2 doubles the datarate of the IDE signals with respect to PIO4 and MWDMA2, so apart from replacing the crucial components of the IDE signalpath, like flatcable or Harddisk, I’m presently out of ideas. Note that a 80-wire IDE cable is not required here, but is only needed for UDMA4 (66MB/s) or faster.

Regards, Jan

Well, the thing is BIOS will just allow Auto or Disabled options when it comes to UDMA. As for PIO it allows you to force to any specific mode. So, when I tick DMA option in Win98 and I disable UDMA in BIOS, windows boots and DMA option in it stays active, although it's not for real as Atto benchmark confirmed.
On the other way around, if I leave DMA option enabled in win98 and actually set to AUTO in the BIOS, windows will just fail to boot.
If I set it to AUTO in the bios, and leave the tick disabled in win98, it boots just fine.

I don't know. Really this is the only setup over dozens I've built over the years that never really worked with DMA. It might be interesting trying this with Windows ME just to see the difference (if Windows ME wasn't more like meh).

A quick update: I found a 1.3GB HDD for my 486, so I will cascade-swap disks and put a 8GB drive in this P166 DMA bugged mobo. It might start working with a proper compatible UDMA HDD. Stay tuned.

And nothing even this time. Now the PC is running a compatible 8G hdd, and still refuses to boot while DMA is turned on. Still using the patched bios.
My last chance would be to swap memory modules I guess.

Just tested the RAM with memtest86 and it passed 6 tests in a row.

Tested the PC with another DIMM bank, no luck.

Tested the PC with 2 x 8MB dual side 72pin banks, no luck.

Beats me, it really seems a buggy/faulty Mobo, or either I'm missing something with jumpers/something else. I've been trying tweaking BIOS settings for a while now, but I still don't see any light.

It might be admitting defeat, but you could add a pci ide card.

The promise ultra66 cards are pretty nice and would be an upgrade from a fully functional piix3 or piix4.

douglar wrote on 2025-02-28, 04:12:

It might be admitting defeat, but you could add a pci ide card.

The promise ultra66 cards are pretty nice and would be an upgrade from a fully functional piix3 or piix4.

Ah, see, I wouldn't have thought of the chance. Thanks for the idea, I found one and I will give it a last try.

Did it! I bought a promise fast track 100 tx2 and it now goes as up as 25MB/s!
Thanks for the hint, I wouldn't have though of that, never used PCI IDE cards.