Correction, adjusting the PCI clock by messing with the UM8886 work!
Thanks mr. oscilloscope for the confirmation!

ok, I will stop posting my M919 benching 🤣... Been testing that thing the whole evening

Anybody can fix the 440LX/440EX not supporting SSE CPU-s or know why it cant be used with 66FSB SSE CPU?
They can boot up to with Celeron 766Mhz with Slot1 to Socket370 or directly at Socket370.Celeron 533 (MMX)
works without issues and it is maximum Celeron CPU. Tiny OC possible, but would love to have it to have at SSE.

I recall the intel 440LX board working happily with Pentium 2 233MHz. Never tested one with later models. The cpus are practically identical and it probably needs updated CPU list in bios.
Intel boards are very pickle with cpu speeds and what they support.

Rav wrote on 2026-05-26, 05:35:

I am looking for Intel 420* Documentation (there 486 PCI offering. I only found for the SIO...)
If anyone have any pertinent informations

Does this help?

ChrisK wrote on 2026-07-01, 15:44:

Rav wrote on 2026-05-26, 05:35:

I am looking for Intel 420* Documentation (there 486 PCI offering. I only found for the SIO...)
If anyone have any pertinent informations

Does this help?

YES!
Thank you!

You're welcome!
😀

Hi. It looks like UCP2 isn't working with the Ali M1429G chipset. The settings aren't applied after rebooting with the UCP 2EXE /autoload switch. I tested it using your previous tool, UCP.EXE /chip ali1429, which works perfectly.

Pr1ha wrote on 2026-07-28, 17:52:

Hi. It looks like UCP2 isn't working with the Ali M1429G chipset. The settings aren't applied after rebooting with the UCP 2EXE /autoload switch. I tested it using your previous tool, UCP.EXE /chip ali1429, which works perfectly.

Thanks for letting me know. I do use ucp2 / autoload on my system here but it's a PCI based system. There might be something wrong with the ISA one. I take a look asap and will include the fix in the next revision.

I did test on my Acer that do have a 1429G.

There seam to be something sus at the level of inp/outp, it work totally fine for the cyrix (just tested on real hardware too, and it's probably why I did not catch the issue as I did not test any other real in/outp hardware), and /autoload do work, but I think it's liked to some "delay" shenanigans that are required between theses commands. I used to have them in my real mode ucp, but the new one I do not, I did try to add delay (using delay()) command but it does not seam to work. Not all registers properly update. I probably have to convert the "ISA/IOPORT" path to assembler to have proper control over the code and delays.

I could detect unrequested register "change" (like if the read did not read the proper value a second time) and also the opposite, not all registers apply.

I know that some 286 and 386 need such delay, did not expect the 1429G to have that problem.

Pr1ha wrote on 2026-07-28, 17:52:

Hi. It looks like UCP2 isn't working with the Ali M1429G chipset. The settings aren't applied after rebooting with the UCP 2EXE /autoload switch. I tested it using your previous tool, UCP.EXE /chip ali1429, which works perfectly.

Here the alpha release for the next version, it include two fix to address your issue (Converting the ISA/IOPORT code to ASM, And fixing the chipset file for the ALi 1429/G as I did indeed fsckup (I ported that file initially from the original real mode ucp and the way it work is literally upside down.. Made a few conversion errors... oops).

The /autoload always worked, it was just the register updating that was not properly applied.
Please delete the ucp.cfg file (where you run ucp2.exe) as it's bogus.
Also I recommand you to run it before other chip patching (so don't run the old real mode UCP before, do it on a clean boot). That is only cosmetic if you want to have the proper default displayed in the program.

Note : Do not use if you have a PCI S3 Virge or Trio 3D. I broke that part by updating the isa/ioport codepath but it should work totally fine for everything motherboard including the one using IOPORTS and INCLUDING your ALi 1429G based motherboard. That version also include a few other minor fix, full implementation of ALi 1531/33 (Including all power saving stuff), and a **Partial** Implementation of the Intel 82420EX.
I should release a proper Q&Aed version this weekend (including reviving the PCI S3 video cards stuff and finalising the Intel chipset)

Rav wrote on 2026-07-30, 02:21:

The /autoload always worked, it was just the register updating that was not properly applied.
Please delete the ucp.cfg file (where you run ucp2.exe) as it's bogus.
Also I recommand you to run it before other chip patching (so don't run the old real mode UCP before, do it on a clean boot). That is only cosmetic if you want to have the proper default displayed in the program.

Thanks! I tested it, and it now works with the new alpha version of UCP2.

Can you tell me how to enable L1 WB using UCP2 when using a Cyrix 5x86 on an ALi M1429/G? The BIOS doesn't recognize the processor and can't correctly set bits 4 and 2 in register 19h.

Pr1ha wrote on Today, 08:43:

Can you tell me how to enable L1 WB using UCP2 when using a Cyrix 5x86 on an ALi M1429/G? The BIOS doesn't recognize the processor and can't correctly set bits 4 and 2 in register 19h.

Changing cache mode is not yet supported. It's the only missing thing that is in the old UCP but not in UCP2. But now that I did revive my Ali1429G 486, maybe I will implement that soon (I need to reimplement the cache purging dance and test it properly). I will let you know when it will be available.