maxtherabbit wrote on 2026-05-19, 12:51:

I like it how it is in the picture, but I don't like the weird abbreviations. Keep the help and use full words in the setting names

Yeah, the weird abbreviations are the whole reason I did not implement more chipsets before doing the new setting UI.

Now that the UI is fixed, I have to go through all the current supported chipsets and fix them all, then I can add more.

5/22/26 : Updated UCP2 to 2.0.2
- Changed the option page layout to optimise available space for option strings
- Removed silly abbreviations as they are no longer needed
- Added a few SiS chips (5511/2/3, 5571/2, 5581/2, 5591/2, 5595)
- Update save format (Older saves are converted)

Is this tool similar to a mating between TweakBIOS and PCCHIP34? It would be nice to see support for all three popular 486 PCI chipsets from UMC, SiS, and ALi.

feipoa wrote on 2026-05-22, 06:56:

Is this tool similar to a mating between TweakBIOS and PCCHIP34? It would be nice to see support for all three popular 486 PCI chipsets from UMC, SiS, and ALi.

Yes it is.
As for the many 486, I already implemented the code for ISA chip I/O configuration. The only two things missing is little code to detect the absence of PCI so it switch to ISA chipset mode and a little menu to select said ISA chipset on the first launch.

I also have to patch the ISA stuff so it actually probe if said device exist at launch (Seam to be possible, but different for each chipsets). To prevent patching the wrong chip if someone move an hard drive or select the wrong chipset. For PCI I simply scan all PCI devices and detect if the saved config match what's actually present in the system.

I just fast-read M1489 docs, and now I think it's not possible to properly ID the ISA chipset to confirm if it's present in the system I might have to figure out other way... Maybe instead, trying to detect if it's the same system. For sure I can't just checksum F000-FFFF as if someone using things like QEMM Stealth mode, that's going to be unreliable.

Maybe checksumming the "CMOS"... If it's different ask to user to confirm it's the same system/chipset or something... If someone have a better idea.

CTCHIP34 makes the user select the chipset for 3/486 boards. I suspect you will have to do the same. You could just do a popup menu at program start with the default option being "PCI" and have the various ISA/VLB chipsets below that

maxtherabbit wrote on 2026-05-22, 14:51:

CTCHIP34 makes the user select the chipset for 3/486 boards. I suspect you will have to do the same. You could just do a popup menu at program start with the default option being "PCI" and have the various ISA/VLB chipsets below that

Yes, exactly, and unlike CTCHIP34, UCP2 save the settings and next time you use it in continue with these.
Asking for which ISA chipset on first launch is the idea, but on next launch I am figuring out a way to detect if it's the same system or no (Does it load from the save or ask you to clear the config (ucp2 /clr ))

As for asking for PCI or ISA, there is no need, I can simply fallback to ask for ISA chipsets if the BIOS don't support PCI. Apparently BIOS function B101h of INT 1Ah will return EDX = 20494350h (' ICP') if the PCI Configuration Mode is available. 486 BIOS that have PCI support can be autodetected. Even like the Ali M1489, who is fully "IO Port configurable", Does have a presence on PCI bus and can be automatically identified.

5/25/26 : Updated to 2.0.4
- Allow system without PCI to select there ISA/VL chipset from a popup on first launch, with a list of ISA chipsets
- Updated the save format so proper ISA support
- Added a mechanism to support slow PCI boards (the ones with PCI bridged behind a VL Bus). These are still considered PCI but do show a popup to allow user to select there ISA northbridge (It show possibly compatible chipset, for example if you have a Via 82C505, it will allow you to choose between Via 82C496G and 82C570M... It won't show Opti or Ali options.)
- Added a few chips (Ali M1429/G, Ali M1489, Opti 802G, Opti 822, Opti 895, SiS 471, SiS 496/497, Via 82C496, Via 82C570M, Via 82C505, um888*)

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

So far I did not put any power saving stuff in most chips (only some on some). Mainly because there lot of options and I don't know how useful that is (do people just disable that anyway).
I wanted to support more chipset first, faster, than add extra stuff if people want it.

Should I began to add power saving stuff in the currently supported chip or I forget it for now?

Hi, @Rav,
Personally I think it will be easier to add all features for the chipset, when your're adding the new chipset support, than first adding support for a huge number of chipsets and tweking them one by one after. But you best know which way would be easier and less time consuming for you.

aVd wrote on 2026-05-26, 06:31:

Hi, @Rav,
Personally I think it will be easier to add all features for the chipset, when your're adding the new chipset support, than first adding support for a huge number of chipsets and tweking them one by one after. But you best know which way would be easier and less time consuming for you.

It's probably the same (except the amount of time used per chipset). Pretty much all the chip I added have power saving stuff in them so I can go back to all of them to add it later. The current configuration saving feature support adding, moving, and deleting options already (if you use the software to configure your chipset in the current version, and you upgrade at a later date to a version that add power saving options, then it will still load all configured stuff properly and will also show the power saving options with there default settings).

I initially wanted to have like about 20 supported chipsets to tweak first to test and get feedback on the software first, so more people can use it and get an idea. I think I have more than 20 chipsets supported now so if people actually want it then I can begin to add the power saving stuff.

In most chipset (except the poorly documented one), the power saving stuff easily have like 2-5X more configuration stuff compared to all the rest of the configuration option, if not more.

But first I finish to add the two new features I am working on right now:
- INS/DEL Device (allow users to add/remove non autodetected stuff, like that Cyrix 5x86)
- Function to flush all the caches, so I can add cache settings that make the cache incoherent when changed

Unfortunately, I don't have the Intel 420 chipset datasheet.

I didn't see documented support listed for the UM8881/8886. This is one of the more popular 486 PCI chipsets used around here. If you are interested, someone reversed engineered the datasheet for this chipset, located here: https://github.com/pc2005cz/um888x_documentat … /tag/2026-02-16

feipoa wrote on 2026-05-26, 07:13:

Unfortunately, I don't have the Intel 420 chipset datasheet.

I didn't see documented support listed for the UM8881/8886. This is one of the more popular 486 PCI chipsets used around here. If you are interested, someone reversed engineered the datasheet for this chipset, located here: https://github.com/pc2005cz/um888x_documentat … /tag/2026-02-16

Oh my bad
UM8881F/8886* are supported in UCP2 I added it yesterday, I just... managed to skip it when I added the new supported chips to the list...
Only unsupported 8886 related thing is the 8886AF IDE part as it's ISA and there is no DID difference between 8886A and 8886AF

I indeed got the specs from pc2005cz.

Ahh, that's good news. This looks promising.

Rav wrote on 2026-05-26, 06:48:

I initially wanted to have like about 20 supported chipsets to tweak first to test and get feedback on the software first, so mo […]
Show full quote

I initially wanted to have like about 20 supported chipsets to tweak first to test and get feedback on the software first, so more people can use it and get an idea. I think I have more than 20 chipsets supported now so if people actually want it then I can begin to add the power saving stuff.

In most chipset (except the poorly documented one), the power saving stuff easily have like 2-5X more configuration stuff compared to all the rest of the configuration option, if not more.

But first I finish to add the two new features I am working on right now:
- INS/DEL Device (allow users to add/remove non autodetected stuff, like that Cyrix 5x86)
- Function to flush all the caches, so I can add cache settings that make the cache incoherent when changed

I agree. Now you have support some of the most popular and widely used chipsets of the '90s, so it's time to add all the tweaking features for them, so the users can provide feedback if the current UCP version works fine with them.

5-28-26 : Updated to 2.0.5
- Fixed minor bugs
- Fixed the main config loading loop (it only affected users with 2 configured ISA devices or more, so... no one!)
- Fixed a race condition for loading configuration on pure isa hardware (ugly fix as using volatile and memory barriers did not work)
- Added ISA/IO Ports devices (UMC UM8886AF and Cyrix 5x86)

To add ISA device or a CPU, one must press the "INSERT" key in the main menu and choose said device from the list as theses are not autodetected. CPU Autodetect might come later.

I did not test theses two additions

Cyrix is eratic on 86box, some configuration have more editable registers, so I guess register changes are not properly implemented on the emulator. My Acer with a 5x86 is out of commission at this moment and While I do have a M919, it's missing the power transistors... Maybe this weekend I will fix one of theses two to test the cyrix properly assuming I don't get positive feedback

8886AF, well, no machine have that in 86box look like, can't test. My M919 have a 8886BF so that one have an integrated PCI IDE controller... Need a board with the 8886AF to test that ISA Integrated IDE controller. Maybe some M919 have one, who know (86BOX's M919 have a 8886A and when I try the IDE part it does funny things...

I got the soldering stuff out and finally fixed the M919 (Been like 3 years it's waiting to get fixed).

I confirm, there is something wrong with the cyrix. Do not toy with it until next release (release won't append until I fix it)
There some other issue with ISA look like (might be linked as they use pretty much the same code path.)

Ok, for the "ISA" stuff, that was confusion, it's a different thing from poorly documented 8881/8886 chipsets
It look like the 8886BF on my M919 is a 1060:886A (UMC8886A) + 1060:673A (UMC8886BF). It's in fact a single chip but it look like the BF chunk is in fact only the PCI IDE controller while the rest report as a 8886A
So there is 3 pci devices for the set of two chips (It's not uncommon, many other PCI chipset contain many PCI devices internally but the other documentations I worked on where pretty clear).

I have to fix the 8886 stuff so the BF only include the IDE chunks and nothing else.

Dont use on 8886 based board until next release.

5-29-26 : Updated to 2.0.5.1
- Hotfix (Cyrix 5x86)
- Hotfix (UMC 8886BF is a 8886A coupled with an extra internal PCI IDE controller on a "different" PCI device)

All is fixed
Funny how my Cyrix 5x86 Can't withstand "Linear Address Burst Cycles" (the feipoa optimized cyrix settings). Tested in both my Acer a few years ago and my freshly fixed M919. In both case, it reboot.

Been testing setting with 8881, seam to work, I optimised my FPM for 60Mhz BUS operation.... Now I want EDO.... BUT, the option to set PCI speed don't seam to work, Any setting I set do nothing different while the video card memory speed increase when I increase the FBS. Maybe it's locked after being set or something.