Nemo1985 wrote on 2021-04-17, 10:51:

appiah4 wrote on 2021-04-17, 07:24:

Nemo1985 wrote on 2021-04-16, 21:46:

My motherboard is just like yours, just a bit older (95 date code), do you have a umc cpu to try by any chance?

I have a UMC U5S-33, do you want me to try it? I don't see why it wouldn't run, UH19 has an MTL jumper setting document with U5S listed there, have you tried these?

http://www.win3x.org/uh19/public/motherboard/ … anual/33492.pdf

On my motherboard it works fine, but according to the pdf made from Feipoa (which is better than common manual) says: RN11 is always open on MB-8433UUD versions 2 & 3. RN11 needs to be closed for U5S and U5SX.
And in fact it is open on our rev but still on mine mb works fine.

What feipoa’s manual states is in accord with the MTL manual I linked above and even if it were not I would trust his word on that regardless.

Yeah, that comment on that RN11 was from another source. I still haven't verified this. Perhaps the v2 and v3 boards found another way to implement U5SX support.

This is how it is described in an original manual:

Is this motherboard known to be quirky with CD-ROM drives? A previously working drive I hooked up to it has caused all kinds of shit to go haywire - hard drive gets detected at wrong capacity (regardless of being on the same chain or not), boot hangs, nothing gets detected.. Regardless, the CD-ROM never gets detected. I'll try with another optical drive soon, just to make sure mine didn't kick the bucket, but has anyone had similar experiences in the past?

appiah4 wrote on 2021-04-17, 17:30:

Is this motherboard known to be quirky with CD-ROM drives? A previously working drive I hooked up to it has caused all kinds of shit to go haywire - hard drive gets detected at wrong capacity (regardless of being on the same chain or not), boot hangs, nothing gets detected.. Regardless, the CD-ROM never gets detected. I'll try with another optical drive soon, just to make sure mine didn't kick the bucket, but has anyone had similar experiences in the past?

I've tried with two other CD-ROMs, same behavior. Whenever there is only the CF-IDE adapter in the system, the card is detected in the correct geometry as 2048MB, but when I add a CD-ROM it gets detected as 2014MB with the wrong geometry, and if I manually add it to BIOS then I get a Hard Drive Failure (80) error at POST.

No idea what to do here, sounds like a BIOS issue?

EDIT: Resolved by moving the CF-IDE and CD-ROM both onto the Primary Channel, disabling the secondary IDE and using a 40 stripe IDE cable instead of an 80 stripe PATA cable. I don't know which of these did it, but they are behaving now.

Please check my manual. This issue was covered.

If you are using a CF card, set the IDE Primary Master PIO to PIO-4 instead of AUTO. I have
experienced problems when booting to a CF card if AUTO is set.

IDE Primary Master PIO - Auto [CF cards do not boot properly when set to Auto. Set to PIO-4 when using CF cards]

Also, be sure to set IBC DEVSEL# Decode - Slow

In general, I use a SCSI or Promise IDE controller rather than the on-board IDE.

Yeah, I often experience issues with 80 pin cables on these older motherboards so I don't use them.

Thanks, I managed to (mostly) resolve this by changing the optical drive. Strangely enough, the LG 52x CD-ROM that came out works with other boards, so it probably has to do with some kind of quirk with the IDE Controller BIOS on this board. Not that it's working flawlessly with the Pioneer 16x DVD-ROM I put in, it hangs at PnP detection at every other soft boot or so but at least works after every reset. I've also had some data corruption on my CF card once (crosslinked files) but I'm not sure if that's going to be a common thing or not. Overall, I'm pretty pleased with how this box turned out, and after finalizing its software setup I may post about it in the Retro Rig thread. Thanks for all your help feipoa, you're an awesome chap.

If you continue to have issues with cross-linked files, you may want to go with a PCI Adaptec SCSI or Promise IDE card. Best overall is an Adaptec 2940U2W or 2940UW. They should allow for CD-ROM boot option and have working DOS drivers. The Promise Ultra100 is a nice and simple to use card, but they have issues with soft-resets on these older systems and don't have a driver for DOS so you can read the CD-ROM. Nonetheless, I use the Promise Ultra100 on at least two setups and just use the reset button.

feipoa wrote on 2021-04-19, 20:58:

If you continue to have issues with cross-linked files, you may want to go with a PCI Adaptec SCSI or Promise IDE card. Best overall is an Adaptec 2940U2W or 2940UW. They should allow for CD-ROM boot option and have working DOS drivers. The Promise Ultra100 is a nice and simple to use card, but they have issues with soft-resets on these older systems and don't have a driver for DOS so you can read the CD-ROM. Nonetheless, I use the Promise Ultra100 on at least two setups and just use the reset button.

Thanks for the advice. So far so good, but if corruption continues I'll start hunting down a Promise IDE controller. I believe I do have a spare PCI ATA100 controller lying about, but it is not a Promise one but rather an ITE IT8212F RAID Controller I flashed with the ATAPI Bios. I had wonted to use it with my Deskpro EN but that one failed to boot from the card, I may give it a try in this system nevertheless.

appiah4 wrote on 2021-04-19, 21:02:

feipoa wrote on 2021-04-19, 20:58:

If you continue to have issues with cross-linked files, you may want to go with a PCI Adaptec SCSI or Promise IDE card.

Thanks for the advice. So far so good, but if corruption continues I'll start hunting down a Promise IDE controller.

A Promise IDE controller is a good start. But you still need to make sure the PCI bus acts fast enough, otherwise you might get strange issues. I experienced problems with PCI @ 20MHz at FSB40, and with some PCI performance settings (IIRC it was about write posting or bursting) disabled. For UMC8881 boards with PCI masters, the general advice: "Disable anything that might make trouble" might be the wrong approach.

I can confirm the following revision seems to work without issue with simultaneous keyboard and ps/2 mouse input.

So I think I also just confirmed you can kill one of these boards by plugging 28 pin sram chips into the 32 pin sockets incorrectly. I was swapping out the 512k chips for the original 256k chips in my original board. I saw a jp1 label and thought I was lining them up correctly with pin 1, doh! So Is this a thing? Do I need to pour one out for this board? I tried reseating the modules correctly and then swapped back to the 512k chips. It’s powering on but no video, keyboard lights stay solid.

*update* Apparently just killed the bios eprom, swapped in another chip and we’re good! Although these sram chips may be sus now.

I had to downclock to 100MHz. PSA: You may think your IBM/Cx 5x86 is stable at 120MHz for running Quake timedemos but long term use showed me they get really finnicky..

Chadti99 wrote on 2021-07-11, 18:49:

*update* Apparently just killed the bios eprom, swapped in another chip and we’re good! Although these sram chips may be sus now.

It's unlikely you kill something by plugging the SRAMs in the wrong position. You mostly kill chips if you push high currents through it. Pushing high currents needs a sender of the current and a receiver of the current at different voltage. The SRAM sockets have three pins that are capable to drive current strong enough to kill chips: Pin 14/16 (lower corner) is connected to GND, Pin 28/30 is connected to +5V and Pin --/32 is connected to +5V, too. (Pin numebrs given for 28-pin / 32-pin chips). All other pins are "just" data pins and are unable to deliver killing current if you keep them inside 0V..5V. Your misplugged SRAMs didn't connect GND, just +5V. There was no way to close a circuit that damages these chips, so they are extremely likely just fine.

On the other hand, if you turn a chip by 180°, you connect +5V to the GND pin and GND to the +5V pin. The chip doesn't expect a strong mis-polarized supply, and usually dies trying to short that reverse voltage out. It shorts negative voltages kind-of on purpose, because the most common cause of negative supply voltage is an electro-static discharge. Shorting the voltage from that discharge prevents damage to other parts of the chip, and typical discharge energies are low enough that they don't kill the chip when they are shorted out. A PC supply can deliver much more energy than a ESD (because the supply delivers current continously, not just a one short spike), and the chip dies trying to short the supply.

mkarcher wrote on 2021-07-12, 21:35:

Chadti99 wrote on 2021-07-11, 18:49:

*update* Apparently just killed the bios eprom, swapped in another chip and we’re good! Although these sram chips may be sus now.

It's unlikely you kill something by plugging the SRAMs in the wrong position. You mostly kill chips if you push high currents through it. Pushing high currents needs a sender of the current and a receiver of the current at different voltage. The SRAM sockets have three pins that are capable to drive current strong enough to kill chips: Pin 14/16 (lower corner) is connected to GND, Pin 28/30 is connected to +5V and Pin --/32 is connected to +5V, too. (Pin numebrs given for 28-pin / 32-pin chips). All other pins are "just" data pins and are unable to deliver killing current if you keep them inside 0V..5V. Your misplugged SRAMs didn't connect GND, just +5V. There was no way to close a circuit that damages these chips, so they are extremely likely just fine.

On the other hand, if you turn a chip by 180°, you connect +5V to the GND pin and GND to the +5V pin. The chip doesn't expect a strong mis-polarized supply, and usually dies trying to short that reverse voltage out. It shorts negative voltages kind-of on purpose, because the most common cause of negative supply voltage is an electro-static discharge. Shorting the voltage from that discharge prevents damage to other parts of the chip, and typical discharge energies are low enough that they don't kill the chip when they are shorted out. A PC supply can deliver much more energy than a ESD (because the supply delivers current continously, not just a one short spike), and the chip dies trying to short the supply.

Thank you for the thorough explanation, I managed to kill eeprom somehow but everything else seems fine.

Has anyone had any luck posting at 200Mhz (50x4) with an AMD 5x86 with this board? I’ve tried a couple ADZ chips and a beefy copper cooler but no luck so far. I was able to post at 160(40x4) and 150(50x3). I suppose I should try 180(60x3) next. I know it is highly unlikely to run stable at 200 without a peltier but just curious how difficult it is to post at that speed?

Success! I had an ADW version of the chip that posts, good times!

Simultaneous mouse and keyboard input works for version 3.1 using the FXT/EYA chips shown.

Best Quake score I could pull with this board running an AMD 5x86 at 180MHz(60x3).

Didn’t see it already covered, apologies if so, but the Voodoo Rush works on this mobo. Def slower than the Banshee and Voodoo 2 in GLQuake and 2d sharpness isn’t good. This is the Intergraph Intense 3D. GLQuake results with AMD5x86@180 w/sound.