I would:
1. Try with a known good high quality PSU
2. Clean the edge contacts on the CPU
3. Check for bulging capacitors on the motherboard.

+5V isn't as big of an issue with Slot A as it is with later more power hungry Athlons. Not irrelevant but not the prime suspect.
If this is a hardware issue I'd much rather suspect tired capacitors than an originally underdimensioned PSU.

Is there a later BIOS revision avaliable? Did you try it?
The Irongate chipset has lots of steppings and some quirks. What stepping is this?

Ok thanks very much for the replies , I'm gonna strip it completely , have a really good look at the motherboard and clean everything thoroughly and swap out the PSU for a really good Enermax 535w one I've got which has been recapped, I'll get pics whilst I'm there of the chipset etc , capacitors look great to be honest from what I can see but I know they could be dried up.

As for the bios , no I haven't tried that , tbh I was a bit nervous of doing that with a system that could lock up whilst it's flashing, I've found bios files on MSI's website , one of them has an insteresting update regarding AC power loss ??

https://www.msi.com/Motherboard/support/MS6191#down-bios

No flash utility on there I can see though ?
This is gonna be an OEM bios on the board originally though right , kinda gimped with it being PB ?

Yeah, I wouldn't flash the BIOS until you get it running stable unless the BIOS chip is removable and you have an EEPROM programmer.

If PB bothered to modify the stock MSI BIOS then the MSI BIOS update may not go on there without force flashing it.

Well I've stripped everything and uber cleaned everything including all contacts and slots , felt like I've had the whole thing swimming in Isopropyl. 😂
Put it all back together with a known good PSU and so far no crashes lock ups or problems, installed windows on a fresh drive and it's sat running Quake demos with no glitching or stuttering so far.
Fingers crossed , pretty sure it was th...............................................................

Aww fs , it's literally just locked up whilst I was typing this , Quake demo went all quiet after over 1/2 hour of running and a solid lock up , reset it and no display again. Argh !! 🤬🤬🤬🤬

The bios is removable but I don't have an EEPROM programmer. 🙄

Here's some pics of it , the bios revision v1.3 , it says A6191p2 V1.3 on boot screen.

Some more

Good news: This is the later C5 stepping which supports super bypass.

Can you boot into DOS from floppy?
If so you can flash the EEPROM on the board using UniFlash.

Yeah I read about the revisions when you mentioned it , thanks for that.

Yes I can boot into Dos from a floppy but it has also locked up from there when I tried to do a windows re-install(the first time), although it's not done that since I rebuilt it , only in windows.

Gonna have to contact my friend and make sure he's ok with the risk of a bios flash first.

I really want to get it sorted because he doesn't really want a lot for it considering what's in it and how mint the case is. ( Athlon + 3dfx etc)

Thanks for all the help , much appreciated, I think I'll go with the bios flash if he's ok with risk , I've found another post on Vogons where someone has successfully flashed the ms-6191 to V1.6.

How do I flash my BIOS

Good luck!
If you have a NIC with the same EEPROM socket you may be able to recover a bad flash using that.

Maybe something is overheating? If I understand correctly it was working ok until you started running demos, then it crashed, then would start up ok after a few seconds turned off, but would crash quite quickly after that. It was then ok for a while after you'd stripped it all down, but then crashed again. Which sounds like it might be due to a component having a chance to fully cool down. I have an Abit KA7 that had stability problems with what turned out to be overheating SDRAM, probably due to a fold in a ribbon cable creating an area with very little air movement right near the top of the RAM slots. When I pointed a fan at the offending stick then the problems went away.

I did have it lock up a few times whilst in the BIOS settings, but that was when the memory timings were too fast. Have you got some safe values selected for the RAM? Might be worthwhile running memtest86 (give it a few hours) to see if it throws up any errors before it locks up.

Funny enough my FIC SD11 Slot A is similar.
1st boot it'll lockup before post. Hit reset and it'll behave just fine, well by Irongate standards.
If I'm using it it'll behave but if it idles too long it'll lock up even though power saving is disabled.

it was physically found in a dumpster and I'm still using the original PSU so I'm inclined to think caps or similar but the PC doesn't see much use so I'm prepared to live with it.

Thanks very much for all the replies , I've come to the conclusion it needs a board or recapping , with it having a Voodoo in it I always run them with a fan blowing on it , so I just plonked a 120mm led fan in front of it whilst testing and when it did the lockup and no post on reboot the fan didn't light up and on some occasions it didn't even spin even though the CPU fan did, every time it did boot properly the fan spun and lit up , it has to be some form of power delivery problem , it also locked in windows and the fan stopped then also.

I've tried several different Sdram sticks in and I've also flashed it to the latest MSI Bios which worked just fine but the same problem remains.

I agree with the something overheating idea and it does seem to lean that way but it's definitely not ram.

Again , thanks for all the input , really appreciated. 👍

Lee

ram test and another gpu in place test.
But gpu been thermal pasted?
Thermal build up by lengh of operation time is my suspicion.
I just resurected a p4 board the gpu showed garbage and got blue screens after re pasteing the heatsink and adding rare known brand new ddr400 plus software updates and drivers i was in the green light.

zapbuzz wrote on 2021-10-26, 15:44:

ram test and another gpu in place test. But gpu been thermal pasted? Thermal build up by lengh of operation time is my suspicion […]
Show full quote

ram test and another gpu in place test.
But gpu been thermal pasted?
Thermal build up by lengh of operation time is my suspicion.
I just resurected a p4 board the gpu showed garbage and got blue screens after re pasteing the heatsink and adding rare known brand new ddr400 plus software updates and drivers i was in the green light.

I've tested several different sticks of known good ram and also different graphics cards both AGP and PCI.
but thanks for the suggestions 👍

loose/worn old memory socket pins a few gentle nudges? my 21 year old needs that sometimes

What was the 120mm fan plugged in to? A motherboard header, or direct on a PSU molex? Given the power distribution fail/fan stopping, I'd have a quick look at what I think are two PTC/thermal resetting fuses near the ATX connector (green components marked 1X1, F3 and F4). If you can check if one gets hot, or measure the resistance across them soon after the board locks up that might give a clue. If one is tripping it wouldn't tell you why, but might help track the cause.

As well as that, I thought of something else. This would be a last resort type measure, but if nothing else seems to help, it might worth cracking open the CPU case. Out of curiousity I opened mine up (early Athlon 500) and found that the yellow thermal epoxy/glue stuff they used between the heat spreader and main heatsink only had patchy contact. Also the grey thermal goop they used between the cache chips and the metal heat spreader wasn't actually making contact for one of the chips. In my case it didn't seem to actually be causing any problems, but I fixed it anyway.

So if it is a heat problem, it could be with the CPU/cache. But I didn't enjoy getting in to the CPU case, it felt like I was one slip away from breaking the CPU PCB. So I wouldn't recommend doing that unless you run out of other things to check. And even then it's probably worth trying to find another SlotA CPU to try first.