Mine is killing processors in socket 2.
I know it is missing one SMD capacitor C619 and i fixed one other.

I'm not sure I understand if the instability has been fixed or not - it sounds like maybe your BIOS settings fixed it?
If not, I suggest you focus on getting it to pass Prime or memtest86 or other such stress tests reliably. You can do CPU stress tests like mprime using some diagnostic boot CDs (this way the integrity of the OS isn't a factor). If it has trouble with basic stress tests then there's not much point trying to do anything else with it.
The 30 minute test is encouraging, but not completely convincing. Maybe leave a stress test running overnight, FSB overclocked by about 3% if possible, and all memory slots full. It should be able to handle that if it's stable. If not, you might be teetering on the edge and some scenario in SeaMonkey is pushing it over. Basically I suggest proving to yourself that the hardware is solid beyond doubt before getting into the complication of problems inside the Windows environment.

Spontaneous resets sound more like a power issue. I think execution errors would be more likely to freeze it but not reset it. I agree the PSU is probably good, but it's possible there's noise or fluctuations happening which aren't visible without an oscilloscope. If you happen to have other PSUs available, I'd try swapping it as an experiment. But I agree that the measured voltages look good.

If you're sure the basic stability is good, then I don't know about the other problems, but I'd probably start by simplifying what cards are plugged into it as much as possible. Also, you could try using a bootable linux CD/DVD and see if it's having similar issues there. If not, then it points at problems specific to the Windows installation - misbehaving drivers etc.

I looked up a little about that board, and at least in the "HotHardware" review, it looks like their board had a mix of some cheap caps and also some Rubycons, which were good, but of course they're pretty old now. Anyway, your caps might be different than that review showed. If you feel a bulge on any of them they're definitely bad, if not, then it's inconclusive.

Thanks. I'm starting with memtest86 as I've already got that on cd.

Overclocking to 138/34 [the lowest overclock available in the BIOS] led to memtest86 crashing after 3.5 minutes with errors. Failong address was around the 650mb mark.

Turning up the cpu fan speed in case this was heat-related, and rerunning memtest86 gave a 55 minute run before a crash with 2 million errors and counting, with the failing addresses around the 150mb mark.

Doesn't sound good, I'm afraid. It might marginally work at 133MHz but it doesn't sound solid enough to be trustworthy.

Are there any settings in that BIOS, or on the motherboard jumpers, where you're supposed to set the PCI/AGP ratios? If it's a jumper it might also be labeled as something like "100/133". I think jumpers like that were fairly common in that time period, and I think they really serve to set the bus ratios (PCI at 1/3 vs 1/4 of the FSB clock, etc).
If the ratios are set wrong it could be running the PCI bus way too fast, which would explain instability.

Does that board let you drop down to a 100MHz FSB? If so, maybe that will stabilize it. To test this, I'd still try testing it slightly overspeed, at say 103FSB or so. The point of this is again that by going a slight percentage over, it causes the PCI/AGP/etc buses to run slightly fast, so everything in the system will be pushed just slightly harder than it will be at the operational target of 100FSB.
100FSB isn't a satisfying solution but if it makes it stable, then the other problems could be revisited to see if they go away at that point.

Ok I'll have a look at the bios and jumpers later.

In the meantime I put the cpu fans at maximum and took the side of the case and reran memtest86. It's been running for two hours now.

EDIT: memtest86 ran for over 7 hours without errors. Seems to indicate some of the instability was due to temperature. The cpu alarm is set at 50 [lowest I could set in the bios, Athlons can go much higher] and never went off; alarm does work because it went off during stress testing at the weekend. Maybe the temperature of something else is the problem?

Sorry if this suggestion is to obvious, but is there perhaps a cap or two missing or bloated? Socket A boards tend to have lots of cap problems

I'll have to get a magnifying glass and a torch to see. The tops of the caps certainly all look fine; it's hard to see the bases right now but certainly nothing obvious bloated.

What PSU model do you use? Dual sA boards have strange powering scheme. At least Asus A7M266-D and Tyan boards draw power for CPUs from both +5v and +12v rails. That means you need ATX12V PSU that is capable of high load on +5v. Latter PSUs have very weak +5v output.

kokornov wrote:

What PSU model do you use? Dual sA boards have strange powering scheme. At least Asus A7M266-D and Tyan boards draw power for CPUs from both +5v and +12v rails. That means you need ATX12V PSU that is capable of high load on +5v. Latter PSUs have very weak +5v output.

This and caps. I have one of these boards, doing the EXACT same thing. While none of the caps are bulging on the top, I later found that a few are leaking from the bottoms. Just replace the caps to be sure. It's time for them to be replaced anyway.

Anyone tried any 3Dlabs card with this MB? Not a single card from 3Dlabs isn't working for me.

Oxygen GMX - not detected graphic card, only pc speaker beeps
Oxygen GVX1 and VX1 - freeze at windows boot screen
Wildcat VP560 and VP990 Pro - bios screen ok, but monitor turns off when have to show windows boot screen - only black screen.

I do also own an Asus A7M266-D

My board is boxed and brand new... and full of bloated caps. If a "new" Asus dual Socket-A board has bad caps I can only speculate that a used MSI dual Socket-A board... 😁

Skyscraper wrote:

I do also own an Asus A7M266-D

My board is boxed and brand new... and full of bloated caps. If a "new" Asus dual Socket-A board has bad caps I can only speculate that a used MSI dual Socket-A board... 😁

nah 2001-2004 was the ASUS Dark days, you could get a more reliable product from PC Chips or ECS than ASUS in those 3 years.

I had similar problems with cpu's that went bad.. but still worked 90% of the time.

Still don't know what exactly is "broken" but even had an notebook where the i5 had strange crashes, could run for hours without issues and then hung up on closing Word or Access, sometimes blue screened while nothing was going on etc. 😵

Took two weeks to find out the cpu was the issue. The notebook now runs already for 2 months without any issue anymore. 🤣