It could be that there is a piece of failing hardware. You're on the right track on the different things you're trying. Have you tried a different motherboard?
It could also be that some component(s) are marginal -- running correctly most of the time with occasional failures at the margins. You might try backing off some of your BIOS settings to more conservative values, especially for RAM timings. Also maybe try boosting the voltages one notch for various processor and/or RAM voltages (don't over-do it -- just a little bit).
I had instability in my machine for awhile. It was very annoying. I had four sticks of RAM (actually six, but of course only four fit at a time). Four of them were 1GB sticks and two were 2GB sticks. Obviously I wanted to go 2+2+1+1 to get 6GB. I ended up rotating through the pairs to finally isolate that one or both of the 2GB sticks was not entirely stable. It worked most of the time, but I would get a crash about once a day on average.
I ended up doing an RMA replacement on the RAM, and the new RAM is working perfectly. But until then I was trying a lot of the same things you're going through.
I haven't got another mobo lying around. I pretty much only have the bits that make up the computer. I'll head to a shop today to get them to take a look at it over the course of a week.
I did back off on the timings of the RAM as one of my strategies. At least the timings I know about that is. I bought GEIL PC3 12800 CL 8-8-8-28 RAM. That's how it's marked at least. This actually reverts to the first SPD value when the BIOS is set to defaults, i.e. PC3 10600 CL 7-7-7-24. I've tested at both these bandwidths/latencies. Both throw up errors.
I've also cranked it down PC3 10600 CL 9-9-9-x. This seemed to give me more errors but I can't say for sure.
I didn't want to increase voltages just yet as I felt that was a last resort. It's getting close to that tho' now. I'll give it to the people who sold it to me first to see if they can try different mobos/PSUs/RAM/etc.
Do you run Windows 7? You could try using the Windows 7 memory test for faster results.
I've found that sometimes memtest86+ has to run for hours before detecting an error that the win7 memory tester catches on the first pass.
Yeh I 'run' Win7 but at this point I can't boot into it; we've become that unstable. As for Memtest86+ taking a while, as I type I've clocked up, on the current session, 1479 errors thus far. I'm thinking about shooting for the world record.
if you have a digital multimeter you can test the voltages of your psu from a spare molex or sata connector. do you have a UPS or surge protector? maybe something got fried during a thunderstorm...i am leaning towards your psu being at fault...do you have a spare to test with? if not my guess is mobo. good luck!
Hey ryko. Nope I don't have a meter. I'm guessing the shop will tho'. I don't have a UPS. I'm running the monitor and computer off a power strip that purports to be a surge protector but I wouldn't be sure. I've tried two different power boards. The one I'm using now is brand new.
In fact, I think you're on the right track. I now believe this is probably a PSU issue. I live in Bangkok. At this time of year we have the most amazing electrical storms almost daily. Sometimes, when the strikes are close it sounds like 3-5 atomic bombs going off at once. Can be pretty cool.
Thanks for all the help thus far folks. I'll be back in touch when I've heard back from the shop.