Stability issue (occasional hang/BSOD)

[mikeymikec]

I built a computer in 2010 of the following spec, and upgraded it in 2014 to Win7-64:

AMD Athlon II X2 240
ASUS M4A78T-E (latest BIOS)
4GB DDR3-1333 (Kingston HyperX)
500GB Seagate 7200.12 SATA2
DVDRW
Corsair VX450W

It took the WIn10 upgrade without any problems. Recently I spotted that its HDD was failing so it was time for a new one. We took the opportunity to upgrade the following:

CPU: AMD Athlon II X4 630
SSD: Samsung 850 PRO 256GB
RAM: An additional 4GB (Crucial DDR3-1600, just basic RAM, not anything with unusual timings etc)

I installed Win10-64 1607 on it. Setup went fine, but during the first Windows session with it connected to the Internet, it BSOD'd (IRQ NOT LESS OR EQUAL). I thought as it would have been installing drivers at that time that it might be a one-off, but then when I told it to do a full chkdsk it hung during the check. Major alarm bells ringing!

Since then (over the course of this week) it has generally worked fine (though it racked up another BSOD: KMODE_EXCEPTION_NOT_HANDLED, and another chkdsk hang), and all of this made me suspect RAM, though I've given it 8 hours of memtest86+ 5.01, 1-2 hours of memtest 4.3, the Windows memory diagnostic, 1 hour of blend Prime95, half an hour of max heat prime95 (large FFTs IIRC), no problems at all during those tests.

While the chkdsk hang is an odd symptom, I've stuck with that for the sake of testing. I'd say 2/3's of the time with the full 8GB it wouldn't manage to do the check and reboot normally. When I removed the new 4GB module (yesterday morning, including 6 chkdsks) it has worked fine every time, but being an intermittent issue I'm still nervous about it as I probably won't see this computer in a good long while (family member doesn't live near me, it's dropped off for me to do the upgrade).

For the sake of testing, I've now swapped the modules (so just the newer module is in as of half an hour ago) and I'm running another full chkdsk right now.

When the chkdsk operation has completed successfully (ie. Windows reboots normally and a wininit entry is in the event log), the tests have always said the drive is fine. SMART data for the drive is also fine.

I think there's a vague chance that the second-hand CPU may be an issue but I don't know of any better ways to test it. I think it's most likely however that while the new memory module probably isn't faulty, perhaps the system doesn't like those two paired up? I've paired up countless non-identical RAM parts in the past and I can't remember the last time I experienced a problem due to mismatched RAM so I'm not sure what the symptoms could be for that.

 

[VirtualLarry]

Man, that's such old kit. It could, in theory, just be the motherboard acting up in general.

Other than that, you are running a 4GB stick, and an 8GB stick together? I know that you can do that with Intel systems since the 865 chipset ("Intel Flex Memory Access"), but can you do similar with AMD? (I guess, since the two memory controllers are seperate?) Did you set "GANGED", or "UNGANGED"? Since the RAM amounts on each memory controller are different, I would think that you would want to set that to "UNGANGED", which is two independent 64-bit channels.

 

[mikeymikec]

2x4GB in total (old and new). I've never ganged memory myself, from my understanding of its function it makes no sense for the average user to use that feature. I can double-check that when both are in that it's set to unganged but I'm about 90% certain it was (and wouldn't it make a issue with a pair of modules working together less likely?).

 

[RLGL]

Mixed memory can do strange things, personally I would have scrapped the system and started with up to date parts.

 

[VirtualLarry]

personally I would have scrapped the system and started with up to date parts.

If it was unstable to start with, then that would be a no-brainer. It's a little harder to justify, if the rig was "working", and then upgraded with some (by now, very inexpensive) parts.