memtest86+ v5.01 hangs during SMP-enabled memory test only

[mikeymikec]

This is with an older machine, Core 2 Quad Q8300 with 6GB RAM (four slots, 2x2GB, 2x1GB, all DDR2-800).

I've tested each module in turn with SMP enabled, no problems detected.
I've tested each pair of modules with SMP enabled, no problems detected.
I've tested each pair of modules in different slots to usual with SMP enabled, no problems detected.
I've tested all four modules without SMP enabled, no problems detected.

I've run Prime95 for half an hour on the machine in question with the 2x2GB modules without any problems, I'm running it right now with all four modules.

The computer's symptoms are intermittent, things like if it's left to its own devices after booting Win7, it sometimes hangs. There's been a few BSODs that IMO blatantly point towards faulty memory, but the dates of the BSODs are really sporadic:

30/03/15: MEMORY_MANAGEMENT
22/05/14: 0x00000124 (WHEA_UNCORRECTABLE_ERROR according to google)
29/03/14: IRQL_NOT_LESS_OR_EQUAL
21/03/14: SYSTEM_SERVICE_EXCEPTION
18/03/14: PFN_LIST_CORRUPT
15/02/14: 0x00000124
15/01/14: IRQL_NOT_LESS_OR_EQUAL
14/01/14: UNEXPECTED_KERNEL_MODE_TRAP
13/01/14: MEMORY_MANAGEMENT
10/01/14: SYSTEM_SERVICE_EXCEPTION
06/04/13: 0x00000124
10/12/11: 0x00000124

One interesting thing about those dates is that none of them are in warmer months.

BlueScreenView always lists Windows system files in the 'caused by driver' tab (ntoskrnl, hal.dll, win32k.sys).

Re: hanging - the machine apparently has hung 5 times in March (2015), twice in Feb, and three times in Jan.

One theory I tested was that perhaps when memtest is running in SMP mode with all four modules that it's pushing the system harder (and using more power) and that's why it always hangs in the first five minutes (it might even be exactly 05:37 into the test as well), but a spare PSU didn't make any difference.

My only remaining theory currently is that the board is developing a fault that revolves around all four memory module slots being occupied. It's a problem I've seen before in other older PCs but I felt like trying to narrow it down further if anyone has any ideas. The big question is whether the PC will be stable with 2x2GB in future of course.

 

[Soulkeeper]

last time I checked memtest in smp mode was experimental and had bugs, so I wouldn't rely on that alone.

You might try relaxing the memory latencies and adjusting the voltage slightly

 

[hennessy1]

I have had constant issues with the smp mode on v5.01 from x58 on up and many different types of memory. I will even get errors in smp mode on the first pass but do 20 or more passes without it on and no errors.