Memtest freezing, errors running test on other CPUs

[Yell0w]

Hello, I hope someone can shed some light on my issue.
Specs:
MB: Asus Sabertooth 990fx R2
CPU: AMD FX 8350
Liquid cooled (Peak temp during test 50C)
8 GB ram (2x4)
850w power Supply
Radeon R7 260x

everything is new except the power supply (about 4 years old), and the hard drive.

I've been getting random shutdowns without warning, it just turns off. After running some other diagnostic programs and passing everything I decided to uses Memtest86 V5.1.0. I've tried this with both sticks in, and each stick individually. Running memtest single CPU on CPU0 passes all 10 tests, changing to any other CPU freezes memtest, sometimes on test 2 sometimes on test 3, total freeze, have to reboot. If I run all 8 cores on parallel I get a ton of errors and it freezes on test 2 CPU7, or at least CPU7 is what I can see on the screen not sure if other CPUs had errors, have to reboot. Running round robin freezes on test 2 CPU0, have to reboot. Running sequential freezes on test 3 CPU0, have to reboot. Any idea? I'm at a loss, not sure whats wrong with it.

[David (PassMark)]

If you are water cooled then I also expect you have overclocked the machine?
Try returning it to default clock speeds.


Otherwise as an experiment, try bumping up the voltage level for the RAM by 0.05V.
If you are using the XMP settings, try not using them (effectively under clocking the RAM).

[Yell0w]

Thanks for responding.
Haven't overclocked it yet, it started doing this right after I built it. I did a sysprep to move my hard drive to the new build. I'll change the RAM voltage and see if that helps. Could it be a CPU issue? Since when ran in parallel it show errors on some of the cores?

[Yell0w]

ok reduced the RAM voltage from 1.49 to 1.45.. ran test in parallel got to test 4 and stopped, the timer was still running but it wasn't doing anything. tried to run single core on cpu1 and it froze at the beginning of the test. Tried to run sequential and round robin and the same thing happened. Here's a pic of the parallel test..

[David (PassMark)]

I was suggesting to increase the voltage, not reduce it.

Yes, it could be CPU or motherboard, but without some very high end test & measurement equipment, it isn't possible to know. Instead you are stuck with trial an error.

[Yell0w]

Sorry misread your post, I increased it to 1.55.. Ran the tests again and got the same results. I also replaced my PSU just in case it was going bad, no change. I can still return my mobo and CPU and ram if needed for an exchange.

[David (PassMark)]

If you had some spare parts I would suggest swapping some parts in and out. RAM, then CPU.

[orioon]

Is the machine really shutting down or is it restarting?
Only seen this behavior when running in a temperature or voltage limitation.

Which tests beside Memtest86 did you run?
Do you encounter the issues during normal use? (e.g. Web browsing)

You could try open the case and let a big fan blow inside it.
Especially on watercooling without much air flow hot spots can occur on voltage regulators and other components.