Abrupt end on Memtest86.log, PC reboots, but zero errors logged?

[leepox]

Left my pc to run memtest86, when I got back an hour later the Windows login was on the screen so PC must have rebooted.

I looked at the log file and it just abruptly ends on these last few lines:

2020-07-26 16:25:50 - Running test #5 (Test 5 [Moving inversions, random pattern])
2020-07-26 16:25:50 - MtSupportRunAllTests - Setting random seed to 0xDA2ED686
2020-07-26 16:25:50 - MtSupportRunAllTests - Start time: 1859209 ms
2020-07-26 16:25:50 - MtSupportRunAllTests - Enabling memory cache for test
2020-07-26 16:25:50 - MtSupportRunAllTests - Enabling memory cache complete
2020-07-26 16:25:50 - Start memory range test (0x0 - 0x430000000)

This happened on the second pass, went through first pass test 1-10 without errors. Does this mean that the RAM is good and it is a question of power? Either not enough voltage to RAM or my PSU is crap?

[David (PassMark)]

Sudden reboots are hard to diagnose.
Yes, it could be the PSU, but it could equally be that you had a short black out of your mains power supply. Or the CPU or RAM could be bad. It isn't conclusive, except to say something isn't right.

Is it random? ie. If you repeat the test does the problem happen at the some point in the testing.

[leepox]

I'll do another test, perhaps the IF clock is too high on the 3900x. Will put everything back to default and start from there.

[David (PassMark)]

Yes, (obviously), if you are overclocking and it fails, then drop the clock speed.

[leepox]

So I have now done two separate runs, one with the memory at default settings (3600Mhz@1.35V) and one where I overclocked it to 3732 (3733 MHz@1.38V). On the first one left the FCLK on my 3900x to auto, and on the second one decoupled it but ran it at 3600 FCLK. Both runs finished the 4 loops with zero errors.

So now, I'm gonna do a third run with the IF at 1867 MHz and putting the memory back to default (3600Mhz).

I've got a hunch that since I have seen zero memory errors so far that the memory is perfectly capable at those speeds and it's the CPU IF that's struggling for stability. Perhaps I need to bump up the SoC, VDDP, VDDG voltage a little bit more? Currently 1.1, 0.92, 0.98 respectively

[leepox]

Just to update. I have managed to get rid of the random restarts by lower SOC, VDDP and VDDG voltages at higher IF (1867). Undervolting these parameters was the key but I still can't wrap my head around it as to why.

Now, the next mission is to try and break the RAM and get tighter timings now that the CPU is stable? XD