Sorry for ramble, need help dissecting my Memtest86.

[WonderBoy]

I'm doing Memtest86 to try to isolate some POST issues during restarts (and not cold starts).

Doing some browsing here, I have noticed some comments about how Memtest86 shouldn't really stress the CPU and in past experiences I vaguely recall fans being very quiet during the whole stress test. I seem to be having a different experience with it today, and am hoping someone can shed some light on that.

My setup is:

ASUS ROG STRIX X670E-F Gaming Wifi
AMD Ryzen 9 7950X3D
EK Nucleus AIO CR360 Lux D-RGB
G.SKILL Trident Z5 Neo RGB Series DDR5 64GB (2x32GB) 6000MT/s CL30-40-40-96 1.40V
MSI RTX 4090 SUPRIM Liquid X 24G
Samsung 990 PRO 4TB

1) My CPU is sometimes getting pretty hot (81+) during Memtest86 even though OCCT and Prime95 never even got close to that (55-65c depending on core). Is that normal or weird?

2) I notice my AIO fans going crazy in short bursts even when the CPU temp is showing a very low 39c. It sounds like the fans are suddenly jumping to 100% and it happens sporadically, though it's especially noticeable / consistent for Test 13 Hammer. Is this normal behavior? In my other stress tests, I noticed my RAM could reach temps of 79c even when the GPU and CPU never get past 65c. I'd never noticed RAM temps before all this but I'm surprised that they could jump past my GPU / CPU by 15 degrees during stress testing...Is that normal or weird?

3) Memtest shows RAM temp of N/A, so I couldn't confirm if maybe the RAM was triggering the fans to turn on. Does that even sound plausible? Can AIO fans or chassis fans react to hot RAM even when the CPU is cool? All my fan curves and settings are default. I do have EXPO enabled. After 4 passes, the cumulative error count was 0.

I'm interested in any insight on any of these things, even if not directly relevant. BTW, I had everything set to default values. All tests, multi-core, both sticks of RAM.

[David (PassMark)]

If this is water cooled I would think there could be a disconnect between the temperature of the water in the loop and the CPU. i.e. it could take a while for one to effect the other.
The fans might be driven not by the CPU temp, but by the water temp?

It is unlikely that any of the fans are being controlled via the RAM temperature. It isn't trivial to get the RAM temp, and some sticks don't even have a temperature sensor.

It would be surprising if MemTest86 made the CPU hotter than OCCT or Prime95. But maybe the fan control for your cooler isn't controlled by BIOS and needs Windows to be booted to work as expected?

[WonderBoy]

Thanks for the insights. Still on my search, but wanted to update: on further testing with OCCT it does feel like Memtest86 was able to push my CPU to higher reported temps than I've otherwise seen (81c in memtest but the more strenuous tests I've done on OCCT can maybe reach the low 70s). Not sure if that means anything to anyone, just thought I'd put it out there.

I can't seem to find any errors on my RAM using Memtest86 and other apps after many hours. Would you say that rules out the RAM sticks as the cause of POST failures? (It often gets hung on the DRAM Q-led, but not always.)

[David (PassMark)]

If MemTest86 passes, it makes it less likely that bad RAM is the cause. Nothing is 100% sure however.
Running with one RAM stick, then the other might be another test you can do.