Unstable 3600 system

[porina]

Main components:

 

CPU: 3600 stock

Mobo: Asrock Fatal1ty B450 Gaming-ITX/ac, bios 3.50 (1.0.0.3ABB) - latest at time of writing

Cooler: Noctua D9L

Ram: 2x4GB Crucial Ballistix 2666

GPU: RX 580 (not under load)

OS: Win10 1903

PSU: without checking, probably a Corsair CX450 or CX450M, or better.

 

I was trying to do PrimeGrid work on it but found it rebooting randomly. The work is comparable to Prime95 FFT sizes around 800-900k. I was running one task per CCX (2 tasks, 3 threads each). Temps were around 80's C at stock. I tried turning on ECO mode which seemed to help, but it only reduced the chance of it happening and didn't eliminate it. Temps did drop, as did the clock and voltage.

 

I suppose at some point I should double-check the ram isn't going bad, but if it was, I'd expect to see detected computation errors than a hard reboot. 

 

My similar 3700X system is running the same work fine, as are a bunch of Intel systems. On that note, I did notice there were updated chipset drivers on the 3700X system recently, so I should install them on the 3600 system also. 

 

Any other ideas?

[Caennanu]

Not entirely sure about the mainboard. But is powerdelivery good enough and cooled well enough?

Since its an ITX board, i assume its also in a smaller casing than the 3700x you mentioned.

[porina]

27 minutes ago, Caennanu said:

Not entirely sure about the mainboard. But is powerdelivery good enough and cooled well enough?

Since its an ITX board, i assume its also in a smaller casing than the 3700x you mentioned.

NZXT Manta, so it is basically as big as micro-ATX cases. I have two 140mm fans intake, one exhaust. Can't really do more than that.

[Caennanu]

hmm then thats probably not an issue.

i know with the 3000 series the infinity fabric is lose from the memory clock.

i do not know if it automatically sets it to 2666mhz if you install that memory, so maybe its set higher and causing a clock mismatch? Worth checking i geuss?

After that, the only thing i can think off is indeed defective memory, or a bad sector on the harddrives.

[porina]

4 minutes ago, Caennanu said:

i do not know if it automatically sets it to 2666mhz if you install that memory, so maybe its set higher and causing a clock mismatch? Worth checking i geuss?

I'll doublecheck but it seems highly unlikely.

 

4 minutes ago, Caennanu said:

After that, the only thing i can think off is indeed defective memory, or a bad sector on the harddrives.

On the note of bad drive, I did have that problem with this system in the past. I had a M.2 drive installed, which was unfortunately right under the GPU. As I was then running the GPU constantly, it cooked it. Now I've moved to a 2.5" SATA drive nowhere near the GPU so that should not be a problem any more.

 

Thinking more, I wonder if there is a protection shutdown setting? If so I wonder if that might be (false) triggering, since the normal operating current/power limits should prevent that from ever being reached.

[Caennanu]

6 minutes ago, porina said:

On the note of bad drive, I did have that problem with this system in the past. I had a M.2 drive installed, which was unfortunately right under the GPU. As I was then running the GPU constantly, it cooked it. Now I've moved to a 2.5" SATA drive nowhere near the GPU so that should not be a problem any more.

It shouldn't, but not everything ordered is flawless, i think it would be worth it to do a disk check. Atleast you'll know its good or not

 

7 minutes ago, porina said:

Thinking more, I wonder if there is a protection shutdown setting? If so I wonder if that might be (false) triggering, since the normal operating current/power limits should prevent that from ever being reached.

I like the thinking, its in line with the cooling i suggested earlier.

No idea how to check this tho. I remember from old bios (as i haven't checked this in recent pc's) you could set thresholds yourself, somewhere in the hardware monitor.

Maybe that one has odd settings and is causing a unjustified thermal shutdown. Should also create a logbook entry tho?

[porina]

40 minutes ago, Caennanu said:

No idea how to check this tho. I remember from old bios (as i haven't checked this in recent pc's) you could set thresholds yourself, somewhere in the hardware monitor.

Maybe that one has odd settings and is causing a unjustified thermal shutdown. Should also create a logbook entry tho?

On Intel systems for example, there is an over-current setting in bios. That should activate in hardware/bios before software can do anything. I don't know for sure if there is similar on AMD. Will have to look.

[Mathieu9836]

How do it reboot, all of a sudden without warning or the OS just reboot by itself?

[porina]

1 minute ago, Mathieu9836 said:

How do it reboot, all of a sudden without warning or the OS just reboot by itself?

The system is headless so I only remote-in to check it now and then. I usually find all the software that was running isn't. It is set to login without interaction. I have eliminated Windows Update reboots as a possible reason.