Power loss in-game

[rohan.goddard]

Hi all,
Bought and built a new system on thursday, it was running fine (with one exception) from friday until last night.

The exception is that the second monitor doesn't wake properly on startup. But cycling an option in the radeon settings app gets it to wake. I never encountered this with the monitor on my old system, the only thing different for the monitor is the GPU out is now full DP rather than miniDP so I had to get 2 new leads.

The problem that began last night is the system loses power/resets in playerunkown's battlegrounds (The only other game installed is binding of isaac, which does not cause this). Event viewer reports it as kernel power loss but I don't see anything else relevant. It crashes about as soon as the lobby loads (which is a 3d environment) so it's not an overheating issue. But I can't remember if the first instance was after playing for a while or i restarted the game... it was late haha.

System
I7 7700
Gigabyte b250-d3h
16gb gskill fortis (when in-store selecting parts they didn't have my first pick and I didn't realise these were branded as "amd" - for whatever it's worth.)
MSI rx580 8gb
WD Blue 500gb M.2 SSD
WD Green 1tb HDD (the only old part, simply for bulk storage)
Coolermaster V650
Win10 home pro

Drivers report as being the latest.
Will run a memtest, any other ideas? It ran fine for a few days.
edit: Possibly a bios issue? gigabyte site has an F7 version mentioning something about a HT (I assume hyperthreading) flaw...

 

[helpstar]

update the BIOS of the motherboard in BIOS itself using q-flash (read manual how to)

eventually update the firmware of the SSD

run temperature tests
intel processor diagnostics utility
furmark for about 10 min

 

[rohan.goddard]

SSD was up to date
intel diagnostics passed
bios is now updated

furmark ran for 3-4 minutes before a reset in the same manner. temps were getting up to ~70c? (consistent with this: http://www.tomshardware.com/reviews/amd-radeon-rx-580-review,5020-7.html)

Yet to run the memtest, seems a bit more involved. Will give it a shot now I suppose.

 

[helpstar]

Are you sure with the model type of your power supply?
Couldn´t find coolermaster with VX, just V or VS,...
Or is it a Aerocool brand?

check all drives as well with wdlife or similar from WD

 

[rohan.goddard]

Yep, the wd software I had tried and found no problems.

My mistake, it is just a v650.

Memtest is running right now.

Also, the second screen issue is not present during the test or in bios/post screen, but the monitor does not wake properly in Windows.

 

[helpstar]

Did you freshly install windows with the new components?

run 3dmark timespy and click "compare result online" after it. Copy the address which comes up in your browser and post it.

 

[rohan.goddard]

Yep, it's a clean install to brand new components. Memtest has just finished at 100%. I'll pop it back on the desk and double check power connections after trying your last suggestion.

 

[rohan.goddard]

Ah, timespy requires the base 3dmark program, which is 30US. I've not really got any interest in benchmarking otherwise, or i would consider it.

I did run furmark again, with the same result, it crashed at about 3.5 minutes with normal temps.

edit: nothing obviously amiss with power leads.

 

[helpstar]

3dmark is for free, if you chose the demo version on steam or get it from a not so difficult source like:
https://www.techspot.com/downloads/5775-3dmark.html

set ram frequency to 2133MHz

 

[rohan.goddard]

Thanks - i didn't notice the demo option. Downloaded via steam and ran twice. First time it crashed shy of 1 minute, on the second it seemed to be 30-40 seconds in.

If memtest86 returned 100% pass on the ram, would changing the speed really help?

 

[helpstar]

How hot is the CPU while crashing?
have a look with amd overdrive or coretemp

reset BIOS (clear CMOS by jumper)

changing RAM speed can help to get the system more stable, it´s just a little workaround

 

[rohan.goddard]

Temperatures are all normal when it crashes.

I tried going back one GPU driver iteration, that solved the relatively minor screen startup/waking issue.

Tomorrow I'll be talking to the vendor or possibly trying to RMA - but since I believe it's a hardware problem and I'm not sure which component I think having the vendor test what they've sold is the next step.

 

[rohan.goddard]

Turns out this was the PSU. I was really leaning toward a faulty GPU after i swapped in my old 7870 and it benched/ran games fine (albeit slower of course).

Took it back to the store, my GPU ran fine on their test bench. Brought in my system and it still crashed. Refunded the PSU and got another, running fine.

That signal issue remained though, so I'm running 17.7.1 AMD drivers which don't seem to suffer from this.