colmanhymx

Try another PCI-E power cable if your PSU is modular. Faulty cables could also cause problem. Depending on your PSU, you could try to run two PCI-E 8Pin cables separately, and not to run one single 8+8pin.

That's my case as well. I'm running a 3700X so also no iGPU. Is there an old GPU you can test with? I borrowed a old GTX460 just to test if it was the GPU acting weird.

That's weird. Have you tried updating your GPU drivers or Reinstalling them? Sometimes that solves the missing temp problem.

Have you tried just running tests without your GPU? If you have iGPU, try running that. I had problems where a dying Vega 64 caused multiple screen blackouts, random shut downs that locks up the power button and needs to be hard-reset every time. Side note, if you need a secondary PSU to test with, if your local stores have a return policy where they accept open returns that's not defective, you could always go buy a new mid-range PSU and test with it. If that solves your problem, keep it. If not, you could always return it.

That's weird given that you have a relatively high-end GPU. Have you tried running MSI's Afterburner to see if the VRAM temps are displayed there?

Hope that solves your problem!

Depending on your location and your local store's return policy, it is possible to buy a new, mid-range power supply and test it out on your PC. If it solves your problem, keep it. If the issue still persists, return it. If your financial situation allows doing so, and that your local stores accept open returns that are not defects, I'd say this could be a way?

Do you have another PSU you can test with?

I have a very similar problem where furmark crashed upon starting a stress test, but the difference is that I'm running a Vega 64. After some inspection, the only anomaly I could find was random and impossible spiking of VRAM temps. Have you checked your VRAM temps as well when you were checking your GPU Core Temps? Btw, I capped my games to a max of 60FPS and they all seem to run fine. I tried running 144hz on a Samsung Q-LED panel and it would either crash totally, shutdown and require a hard reset; or it the screen would go black for minutes then come back running fine. Since then, I have been using a old Samsung 1080p 60Hz panel from before I upgraded and it ran much more stable, where game crashes would only occur every week or two. If your card continues to crash like this, or even under light or normal loads, I'd say there might be something wrong with your GPU. Mine is crashing just using Chrome and watching Netflix, but then my Vega is also dying and quite possible on it's last leg.

While I was typing that, my screen went out but the PC is still running. Starting to believe it's being caused by a dying GPU.

I think this might just be the ultimate cause. I noticed some interesting random spikes for the GPU VRAM temperature (went from 18 to 3102 which definitely is impossible), so I ran furmark and it crashed, instantly. Same shutdown, can't reboot with power button, ram sticks still light up. Think I may have found my smoking gun, or my smoking hot GPU. I think my GPU fan might be dying? The noise is pretty recognizable but I always just thought that's how loud Reference cards are.

What RAM and Motherboard do you have? Any discrete GPUs?

Yeah, guess I could run memtest. Could it be related to the XMP Profile I'm running?

Somewhere around 50. I have a Noctua NH-U12A so I think CPU cooling is fairly adequate?

I did, just serviced the card last month. I used noctua's NT-H2. Under full load it's running around 70 Degrees? That's also what I thought at first, so I have a monitor running just to check all the temps. When it crashed, the gpu was just running around 55-60.

My desktop is recently experiencing this problem where under light or medium load, it shuts down out of the blue and cannot be restarted with the case's power switch. Only way to restart was to turn off the power supply and do a hard reset. It used to be very less frequent, like once a week or two but these couple days it's been doing this every day... This is also a fairly new and recent build, about one year old so I don't understand what's going on. Is it some sort of hardware problem or is it software related? Specs: Ryzen 3700X MSI x570 ACE Trident.Z 3200Mhz 16GB Samsung 970 Evo Plus 1TB XFX Vega 64 Reference Corsair RM850I Phanteks P300 Windows 10 Pro 2004 (19041.508) Thanks for the help in advance!

Yeah, I had the same thought that my current card is ran in a very moist environment. Yeah, maybe getting a refund is a solution with less hassle, but at around 200usd for a Vega 64 is a pretty good deal, as it still comes with warranty, as for that same price, if I went to the shops and opt for a new card, I think I can only get like a 1050ti or a 1060 3G from an unpopular manufacturer, which makes no sense for me to upgrade from my current 390.

Just got the photos from the seller, pcb seems fine, screws don’t seem discoloured, anything I missed?

Thanks for the advice. Swap is happening tomorrow. Will post pics of new card once the seller send them over sometime later today. Anything you would suggest me to test after getting the new card? Ie things to look for, stress tests to run? As I plan on water cooling it, and as taking off the cooler voids the warranty, if there’s any problem with the card, I would like to rma the card beforehand. Thanks for all your help!

The seller just offered to give me another card, one with warranty, and that the warranty stickers are not broken at the same price. Should I take the offer?

Oh and another thing is that after updating the drivers and gpu bios, it not instantly restarts when I run furmark, but is fine if I game on it. I don’t think it’s an psu issue as I have a RM850i and that my old gpu, a 390 which has around same power consumption ran furmark just fine.

Sorry for very late reply, very busy weekend.

It was used to mine before I got it, mined for around less than a year I think, as the card was purchased on 10/3/2018. I suspect that it was caused by the High humidity here where I live. Tried wiping if off with alcohol and a piece of tissue, got some off but still a lot left. Will post more pics later

I cleaned off the card by wiping with 95% alcohol as that the highest purity I have on hand, I cleaned the dust buildups on the joints of the inductors as must as possible, and it seemed okay-ish, I’ll post more pics of the pcb later

I don’t think a RMA can be done as the little sticker on the screw that’s says void when broken is broken already when I got the card