Troubleshooting power down issue

Hi,

i got a used 1070 (upgrade from 970) three days ago and since yesterday my PC just randomly shuts off after playing games for a while.
By shut off I mean equivalent to just pulling the power plug, the PC is completely powered down instantly and does not start again unless I unplug and then replug the power cord.

I’m just confused since this happened zero times for one day, then once the next day and then 3 times within 30 minutes the day after.
Therefore it should not be a thermal issue, or power draw issue, since I was playing the exact same games and really nothing changed.
One shutdown happened at about 30% GPU usage with CPU and GPU temps in the 60C range.

I already had this type of shutdowns about a year ago but they stopped when i removed the manual OC from my 3600 (although my PC previously worked fine with the same OC for a couple of months).

My main question is, how do I troubleshoot what exactly is wrong since the shutdown seems to hardware related.
Also, I’m dual booting Win10 and Arch, maybe relevant logs can be obtained somewhere, I just wouldn’t know where for this type of issue.

Build

  • MSI B450 Tomahawk Max
  • Ryzen 3600 cooled by Noctua NH-D15
  • 2x 8GB Corsair Vengeance LPX 3000Mhz CL15
  • Seasonic SSR-450RM (or G450) 450W PSU

I would start here. Test the PSU, 450W may be low. Do you have a way to see what the power draw is at the wall.
Also, how new is your power supply?

Since you got a used card, it could be that the card is bad. I have had this issue before with a Radeon AIW that failed slowly from the over clock. I have also had this when a card fell out of the AGP slot and grownded on the mobo, the card would work for a little bit but progressively get bad and then hard shut off the machine. I had to do what you did and eventually the card gave up the magic smoke.

However, since you said that you had this issue before due to overclock, start with the power supply.

2 Likes

Agree with the power requirement issue. Im suprised your 970 didnt have any prior issues because you seem to be tripping some overcurrent protection and it looks like your 970 just barely lives under the limit before. If you still have access to your 970, i suggest you replace your 1070 temporarily and attempt to replicate the issue. If it stops shutting off, you should get a larger wattage PSU, probably something like 650W or higher so you have some allowance. Better yet, search for a PSU calculator online. Get about at least 20% extra allowance on top of what the calculator recommends.

If your PSU still fails with a 970, it looks like your PSU is beginning to fail and you need a new one.

You may also check if the GPU has a modded BIOS (im not exactly sure how to do that). A modded GPU BIOS could alter how the GPU draws power. Ask the previous owner if he/she/they could recall any mods on the system.

Also, I dont think your OS has anything to do with this problem. But there are smarter computer people here that can give a better recommendation.

1 Like

If it’s a power issue I’d say you can try to enable ECO mode from the BIOS to see if this helps the situation. If it does, since the power requirements for the CPU get lowered, you might have too weak of a PSU and should try to switch it out for a more powerful one.

Undervolt the 3060 and 3600.

Thanks for the feedback.

I’ll try undervolting for now and see if that helps.

Although as far as i know the PSU should be plenty for sustained loads.
From reviews the avg 1070 power draw should be around 160W, maybe 200W with insane OC.
The 3600 should draw about 90W max.

Maybe it’s some peak current or something wrong with the PSU, maybe both.

Update

So i tried a couple of things and the ran furmark to check for crashes.

  1. Undervolting 1070: Crashes after like 3 mins.
  2. Undervolt + 50% Power target: still crashes after 3 min.
  3. Reinstalled old 970: Crashed instantly when starting furmark
  4. 1070 in lower PCIe slot: crashes instantly

So after this testing, loading any GPU seems to crash instantly.

Since I don’t know how to narrow down the issue any further, I guess I’ll just buy a new PSU and hope that it fixes the issue.

1 Like

I’d like to clarify a thing - when you say crash do you mean:

the computer completely powers off and you need to turn it back on with the power on button

or

the computer freezes and the monitor shows the last frozen image indefinitely until you manually power it down?

Basically what i wrote

By shut off I mean equivalent to just pulling the power plug, the PC is completely powered down instantly and does not start again unless I unplug and then replug the power cord.

I see a black screen, fans in the case don’t spin anymore, most LEDs are off, pressing the power button does nothing.
The crash is instant, no slow transition, just like cutting the power.
I think when I had the 970 installed, a white LED on the card was still glowing.
But I don’t know if it was actually getting power from the PSU or just from a slowly discharging capacitor.

Now that I think about it, the crashes i had last year were likely not related to power.
I said I OC’ed the 3600, I actually undervolted it heavily and pushed the frequency on all cores, like 1.2V with 4.2 GHz all core iirc.
For perspective, without manual OC my 3600 barely boosts to 4100 MHz all core at 1.45V with the ryzen precision boost overdrive thing enabled.

The weird thing was, that it worked completely fine for about 6 months, then I started having random crashes while playing games.
Removing the manual voltage and clocks and simply turning on the ryzen PBO fixed the crashes so I assumed the OC was suddenly unstable for some reason.

I’ve been using a Corsair TX 750 24/7 for the past 13 years.
It is possible however unlikely that power supplies go bad with age.

I would recommend finding a PSU with dual EPS CPU connectors as the Threadripper and some AM4 boards feature it.

This topic was automatically closed 273 days after the last reply. New replies are no longer allowed.