New PC I built for my sister keeps suddenly restarting

[Howitz]

Hello helpers, I recently purchased a bunch of parts on newegg to build my sister a super nice PC that she could use to stream and play all her games at max settings on 1440p. Here are the specs followed by the problem.

 

Processor: Ryzen 9 5950x with noctua nh-d15 cooler

Ram: 2x Corsair Vengeance 8gb 3200Mhz

GPU: Asus Rog Strix RTX3080 12gb

Mobo: MSI MPG X570 Gaming Plus

Storage: Samsung 970 m.2 ssd 2tb

PSU: Corsair RM750x 2018

 

The problem is random restarts, with event ID41 in event viewer which basically means nothing. She will be watching a video on youtube, playing a game or something and it just suddenly reboots.

What was done so far to no avail.

 

-Checked temps, cpu stays under 80 degree and so does gpu, chipset stays under 70, ssd under 50. so temps seems fine (that was during the 30 minutes aida64 test)

-Use an uniterruptible power supply unit (cyber power pure sine wave CP1500PFCLC) to supply better input current.

-Test ssd using samsung magician, did a complete test that took 40 minutes and no error detected.

-Tried windows 10, windows 11, nothing changes still reboots randomly.

-Tried keeping XMP profile disabled

-Tried setting game mode in ryzen master

-Tried 3 different bios versions

-Ran aida64 complete stress test for 30 minutes, temps were stable and no crash.

-on the cyberpower ups there is an lcd that showed peak wattage of 696 watts under the 100% stress test in aida64 and as i said it didn't crash which seems to indicate that the psu is right on the edge but still high wattage enough to ensure stable 100% use.

-Using high performance power mode in window.

-The ram was used in a previous computer with no issue whatsoever, same for the power supply.

 

I've read about many things, from psu, to bad ram, to bad motherboard or cpu, it seems that it could be basically anything. I think i am still within RMA time with newegg as i received the components less than 1 month ago.

 

So yeah, what would you guys do, do you experience similar issues? what do you suggest i try to fix the problem?

 

Thanks for your input! I might edit my post later if I think about more things i've done.

 

 

 

DxDiag.txt

[G-ZooM]

Have you tried changing in the Bios the behavior of what should the PC do after a power outage? If you can make it so it doesn't restart but keep it shut down, it could help you rule out a few issues.

 

Also try to leave it on the Bios UI for a while, see if it reboots randomly as well or not

[Howitz]

2 minutes ago, G-ZooM said:

Have you tried changing in the Bios the behavior of what should the PC do after a power outage? If you can make it so it doesn't restart but keep it shut down, it could help you rule out a few issues.

 

Also try to leave it on the Bios UI for a while, see if it reboots randomly as well or not

I can sure do this, also what would keeping it powered down after power outage what issues do you think it would help ruling out?

[G-ZooM]

It's probably obvious but... Have you checked that the power switch and reset button pins are well seated on the motherboard? Could the reset button itself be faulty?

I would try to unplug the reset button pins to make sure it's not related to the case.

 

Also, a long shot, but I've seen this behavior in PCs with bad USB plugs making random short circuits.

[G-ZooM]

2 minutes ago, Howitz said:

I can sure do this, also what would keeping it powered down after power outage what issues do you think it would help ruling out?

If it's just a reboot, it could be the reset button or any kind of request made by the bios, a device or something. If it's a full shutdown it's more likely power related, but not 100% sure either...

[Howitz]

2 minutes ago, G-ZooM said:

It's probably obvious but... Have you checked that the power switch and reset button pins are well seated on the motherboard? Could the reset button itself be faulty?

I would try to unplug the reset button pins to make sure it's not related to the case.

 

Also, a long shot, but I've seen this behavior in PCs with bad USB plugs making random short circuits.

been thinking the same so i straight up unplugged the reset button from the mobo. only thing plugged to mobo from front panel right now is power and light.

[Howitz]

3 minutes ago, G-ZooM said:

It's probably obvious but... Have you checked that the power switch and reset button pins are well seated on the motherboard? Could the reset button itself be faulty?

I would try to unplug the reset button pins to make sure it's not related to the case.

 

Also, a long shot, but I've seen this behavior in PCs with bad USB plugs making random short circuits.

should i try unplugging front usb panel of the case from the mobo?

[FeeNiX]

I recently watched a content piece from another YouTuber talking how some 30 series cards can randomly have an extremely short, extra high draw, and if your PSU cannot meet that need the computer shuts down. They also did discuss ripple, but I'm not sure if that was the same issue or not. I believe it was said that the spike on the 3090 could be as high as several hundred watts. If you have a higher draw PSU I would try swapping. I can look for the link to the piece but I don't know if I'm allowed to post it.

 

Also, I was told by some techs a while back, "Intermittent issues? Try swapping the powersupply."

 

Basically, given your load amount, I'd put money on the power supply. Perhaps bump to 1000?

Edited October 15, 2022 by FeeNiX

[G-ZooM]

Troubleshooting is all about removing as many variables as possible

[G-ZooM]

5 minutes ago, FeeNiX said:

I recently watched a content piece from another YouTuber talking how some 30 series cards can randomly have an extremely short, extra high draw, and if your PSU cannot meet that need the computer shuts down. They also did discuss ripple, but I'm not sure if that was the same issue or not. I believe it was said that the spike on the 3090 could be as high as several hundred watts. If you have a higher draw PSU I would try swapping. I can look for the link to the piece but I don't know if I'm allowed to post it.

Also I was told by some techs a while back, "Intermittent issues? Try swapping the powersupply."

Basically given your load amount, I'd put money on the power supply. Perhaps bump to 1000?

Or if you have a cheap GPU lying around it could be worth swapping it to see if it keeps happening. Might be less trouble than another PSU.

[Howitz]

3 minutes ago, FeeNiX said:

I recently watched a content piece from another YouTuber talking how some 30 series cards can randomly have an extremely short, extra high draw, and if your PSU cannot meet that need the computer shuts down. They also did discuss ripple, but I'm not sure if that was the same issue or not. I believe it was said that the spike on the 3090 could be as high as several hundred watts. If you have a higher draw PSU I would try swapping. I can look for the link to the piece but I don't know if I'm allowed to post it.

 

Also, I was told by some techs a while back, "Intermittent issues? Try swapping the powersupply."

 

Basically, given your load amount, I'd put money on the power supply. Perhaps bump to 1000?

definetely going to try to plug the graphics card to its own 2nd power supply and see if the issue is still there. I can do that right? no need to outright replace the power supply if i can have one take the gpu load and another take the rest of the system?

[FeeNiX]

My reading from the rules seems to indicate that since this isn't self promotion, I can post it.

 

https://youtu.be/zG71dQ9qRbc

[FeeNiX]

4 minutes ago, Howitz said:

definetely going to try to plug the graphics card to its own 2nd power supply and see if the issue is still there. I can do that right? no need to outright replace the power supply if i can have one take the gpu load and another take the rest of the system?

I would be very uncomfortable with the dual PSU idea, but I don't have and data on doing that either way. A low end Video card would be my preferred way to test versus two PSUs

The addition of another powersupply won't rule out some other sort of issue with PSU. Make sure each cable into the card has its own cable direct to the powersupply if possible.

 

If it is the PSU, I would not permanently run on two PSUs.

 

Also, see how sometimes they add the watt values across all the different voltages, so you may be maxing out your 12V wattage.

[Howitz]

16 minutes ago, G-ZooM said:

If it's just a reboot, it could be the reset button or any kind of request made by the bios, a device or something. If it's a full shutdown it's more likely power related, but not 100% sure either...

the bios is already set to shut down on ac power failure and the system reboots so what would that indicate? if it reboots instead of staying shut down would that rule out the psu?

[Howitz]

5 minutes ago, FeeNiX said:

I would be very uncomfortable with the dual PSU idea, but I don't have and data on doing that either way. A low end Video card would be my preferred way to test versus two PSUs

The addition of another powersupply won't rule out some other sort of issue with PSU. Make sure each cable into the card has its own cable direct to the powersupply if possible.

 

If it is the PSU, I would not permanently run on two PSUs.

 

Also, see how sometimes they add the watt values across all the different voltages, so you may be maxing out your 12V wattage.

i definetely have a lower end gpu on hand, a fricking gtx560 lol.

[Howitz]

alright so tomorrow i will swap the rtx3080 to the gtx560 and if it still crashes that will rule out the gpu and psu, if it doesn't crash anymore then it's either the psu or the gpu. (most likely the psu)

[G-ZooM]

20 minutes ago, Howitz said:

definetely going to try to plug the graphics card to its own 2nd power supply and see if the issue is still there. I can do that right? no need to outright replace the power supply if i can have one take the gpu load and another take the rest of the system?

Ultimately we will all have to do it with RTX 5000 cards, you're just a time traveler!

 

In an AC situation I wouldn't recommend it, but with DC 12V as long as you don't mix up the two PSUs on the same card it's fine.

[FeeNiX]

That single test won't help. If it still crashes, it could be anything in the system other than the GPU. If it doesn't crash, it is GPU or PSU (or the old GPU is bad which would take us back to it could be anything.). The problem is you're testing with two independant variables, instead of just "bad hardware" you're also testing for a"bad combination due to power draw". 

 

Your GPU takes 3x PCIe 8pins, and your PSU has only 2? Does your board also take a 6 pin?

 

Don't mix up EPS 8 pin and PCIE 8 pin they are not interchangeable.

 

You're so close to the limit I'd actually up your PSU regardless. If you're in the Chicagoland area I can go digging through my shelves to see if I have any larger PSUs for testing, but I'm reasonably sure I do not.

 

Sorry for the delay, I'm typing on my phone.

 

If you've got a rig with a larger supply you could temporarily use it in your sister's box for stability testing. Or if your personal rig is less demanding you could put the 750 PSU in it. 

 

EDIT: In this case your single best test would be different psu of a higher wattage. If you're looking for a new PSU, get one with enough PCIE plugs for motherboard and card. (4?)

 

If you do the two psu test you're going to need to manually short the startup pins on the second supply mother board connector to get it to turn on. Messing that up could hurt I think. Do you have an attachment to do that (typically used to cycle a water pump during fill up of a custom loop.

[G-ZooM]

18 minutes ago, Howitz said:

the bios is already set to shut down on ac power failure and the system reboots so what would that indicate? if it reboots instead of staying shut down would that rule out the psu?

I'm not 100% sure but I think it rules out a power supply issue in terms of outage... If it were just a few milliseconds of outage it would either shutdown or run on the capacitors, but not a reboot. If it's a spike problem from the RTX 3080 transient power then I have no idea what the behavior would be

 

Hopefully your tests will be positive and your sister can enjoy the GTX 560 for streaming on MySpace.

[FeeNiX]

9 minutes ago, G-ZooM said:

I'm not 100% sure but I think it rules out a power supply issue in terms of outage... If it were just a few milliseconds of outage it would either shutdown or run on the capacitors, but not a reboot. If it's a spike problem from the RTX 3080 transient power then I have no idea what the behavior would be

 

Hopefully your tests will be positive and your sister can enjoy the GTX 560 for streaming on MySpace.

I would not consider that to eliminate the PSU. You're on a battery backup. That would eliminate wall power outage, unless you think both the battery and the PSU have issues.I think tripping the current protections inside the PSU might not look the same as AC Power loss.

 

Hopefully I am making sense. We've got a prescription stimulant medication shortage in the USA so I'm operating half asleep.

Edited October 16, 2022 by FeeNiX
Updating with explanation

[G-ZooM]

1 minute ago, FeeNiX said:

I would not consider that to eliminate the PSU. You're on a battery backup. That would eliminate wall power outage, unless you think both the battery and the PSU have issues.

I was thinking of an outage on the 12V DC that the PSU outputs. If the PSU is starting to fail and/or too much current is pulled that would lower the voltage to dangerous territory and create an outage for the motherboard and everything after it.

[FeeNiX]

1 minute ago, G-ZooM said:

I was thinking of an outage on the 12V DC that the PSU outputs. If the PSU is starting to fail and/or too much current is pulled that would lower the voltage to dangerous territory and create an outage for the motherboard and everything after it.

Wouldn't that trip under current protection? Hopefully LTT Lab will let us send in "what if" scenarios to see what all happens.

[FeeNiX]

49 minutes ago, Howitz said:

alright so tomorrow i will swap the rtx3080 to the gtx560 and if it still crashes that will rule out the gpu and psu, if it doesn't crash anymore then it's either the psu or the gpu. (most likely the psu)

Your powersupply voltage table has your max 12 volt rail output as 750w and i think that is where all of your load is pulling from, as you don't have spinning rust, or an optical drive. If you add the wattage numbers for the other voltages it seems that they all add up to that over 30% discussed in the content piece. I don't think you have the extra head room basically.

 

You could try turning off all unneeded motherboard integrated peripherals (USBs, audio, etc) But I have no idea if that will save you any watts on 12v.

 

Also for testing make sure you turn off any auto over locking features on the CPU, like "multi core enhancement" or whatever they call it. I believe some manufacturers leave that on by default still, so you're already sort of overclocking. I'm not sure if that is still something that exists.

 

That could account for why your power draw is higher than PSU calculators estimate. 

[Howitz]

15 hours ago, FeeNiX said:

That single test won't help. If it still crashes, it could be anything in the system other than the GPU. If it doesn't crash, it is GPU or PSU (or the old GPU is bad which would take us back to it could be anything.). The problem is you're testing with two independant variables, instead of just "bad hardware" you're also testing for a"bad combination due to power draw". 

 

Your GPU takes 3x PCIe 8pins, and your PSU has only 2? Does your board also take a 6 pin?

 

Don't mix up EPS 8 pin and PCIE 8 pin they are not interchangeable.

 

You're so close to the limit I'd actually up your PSU regardless. If you're in the Chicagoland area I can go digging through my shelves to see if I have any larger PSUs for testing, but I'm reasonably sure I do not.

 

Sorry for the delay, I'm typing on my phone.

 

If you've got a rig with a larger supply you could temporarily use it in your sister's box for stability testing. Or if your personal rig is less demanding you could put the 750 PSU in it. 

 

EDIT: In this case your single best test would be different psu of a higher wattage. If you're looking for a new PSU, get one with enough PCIE plugs for motherboard and card. (4?)

 

If you do the two psu test you're going to need to manually short the startup pins on the second supply mother board connector to get it to turn on. Messing that up could hurt I think. Do you have an attachment to do that (typically used to cycle a water pump during fill up of a custom loop.

I think you've just fixed it. I've just read the part about your pci-e connectors. I do indeed only have 2 connected to the card (Haven't mismatched cpu and gpu cables tho), but they have a second daisy chained connector, so basically one connector is plugged in two of the card's 8 pin inputs and one of the connector is plugged in the third. Which means the card is supposed to draw 400 watt but is only getting 75 from the board and 150 watt from each of the two cables, (75+150+150) 375 watts rated instead of the 400 needed. Rookie mistake I presume.

 

Now I 100% saw the card draw in the 390's watts on hardware monitor and it worked in aida64 stable for 30 minutes but it could still be the cause of the crashes if the card does have power requirement spikes likes the one you guys have been talking about. I'll keep it on light load until the new 1000 watt psu we purchased arrives.

 

We are definitely using a PSU that is very at the limit and this other problem linked to it is giving me high hopes that swapping for a 1000 watt might fix the issue.

 

Also, I am worried of damages to the circuitry of either the motherboard or the power supply/cables or event the rtx3080 itself with the way i have it plugged, do you guys think there is a risk in supplying it with only 375 watts from the 2 cables? could something burn out? if yes I will just totally unplug everything.

[FeeNiX]

I honestly don't know, however my guess is that there are potential negative effects, as this is why the protection exists in power supplies. The thing I don't know is if you can trip the protection enough times that it stops protecting for some reason. I also think that non graceful reboots can potentially cause windows installation corruption. You could try limiting the power budget in an overlooking app for your video card, which would bring the voltage down, technically. But given transients exceed the budget normally, it might not help.

 

Since I couldn't eliminate the possibility of damage, and the potential damage could be monetarily high, I'd just wait the few days until the new power supply arrives. You could put that spare 560 in temporarily, just to see if the reboots still happen, games won't run as great but the computer will run. I'm pretty sure the GTX 560 draws much less power.

 

I rewatched that content piece, the 3080 is particularly mentioned for high transients, not the 3090.