Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

PC instantly shuts off under load?

Hello I have had an issue where my pc will instantly shut off and require a power cycle (flip my power supply on and off) to be able to turn on again. This has happened only while in games (many different games too).


CPU: AMD Ryzen 9 5900X 3.7 GHz 12-Core Processor
CPU Cooler: NZXT Kraken Z73 73.11 CFM Liquid CPU Cooler
Motherboard: Asus ROG Crosshair VIII Dark Hero ATX AM4 Motherboard
Ram: G.Skill Trident Z Neo 64 GB (2 x 32 GB) DDR4-3600 CL16 Memory
SSD/HDD: Samsung 980 Pro 1 TB M.2-2280 NVME Solid State Drive <- windows 11 is on that
Western Digital Black SN850 1 TB M.2-2280 NVME Solid State Drive
2x Samsung 850 EVO-Series 500 GB 2.5" Solid State Drive
Samsung 860 Evo 1 TB 2.5" Solid State Drive
GPU: Asus GeForce RTX 3090 24 GB TUF GAMING OC Video Card
PSU: SeaSonic PRIME Ultra Titanium 1000 W 80+ Titanium Certified Fully Modular ATX Power Supply
Chassis: Lian Li PC-O11 Dynamic Razer ATX Full Tower Case

I've had this computer build for about 5 months. I've had almost no issues at all with it (other than me being dumb and connecting a USB wrong in an external usb hub). I've had a windows 11 OS since it was released but 3 days ago I reinstalled it fresh after telling windows to keep no personal files on the reinstall. I also did DDU and had to reinstall all my programs from scratch. I took opportunity to OC my 3090. The step I took Core Voltage to maximum Power Limit to max Core clock increment by 10 until system instability during stress test decrement and stress test again repeat for memory clock but by increments of 25

After doing like 4 different stress tests with absolutely nothing wrong (+110 core clock +625 memory clock) I started gaming and I ran into the pc shutting off randomly. I gradually railed back the OC until I reset literally everything (including core voltage and power limit) to their default values again. The PC STILL randomly instantly shuts off during games. (nothing intensive even needs to happen). The consistent games I've been testing with are Vermintide 2, Deep rock galactic (this one is consistent on instantly killing my PC on load), and Red dead redemption 2.

I have power settings in windows on Balanced and Nvidia Control Panel has Prefer High Performance. I assume it has something to do with my power supply but I feel like 1000W should be enough to not instantly shut off when I'm running my specs non OC (also it was fine for the months prior to my fresh install and OC attempt).

After asking people (in the end they couldn't find the solution) I changed my AI overclock setting from Auto to D.O.C.P and back. Neither setting fixes it. I tried reseating my gpu by taking it off the vertical riser I had it on. I plugged it directly onto my mobo and it still didn't fix itself.
I tried OCCT CPU Stress Test, GPU Stress Test, and their PSU Stress test and nothing went wrong. I tried Time Spy Stress Test and the regular benchmark and nothing went wrong. Prime95 and nothing went wrong. I tried doing memcheck on my ram and nothing showed up. I updated my bios and checked for a vbios update and it didn't fix it.

There are 2 things I DID manage to find out about the problem though.
I swapped my 3090 TUF for my old 980ti and the problem did not happen even when I opened up Deep Rock Galactic (which normally instantly crashed my pc on load) and Vermintide 2. So it is isolated to my newer GPU BUT I'm not sure if that means it's the GPU itself has a problem or if the PSU can't handle something with the 3090 (since it is a big increase in power draw)
This may be a big find but unsure. I ran Furmark Stress Test and it would consistently shut down my pc while all the other stress tests did not. I went into msi afterburner and I set my power limit to 80% and I no longer have any problems during the benchmark. I'm still unsure what to make of this since, for one, I can't tell if this means the PSU or GPU is messed up since it's the power draw part of the GPU, and second, the only thing that happened since me not having this problem is reinstalling Windows 11 fresh and me OCing my card (which I reverted back to default and the issues still popped up).

Unsure if this last part is relevant but when the instant shut off happens I can't power back on unless turning the power supply switch off and letting my mobo turn off completely. During this the only things lighting up on my mobo are the start button that's directly on the mobo and the 2 ram sticks I got on there. No case fans, CPU AIO cooler, or gpu are lit up.

Is there anything I can do to fix my problem someway somehow?

Edited by EunoiaJean
Adding FULL specs
Link to comment
Share on other sites

Link to post
Share on other sites

Try turn on logging in GPU-Z and check back the power usage of your 3090 after it shut down. 

| Intel i7-3770@4.2Ghz | Asus Z77-V | Zotac 980 Ti Amp! Omega | DDR3 1800mhz 4GB x4 | 300GB Intel DC S3500 SSD | 512GB Plextor M5 Pro | 2x 1TB WD Blue HDD |
 | Enermax NAXN82+ 650W 80Plus Bronze | Fiio E07K | Grado SR80i | Cooler Master XB HAF EVO | Logitech G27 | Logitech G600 | CM Storm Quickfire TK | DualShock 4 |

Link to comment
Share on other sites

Link to post
Share on other sites

I posted the log right up until the crash here. It's really short because I started my computer and started the log and started fur mark and it crashed a few seconds in. Also I forgot to mention in case it's important I saw some other posts about it. I am connecting my 3090 to my PSU through 2 pcie 6+2 Power connectors. The ones where the 2 pin is separated by the 6 pin but come from the same cable.

crash GPU z Log.txt

Link to comment
Share on other sites

Link to post
Share on other sites

You said 80% power limit is ok? Can you try that and upload GPU-Z log with that power limit set?

| Intel i7-3770@4.2Ghz | Asus Z77-V | Zotac 980 Ti Amp! Omega | DDR3 1800mhz 4GB x4 | 300GB Intel DC S3500 SSD | 512GB Plextor M5 Pro | 2x 1TB WD Blue HDD |
 | Enermax NAXN82+ 650W 80Plus Bronze | Fiio E07K | Grado SR80i | Cooler Master XB HAF EVO | Logitech G27 | Logitech G600 | CM Storm Quickfire TK | DualShock 4 |

Link to comment
Share on other sites

Link to post
Share on other sites

20 minutes ago, xAcid9 said:

You said 80% power limit is ok? Can you try that and upload GPU-Z log with that power limit set?

This log is from a 5 minute long log. I'm leaving the furmark stress test still running as we speak at 80% power limit. Still no issues whatsoever.

80% power limit FurMark Stress Test LOg.txt

Link to comment
Share on other sites

Link to post
Share on other sites

11 minutes ago, EunoiaJean said:

This log is from a 5 minute long log. I'm leaving the furmark stress test still running as we speak at 80% power limit. Still no issues whatsoever.

80% power limit FurMark Stress Test LOg.txt 219.56 kB · 1 download

it looks like based on the power draw hitting 350 or so watts on the gpu is whats crashing it. Maybe its the gpu shutting down to to some kind of filing capacitor or the psu cant handle it with the 6+2s on the same wires. if you know sommeone with a similarly specced psu maybe you can test it out with their psu?

Link to comment
Share on other sites

Link to post
Share on other sites

I unfortunately don't know anyone with a psu as big a wattage as mine. Is there substitute wire or something I can possibly order to replace the 6+2's or is it not that sort of problem

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, EunoiaJean said:

I unfortunately don't know anyone with a psu as big a wattage as mine. Is there substitute wire or something I can possibly order to replace the 6+2's or is it not that sort of problem

it looks like it has more spots on the psu for pcie power. do you have a 2nd cable that you can run from the psu thats separate? based on my reading it comes with 8 6+2s

Link to comment
Share on other sites

Link to post
Share on other sites

i would run 2 separate cables frm the psu and only use 1 of the 6+2s on each cable

 

also you can touch the cables going into the gpu and see if they are getting hot. they shouldnt

 

 

edit 2

if you dont have them for some reason ive called seasonic customer service before and ordered replacement cables

Link to comment
Share on other sites

Link to post
Share on other sites

47 minutes ago, EunoiaJean said:

This log is from a 5 minute long log. I'm leaving the furmark stress test still running as we speak at 80% power limit. Still no issues whatsoever.

80% power limit FurMark Stress Test LOg.txt 219.56 kB · 3 downloads

Sorry can you do it again with Superposition instead of Furmark? 😅 The card is throttling hard at 80% in Furmark.. <800mhz core clock. 
Or just run Vermintide 2. 

| Intel i7-3770@4.2Ghz | Asus Z77-V | Zotac 980 Ti Amp! Omega | DDR3 1800mhz 4GB x4 | 300GB Intel DC S3500 SSD | 512GB Plextor M5 Pro | 2x 1TB WD Blue HDD |
 | Enermax NAXN82+ 650W 80Plus Bronze | Fiio E07K | Grado SR80i | Cooler Master XB HAF EVO | Logitech G27 | Logitech G600 | CM Storm Quickfire TK | DualShock 4 |

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, LavenderPhantom78139 said:

i would run 2 separate cables frm the psu and only use 1 of the 6+2s on each cable

 

also you can touch the cables going into the gpu and see if they are getting hot. they shouldnt

 

 

edit 2

if you dont have them for some reason ive called seasonic customer service before and ordered replacement cables

I think I have the exact setup you're mentioning? I'm unsure of the terminologies but I have 2 separate 6+2's plugged into my PSU. The 6+2's each go into one of my GPU's 8 pin slots.

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, xAcid9 said:

Sorry can you do it again with Superposition instead of Furmark? 😅 The card is throttling hard at 80% in Furmark.. <800mhz core clock. 
Or just run Vermintide 2. 

https://benchmark.unigine.com/leaderboards/?utm_source=launcher&utm_medium=&utm_term=&utm_content=9663c466f8864898996788a0642cdd77&utm_campaign=superposition_1.1.8628
So I had furmark running for like 30 minutes just a little bit before this and then I ran superposition (This is all still at 80% power limit) and my pc crashed. so it looked like 80% power limit isn't immune to my issue either but it definitely helps with stability and keeping it bay most of the time I guess. I ran it again after power cycling and it went the whole benchmark fine. For some reason it says windows 10 but I'm on Windows 11 100% I guess Windows 11 isn't in their detection database or something. I went away for a bit from my desk so the GPU Z Log may have been running a bit longer than the benchmark sorry about that. 

Superposition Log.txt

Superposition_Benchmark_v1.1_14321_1638709481.png

Link to comment
Share on other sites

Link to post
Share on other sites

5 minutes ago, Bad5ector said:

Might be this:

 

tierlist.png.79bf2ccc6e07e5edaf128dcee51318ed.png

How would I go about overprovisioning wattage? Also would the problem in the screenshot still be the case if I've been fine for over 5 months with the rig? The problem did arise quite suddenly and I am power limiting the GPU to 80% which I feel like is a lot given it's 1000W PSU (unless the PSU is REALLY a lost cause)

Link to comment
Share on other sites

Link to post
Share on other sites

5 minutes ago, EunoiaJean said:

How would I go about overprovisioning wattage?

What it means is if you want to go with one of these PSUs then it is best to overprovision. So instead of 1000 watt, get a 1200 watt. It says this in the very next line but uses the terms 1KW and 1.2KW instead. Maybe check out the list and see if your model is one of those affected. There seems like there is a way to work around the OCP if you're daring enough.

Link to comment
Share on other sites

Link to post
Share on other sites

these issues are always 99% psu issue. 

thinking i have a 1000w super psu it cant be it, is exactly what it often ends up being 

 

seasonic isn't that great anyway,  no idea how they got the good reputation (well, ofc i have an idea...)

 

try corsair hx rmi etc, maybe superflower 

My Nvidia GPU undervolt tutorial  !

 

 

The direction tells you... the direction

-Scott Manley, 2021

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

 Seasonic | Focus [OneSeasonic refreshes] GX / GM / PX / SGX - Prime [8] GX (<=750W / =>1000W) / (Ultra) Gold =>1000W / PX =>1000W / (Ultra) Platinum =>1000W / TX =>1000W / (Ultra) Titanium =>1000W / PRIME Snow Silent Gold [550W] - Connect


It seems my PSU is on the the list as needing to be over Wattaged. I know Mark Kaine did say that it's prob a psu issue but I will say again that I feel like 1000W IS overwattaged for my setup but I don't know at this point. Hands are UP.

Sigh I'd rather not have to deal with janky fixes that might brick the whole thing and risk my RMA. Does that mean I have to buy a completely new PSU? And are we completely sure this is the root cause of the problem? I did swap out my 3090 for my 980ti with no problem so it could still point to the GPU or PSU both honestly. Also seeing how Seasonic has a good reputation and apparently they're bad even with a good reputation what PSU do I even pick at what Wattage?

Link to comment
Share on other sites

Link to post
Share on other sites

You're misinterpreting the root cause. It isn't because your PSU is jank per say, it just has a lower tripping point for OCP. At least that was my understanding. You could go with the same model of Seasonic but just get a 1200watter and you'd be fine. Again, assuming it is your PSU. Though really feel like it is at this point.

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, Bad5ector said:

You could go with the same model of Seasonic but just get a 1200watter and you'd be fine. 

thats unless the ocp still triggers. but you're right this isnt the psu being "trash" more like its not designed for that kind of load (ie high transient spikes) as thats a new thing in these dimensions at least.

 

for example: 750w bequiet straight power + my 3070 = nope nope nope , 5 bsods/day, constant "clicking" noises that eventually trigger shut down/crash.

500w bequiet pure power + my 3070 = silent, cool, no crashes whatsoever...

 

(so obviously the 2 psus have different protections, the straight power supposedly more advanced,  thats what we call "griff ins klo" around here... 😅 )

 

 

My Nvidia GPU undervolt tutorial  !

 

 

The direction tells you... the direction

-Scott Manley, 2021

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, EunoiaJean said:

 Seasonic | Focus [OneSeasonic refreshes] GX / GM / PX / SGX - Prime [8] GX (<=750W / =>1000W) / (Ultra) Gold =>1000W / PX =>1000W / (Ultra) Platinum =>1000W / TX =>1000W / (Ultra) Titanium =>1000W / PRIME Snow Silent Gold [550W] - Connect


It seems my PSU is on the the list as needing to be over Wattaged. I know Mark Kaine did say that it's prob a psu issue but I will say again that I feel like 1000W IS overwattaged for my setup but I don't know at this point. Hands are UP.

Sigh I'd rather not have to deal with janky fixes that might brick the whole thing and risk my RMA. Does that mean I have to buy a completely new PSU? And are we completely sure this is the root cause of the problem? I did swap out my 3090 for my 980ti with no problem so it could still point to the GPU or PSU both honestly. Also seeing how Seasonic has a good reputation and apparently they're bad even with a good reputation what PSU do I even pick at what Wattage?

The PSU is the ussue. It measures too much noice given back from the GPU and shut down. Where, for example, a Corsair RM850x has no issues at all, for a Seasonic you need 1200w to run your rig. Avoid Seasonic with the latest high-end GPUs.

Link to comment
Share on other sites

Link to post
Share on other sites

So the PSU can just start the OCP after about 5 months of regular use and no problems? And if seasonic is to be avoided will I be able to somehow get my money back and maybe buy a different company's PSU?

Link to comment
Share on other sites

Link to post
Share on other sites

I posted this in a PCMasterRace Tech Support Discord yesterday and It was mentioned that my Vram may be fried? Unsure of how to even check that or if it's right but I ran the Time Spy Stress Test about 3 times and passed after that suggestion. 2 of the times I was at 80% power Draw and the third time I was heavily OCed at +120 Core Clock and +600 Memory. I reverted back to default values and even though I passed those stress tests when I ran games it instantly shut off again.


Today, I tried to do a hail mary test and I ran the usual tests to see if it would trip OCP at 100% load again. I ran Furmark and it does NOT seem to be the case even when the GPU spikes past 350W. (Current Maximum Board Power Draw reported by GPU Z is 358.7) Memory Temp is going to 96 degrees C and GPU is averaging 63 C. This is under 99% GPU load. I'm unsure why it isn't tripping the OCP but clearly the problem isn't consistent on OCP I would think. I'll post a log of the FurMark Stress Test at 100% power limit.

 

I also am currently running OCCT's PSU Stress test (been about 7 minutes) and I'm currently still running strong at 100% power limit. I'm logging the OCCT Power Test. Both the CPU and GPU are at 100% usage. One thing to note is that the GPU Memory usage is low at around 3.86% constantly. The CPU and GPU seem to be drawing a lot of power each. CPU Package Power is Sitting at ~134.7W and GPU Power at ~349W. I'm gonna let it run the whole hour but I feel like it's a big sign that it didn't instantly die like before for some unknown reason. If it's not instantly shutting off at maximum power draw from both CPU and GPU both at 100% load then surely something else msut be up? or is OCCT unreliable for testing such things.

furmark stress test.txt

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, EunoiaJean said:

So the PSU can just start the OCP

It's not OCP!

 

37 minutes ago, EunoiaJean said:

Today, I tried to do a hail mary test and I ran the usual tests to see if it would trip OCP at 100% load again. I ran Furmark and it does NOT seem to be the case even when the GPU spikes past 350W. (Current Maximum Board Power Draw reported by GPU Z is 358.7) Memory Temp is going to 96 degrees C and GPU is averaging 63 C. This is under 99% GPU load. I'm unsure why it isn't tripping the OCP but clearly the problem isn't consistent on OCP I would think. I'll post a log of the FurMark Stress Test at 100% power limit.

 

I also am currently running OCCT's PSU Stress test (been about 7 minutes) and I'm currently still running strong at 100% power limit. I'm logging the OCCT Power Test. Both the CPU and GPU are at 100% usage. One thing to note is that the GPU Memory usage is low at around 3.86% constantly. The CPU and GPU seem to be drawing a lot of power each. CPU Package Power is Sitting at ~134.7W and GPU Power at ~349W. I'm gonna let it run the whole hour but I feel like it's a big sign that it didn't instantly die like before for some unknown reason. If it's not instantly shutting off at maximum power draw from both CPU and GPU both at 100% load then surely something else msut be up? or is OCCT unreliable for testing such things.

Furmark and OCCT are static loads. Gaming is dynamic, that's the difference.

 

And again, it's not an OCP issue.

Link to comment
Share on other sites

Link to post
Share on other sites

I MAY have found out something. It is cold where my setup is so I have a space heater on and I've never had my new PC Setup and my heater on at the same time before this Winter Season. I recently had a flood and while my stuff was all untouched I just now looked at my surge protector that I have my PC and 2 monitors directly connected to on and saw that the surge protector was off. Unsure if the flood was even related but I'm putting it for full preface. I totally didn't know you have to replace your surge protector every couple years so I didn't notice this was broken and I have been using that surge protector without replacing it since 2016. I have a theory that it's my surge protector tripping so I ordered some new surge protectors and I'm going to see about doing the tests again after they come and I swap everything onto those. 

(full preface I have no idea if this'll work or not but I'm eliminating all variables hopefully it's not all in vain.)

Quick update on the OCCT PSU test it's still going strong 48 minutes in at 100% power limit and I'm still using chrome with a bunch of tabs just fine with no shut off.


edit: I posted the OCCT PSU Stress test log from HWINFO.

OCCT PSU Stress Test.CSV

Edited by EunoiaJean
Link to comment
Share on other sites

Link to post
Share on other sites

3 minutes ago, --SID-- said:

It's not OCP!

 

Furmark and OCCT are static loads. Gaming is dynamic, that's the difference.

 

And again, it's not an OCP issue.

Oh. well would my post I just posted about maybe be a culprit for the problem? and I know I misunderstood it being about OCP but then can you explain exactly what about my PSU is the problem if all the other I posted about still points to the PSU.

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share


×