PC will reboot sometimes when gaming

[ChristTheGreat]

hey guys!

I have a problem, that I am not able to put my finger on it. I play often since march 2020, and I started having issue, I think, when I swapped my 3700x to 5800x, or with the RTX 3070 (before the 5800x)

The PC will sometimes (not often), reboot while gaming. My fans are running full when gaming, CPU is like 72-75C (warzone + Streamlabs), and GPU is like 70C. I had a voltage offset negative, where I dropped it near auto now, since I thought that was the issue.

Grabbed an RX 6800XT (cause yeah, I have alot of problem with NVENC to record gameplay), and it happened yesterday again. No event in windows, it just say the PC reboot/shutdown, no error.


my rig is updated in my pc spec, I also have a Cyberpower UPS 1350va, which I might remove for testing, cause it's like 7yo. Could it be the PSU? Bios Is at the lastest, driver update, windows has been reinstalled.


Thanks for any help!

 

[AltCapwn]

Hello Christ,

if you check your eventviewer, do you have traces of the crash?

 

[oldwalltree]

Hello Christ,

if you check your eventviewer, do you have traces of the crash?

This go into admin view on event viewer and look for critical errors. If you have a generic code that basically reads out sudden and unexpected loss of power then you have a hardware issue. If its something different google away with the code! It will end up being a windows issue most likely.

This go into admin view on event viewer and look for critical errors. If you have a generic code that basically reads out sudden and unexpected loss of power then you have a hardware issue. If its something different google away with the code! It will end up being a windows issue most likely.

Also have you tired a simple CMOs reset since you've swapped all that hardware in?

 

[tabascosauz]

hey guys!

I have a problem, that I am not able to put my finger on it. I play often since march 2020, and I started having issue, I think, when I swapped my 3700x to 5800x, or with the RTX 3070 (before the 5800x)

The PC will sometimes (not often), reboot while gaming. My fans are running full when gaming, CPU is like 72-75C (warzone + Streamlabs), and GPU is like 70C. I had a voltage offset negative, where I dropped it near auto now, since I thought that was the issue.

Grabbed an RX 6800XT (cause yeah, I have alot of problem with NVENC to record gameplay), and it happened yesterday again. No event in windows, it just say the PC reboot/shutdown, no error.


my rig is updated in my pc spec, I also have a Cyberpower UPS 1350va, which I might remove for testing, cause it's like 7yo. Could it be the PSU? Bios Is at the lastest, driver update, windows has been reinstalled.


Thanks for any help!

Unexpected reboots without BSOD on Ryzen usually either memory stability issue, core voltage issue or SOC voltage issue. The first one will probably not give you any Event Viewer entries, but the latter two are liable to leave WHEA events in Event Viewer. Generally, there will be a red "Critical" event, and an accompanying yellow "Error" event. For any WHEA error, the yellow Error log will give you the details about the source.

Also, if you aren't using it already, get HWInfo. You don't need to monitor it all the time, just let it Auto Start at boot and run in the background. It also logs any WHEA errors that come up over time in a counter at the bottom of the list, makes it easy to identify when something's not quite right.

 

[RJARRRPCGP]

Nothing but that usual "Critical" category message=Probably PSU, MOSFET or VRM issue. It's possible that the 8-pin EPS plug has poor connection. That looks like a case of voltage loss.

Loss of RAM stability=Usually a BSOD with "PAGE_FAULT_IN_NON_PAGED_AREA", "BAD_POOL_HEADER" or "0x0000008E". Or another error that reports corruption that wasn't there before.

 

[Bill_Bright]

Since all depends on good, clean, stable power, I always swap in a known good PSU when troubleshooting hardware problems - if for no other reason than to quickly eliminate the PSU from the equation.

 

[ChristTheGreat]

Hello Christ,

if you check your eventviewer, do you have traces of the crash?

only: The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.

Unexpected reboots without BSOD on Ryzen usually either memory stability issue, core voltage issue or SOC voltage issue. The first one will probably not give you any Event Viewer entries, but the latter two are liable to leave WHEA events in Event Viewer. Generally, there will be a red "Critical" event, and an accompanying yellow "Error" event. For any WHEA error, the yellow Error log will give you the details about the source.

Also, if you aren't using it already, get HWInfo. You don't need to monitor it all the time, just let it Auto Start at boot and run in the background. It also logs any WHEA errors that come up over time in a counter at the bottom of the list, makes it easy to identify when something's not quite right.

I'll look at my settings right now, and post it. it was very stable with my 3700x (for memory)

Nothing but that usual "Critical" category message=Probably PSU, MOSFET or VRM issue. It's possible that the 8-pin EPS plug has poor connection. That looks like a case of voltage loss.

Loss of RAM stability=Usually a BSOD with "PAGE_FAULT_IN_NON_PAGED_AREA", "BAD_POOL_HEADER" or "0x0000008E". Or another error that reports corruption that wasn't there before.

No BSOD or anything else :/ but I'll relook the EPS connector.

Since all depends on good, clean, stable power, I always swap in a known good PSU when troubleshooting hardware problems - if for no other reason than to quickly eliminate the PSU from the equation.

I wish I could haha, got this supervona G2 750w on 7/29/2016, so it's like 4y+ old, I guess it should be fine. I get max (on the UPS) around 500-550w under 1100% GPU and 100% CPU.

 

[Bill_Bright]

I wish I could haha, got this supervona G2 750w on 7/29/2016, so it's like 4y+ old, I guess it should be fine.

That's the problem. Guessing could result in spending money on something else that is not needed.

I recommend you try to borrow from a trusting friend or relative to prove one way or the other - especially before spending money on something else.

 

[ChristTheGreat]

Well It's going to be hard to find one to test, my closest friend that has a psu, would be a 500w at max.


I can RMA the power supply, that wouldn't be a problem, it is 10y warranty

 

[AltCapwn]

I would have borrowed you my PSU but it's a dell 500W .

HWInfo might be able to give more details about the crash tho as proposed by @tabascosauz .

 

[thesmokingman]

View attachment 189967View attachment 189968View attachment 189969

lol, you're using level 3 oc! And you're using a negative offset, why???

The system is obviously NOT STABLE because you're overclock is not stable.

 

[milewski1015]

I'll look at my settings right now, and post it. it was very stable with my 3700x (for memory)

"very stable" based on what? Lack of prior crashes? From what I understand, AMD is still ironing out the kinks with the AGESA for the 5000 series CPUs so it's possible what was "stable" before with your 3700X is not with your 5800X

The system is obviously NOT STABLE.

My thinking as well. OP should disable all OC (CPU and memory) and determine if that stops the crashing

 

[thesmokingman]

My thinking as well. OP should disable all OC (CPU and memory) and determine if that stops the crashing

Oh yea his settings contradict each other, oc control set to auto, yet uses asus level 3 oc enhancement which is an extreme level, then slaps on a negative voltage offset on top of that, and I wouldn't be surprised if a lil curve enhancement isn't used.

It ain't the PSU LAMO.

 

[Alexa]

Disable "Global C-State Control" in the BIOS. Fixed every single random reboot on my 3900X. Not sure if it will help here but worth a try.

 

[AltCapwn]

Disable "Global C-State Control" in the BIOS. Fixed every single random reboot on my 3900X. Not sure if it will help here but worth a try.

Had some issue with Intel CPU before because of that too.

lol, you're using level 3 oc! And you're using a negative offset, why???

The system is obviously NOT STABLE because you're overclock is not stable.

@ChristTheGreat if you ignore the fact that he's toxic, he has a point. You can try and put your system to default values and see if your system is more stable, then overclock your stuff following best practices based on your platform.

@thesmokingman You make it seems like everyone should know that. Make us learn and explain to us why it shouldn't be done like that instead of being passive aggressive, or just give some guide or reference at least.

 

[xtreemchaos]

im leaning towards a ram issue id drop the speed and see what happens bud.

 

[tabascosauz]

only: The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.


I'll look at my settings right now, and post it. it was very stable with my 3700x (for memory)

No BSOD or anything else :/ but I'll relook the EPS connector.


I wish I could haha, got this supervona G2 750w on 7/29/2016, so it's like 4y+ old, I guess it should be fine. I get max (on the UPS) around 500-550w under 1100% GPU and 100% CPU.

You should never carry your settings over to a CPU of an entirely different family......start over from a CMOS reset. Leave the auto OC features alone, just set DOCP. At this point I highly doubt its a memory issue but you can easily test that later with HCI or TM5.

Frankly I don't even recall Level 3 to be an option on 500-series boards. Is that a holdover from the 400-series UEFI, that they've tacked on 5000 support to?

If you want to undervolt Ryzen 5000, the curve optimizer is the way to go. Lots of good info here, whether u just want to undervolt or want to undervolt+OC at the same time

AMD Curve Optimizer any guides / experience

I have my 5800X up and running, and would like to do a little undervolting, however I am fairly confused. 1) How can I tell how much of an offset to set for each core? From what I read, it's during idle when too much can cause instability, so how the hell do I test that, and how can I even know...

www.techpowerup.com

 

[ChristTheGreat]

lol, you're using level 3 oc! And you're using a negative offset, why???

The system is obviously NOT STABLE because you're overclock is not stable.

Negative offset, because the voltage is way too high on stock, issue seen on alot alot of board But hey, thanks for pointing that, the 3700x was using LVL 3, because on other level, the boost wasn't normal, so I thought I would leave this on LVL3, because that damn Asus board always been special, IMO. I've not doing advanced settings in computer for a while, so I can miss sometime.

Just don't be rude man. One day, you will ask somethign and someone will find out that somethign is stupid in your configuration or whatever!

I do not take it personal, don't worry

"very stable" based on what? Lack of prior crashes? From what I understand, AMD is still ironing out the kinks with the AGESA for the 5000 series CPUs so it's possible what was "stable" before with your 3700X is not with your 5800X


My thinking as well. OP should disable all OC (CPU and memory) and determine if that stops the crashing

since I can't run prime95, as my normal habit, I let the computer running a couple of benchmark for CPU for hours, been running WCG for a while (like 3-4 days, but CPU temps was like 84C' so I stopped )




By the way, thanks guys for your time.

 

[tabascosauz]

Negative offset, because the voltage is way too high on stock, issue seen on alot alot of board But hey, thanks for pointing that, the 3700x was using LVL 3, because on other level, the boost wasn't normal, so I thought I would leave this on LVL3, because that damn Asus board always been special, IMO. I've not doing advanced settings in computer for a while, so I can miss sometime.

Just don't be rude man. One day, you will ask somethign and someone will find out that somethign is stupid in your configuration or whatever!

I do not take it personal, don't worry



since I can't run prime95, as my normal habit, I let the computer running a couple of benchmark for CPU for hours, been running WCG for a while (like 3-4 days, but CPU temps was like 84C' so I stopped )

No currently up to date board runs Ryzen 3000 at "high voltage" out of the box as is (default is PBO Auto = essentially disabled). If this is the 1.5V you're seeing at idle, there must be a billion threads by now out there explaining why this is been normal behaviour since Ryzen 3000. I'm not going to waste time on this - just google "ryzen 1.5v idle".

As long as stock settings run an all-core benchmark at 1.35V or lower, it's not being overvolted.

All you're doing with a manual offset is killing your single-core boost performance, because PB2 is designed to leverage up to 1.45-1.5V in light boost to hit those sweet numbers. If your cooling isn't ITX-grade and necessarily extremely limited, lay off the "dumb" method of undervolting.

 

[ChristTheGreat]

No currently up to date board runs Ryzen 3000 at "high voltage" out of the box as is (default is PBO Auto = essentially disabled). If this is the 1.5V you're seeing at idle, there must be a billion threads by now out there explaining why this is been normal behaviour since Ryzen 3000. I'm not going to waste time on this - just google "ryzen 1.5v idle".

As long as stock settings run an all-core benchmark at 1.35V or lower, it's not being overvolted.

All you're doing with a manual offset is killing your single-core boost performance, because PB2 is designed to leverage up to 1.45-1.5V in light boost to hit those sweet numbers. If your cooling isn't ITX-grade and necessarily extremely limited, lay off the "dumb" method of undervolting.

I saw that a few board were over volting CPU ( more than the specified limit) but I might be wrong.

Running a Noctua NH-U14S, with a noctua ippc 2000 pmw, 2 front fran noctua (same), and a fractal gp-12 pwm at the back. Still, cinebench r20 get to 85C, on auto voltage. But anywaybi don't do CPU full load every time haha.

Just set back the performance enhancer to default (from a cmos reset), set D.O.C.P to my ram speed, just set yhe Fabric to 1600 to make sure it match .

Sad that 5800x is a hot chip

 

[tabascosauz]

I saw that a few board were over volting CPU ( more than the specified limit) but I might be wrong.

Running a Noctua NH-U14S, with a noctua ippc 2000 pmw, 2 front fran noctua (same), and a fractal gp-12 pwm at the back. Still, cinebench r20 get to 85C, on auto voltage. But anywaybi don't do CPU full load every time haha.

Just set back the performance enhancer to default (from a cmos reset), set D.O.C.P to my ram speed, just set yhe Fabric to 1600 to make sure it match .

Sad that 5800x is a hot chip

And pray tell, what is this "specified limit" on a chip that dynamically varies its Vcore?

It doesn't have to be so hot, that's why I linked you to the Curve Optimizer thread. The beauty of it is that it can be done on a per-core basis that you can tweak to your liking according to the silicon quality distribution on your CPU.

It's also the reason why I opted for the 5900X instead for my NH-C14S to cool; unfortunately it's not necessarily "cooler" just flipped on its head, as multi core benchmarks always top out at 72C, but high single thread boost in games can take CCD1 up to 82C regularly.

 

[DeathtoGnomes]

What case does the OP have? is it getting enough airflow? what Heatsink is being used?

I dont see how a PSU can affect the temps other than being overclocking.

 

[ChristTheGreat]

And pray tell, what is this "specified limit" on a chip that dynamically varies its Vcore?

It doesn't have to be so hot, that's why I linked you to the Curve Optimizer thread. The beauty of it is that it can be done on a per-core basis that you can tweak to your liking according to the silicon quality distribution on your CPU.

It's also the reason why I opted for the 5900X instead for my NH-C14S to cool; unfortunately it's not necessarily "cooler" just flipped on its head, as multi core benchmarks always top out at 72C, but high single thread boost in games can take CCD1 up to 82C regularly.

I am sure, my Chip isna shitty one, like all my CPU I had in my life , the only very good I had, was my opteron 165, already tested at 2.7ghz for stock voltage and a 2500k @ 4.8ghz. My 1700, 2600x, 3600x, 3700x were not good clocker.

Well, I'll test woth level 3 removed as per Thesmokingman saw, and I hope this fix it.

What case does the OP have? is it getting enough airflow? what Heatsink is being used?

I dont see how a PSU can affect the temps other than being overclocking.

Fractal define C with tempered glass, nh-u14s, with noctua ippc 2000rpm pwm 140mm for cpu, 2x front. At 75C, fan speed is set to 100% (loud but alot of air flow)

 

[DeathtoGnomes]

Fractal define C with tempered glass

yea thats the case with those side vents that restrict air flow. TBH, you need a better case that will handle a couple more fans. It might be pretty but its not practical in this instance.

nh-u14s, with noctua ippc 2000rpm pwm 140mm for cpu

thats not really enough for a 5800X, it starts out at 105w (idle) so check the TDP specs for the cooler to make sure its enough for the 5800X when its overclocking itself.

EDIT: one thing you can test without a hassle, take the tempered glass panel off and point a desktop fan at the CPU and see if temps are better.

Spoiler: Side Fan

 

[JcRabbit]

I have a problem, that I am not able to put my finger on it. I play often since march 2020, and I started having issue, I think, when I swapped my 3700x to 5800x, or with the RTX 3070 (before the 5800x)

The 3000 series GPUs are known to have issues with older power supplies.

The problem has nothing to do with how much total power the system is consuming, but with the overcurrent protection in many 'older' PSUs being 'too sensitive'. The PSU automatically turns off if it detects a **sudden** surge in power demand to protect itself and your equipment - the cause would normally be a short circuit somewhere, so before it has time to burn down your entire system the PSU turns all power off.

The 3,000 series cards seem to have sudden spikes in terms of power demand, and even though total system wattage demand is not more than what your PSU can handle, it is never-the-less a sudden event, so the PSU thinks it is dealing with a short circuit and shuts down without warning.

You can search the net for a ton of evidence on these issues (just google '3090 reboot psu seasonic' for instance, although this is in no way restricted to Seasonic PSUs or even 'lower' wattage PSUs).

Kind of amusing as I too have just been bitten by this: when I got my current system back in January 2019 (9900K, 2080ti) I wanted a 1000W Seasonic Prime Titanium PSU but those were nowhere to be found so I had to settle for the 850W version. Still, more than enough for my needs.

A couple of months ago I managed to get my hands on an Asus Strix 3090. The card can peak to 450+Watts, but even with my overclocked 9900K total power usage was well within what the 850W Seasonic PSU could handle (plus in tests and reviews the 850W units were able to sustain a 1000W load indefinitely without shutting down).

Despite having previously heard stories about problems with these Seasonic units and 3080/3090 cards, I thought I was in the clear as none of my games caused my PC to reboot (I'm running 4k @120Hz). Rock solid stable for all this time.

Until I went back to Doom Eternal just a few days ago, that is. That thing is so well optimized it runs at 120FPS on a LG CX 4K display with every graphic setting turned as high as it will go. Within MINUTES of playing that game my PC suddenly turned off. No warning, nothing. Sudden black screen, no signal, system reboots.

That's when I knew this was a power problem. I was keeping an eye on temperatures and GPU power consumption with AIDA64 on my secondary monitor, so I know that wasn't the problem (GPU about 70C, using about 400W+ as normal).

So today I got an Asus ROG Thor 1200W PSU to replace my faithful 2-3 year old 850W Seasonic (which I will RMA on advice of Seasonic support so I can get a newer version which I will use on another system). The Thor is actually a Seasonic OEM (cool because this means it's also compatible with my customized Cablemod cables) but I'm betting the house that, having been manufactured more recently, it no longer suffers from the overcurrent protection issue with 3000 series GPUs. The higher wattage also makes me feel a lot more comfortable and that thing has RGB too!!!! (whoohoo lol).

I haven't installed it yet, so wish me luck please as this can be a very expensive mistake.