Computer shuts down after about 30 minutes of DCS

[sepruda]

Hello, 

I recently upgraded my system (only to improve my DCS experience!), and unfortunately now my PC completely shuts down (without rebotting) after about 30 minutes of time flying. It doesn't matter what I do, just having the plane sit cold and dark for 30 minutes will lead to the same result. 

I first thought it was a heating issue, my CPU was peaking at about 90 celsius (even though it should be able to take that). So I installed an additional fan for the CPU and put in extra cabinet fans. That didn't change anything though. 

Then I tried reinstalling DCS, I tried going back a version, I've updated my Bios and GPU drivers. This only happens for me when playing DCS. 

I've read similar threads, and tried to follow the suggestions there (like this one) - the conclusions of both of those were the PSU that was faulty. 

That of course might be the case for me as well, but I upgraded my PSU in august (80+ gold certified), and it was working fine before I upgraded my CPU. 

Any help would be much appreciated!

My specs: 

TUF GAMING B450M-PLUS II MB
AMD Ryzen 9 5950X
Noctua NH-D15 CPU cooler
4 x Corsair Vengeance 8GB DDR4-3200 RAM
Asus GeForce® RTX 3080 10GB TUF
Corsair RM750 750W 80+ Gold PSU
Intel 665P 1TB M.2 NVMe SSD
Corsair Carbide 275R Tempered Glass Black VG Edition Cabinet
Windows 11

Edited November 14, 2022 by sepruda

[Steel Jaw]

Number one cause of that is going to be either CPU or GPU over heating. 

[Hiob]

Shutting down completely without rebooting could indeed point to your PSU. If it is a multi-rail PSU, make sure that you haven't connected the GPU to one rail only (consult the manual).

If it is the PSU, you should be able to force a black out with other GPU heavy applications as well (MSI Kombustor e.g.)

If you use MSI Afterburner you could try to limit the Power draw of your GPU to - let's say - 200-250W (for testing), and see if it runs stable then.

Edit: Does your CPU reach 90C while running DCS only? That's definitely too much. 90C while running cinebench (multicore), ok. But 90C while running on two cores with DCS - probably not. Could point to a bad application of thermal paste then (or a sub par cooler)

Edited November 14, 2022 by Hiob

[Steel Jaw]

Oh PSU yeah thats good too.

[sepruda]

@Steel Jaw

(LOL just choose a Virgil Cole photo, and then you have a quote from him) 

I'm pretty sure it's not overheating, since it's not hitting above 80 with my new fans. Also, I've played taxing VR titles like Alyx without issue. 

@Hiob It might be that I plugged something in the wrong way, I'm very new to PC building. I have tried running a stress test (can't recall the program now), without issue, and have played other titles without issue. But I'll crack my case open and see if I somehow connected something "one rail".

[Hiob]

9 minutes ago, sepruda said:

@Hiob It might be that I plugged something in the wrong way, I'm very new to PC building. I have tried running a stress test (can't recall the program now), without issue, and have played other titles without issue. But I'll crack my case open and see if I somehow connected something "one rail".

You really need to study the manual of your PSU. Usually they give only 150W (20A*12V is more like 240W) on a single connector. So if you use a Y-Cable to power the two connectors of your GPU with one connector on your PSU, you're probably asking too much. There should be at least two PCIe outlets that are meant for GPUs. You should use both to power a 3080.

I can't find the maunal to your PSU online. But it should contain a table like this. (this is from my bequiet)

Note that a single connector can only support 20-24 Amp. That is not enough for a 3080 under all circumstances.

Edited November 14, 2022 by Hiob

[sepruda]

Thank you for your answer @Hiob!

I'm cracking it open now, found the power table for my PSU. 

 

[BIGNEWY]

have you checked windows event viewer to look for the reason?

[Hiob]

18 minutes ago, sepruda said:

Thank you for your answer @Hiob!

I'm cracking it open now, found the power table for my PSU. 

 

That's not exactly the chart I'm looking for. It only shows the combined Power Output of all connectors. Nevermind. Is your GPU connected to at least two of the 6+2 PCIe Connectors on the PSU side?

[sepruda]

Ive connected both to these, is that correct? Sorry, I'm a real beginner at all of this. 

 

[Hiob]

That should be fine. But I wonder if there is a table in PSUs manual that shows, which socket is connected to which power-rail. Unfortunatly I have no experience with corsair PSUs and they don't offer the manual for download (or I didn't find it).

Edit: Ok I found a manual and it seems that the corsairs really are build(or documented) a little different. The way you connected your GPU should be fine.

Do the other stress tests I recommended, but your PSU should be ok.

Check the event viewer for whea logs as @BIGNEWY suggested

Edited November 14, 2022 by Hiob

[sepruda]

To be clear, my GPU was not connected to those two sockets, I've moved them there (currently running the game for testing again). 

Will look into the event viewer and stress test if the issue persists. 

[sepruda]

I was beginning to hope that switching PSU sockets had solved it, but unfortunately I gotta another shutdown after 53 minutes cold and dark on the tarmac. 

I've looked in the event log, but not sure what to look for. I have one critical "kernel-power" error, just around the shutdown, but doesn't really tell me anything. I've attached the logs around that error. 

I've only just noticed you writing about the temperature in post 3 @Hiob I've applied the thermal paste and installed the CPU cooler myself, so might be a possibility that I did something wrong. Is there a way to track what the highest temperature was before shutdown? I had the cabinet open for my last round of testing, which lasted longer than before, so I guess a cooling problem is still a possibility..

I will try to run the stress tests you recommended.

log.evtx

Edited November 14, 2022 by sepruda

[Hiob]

5 minutes ago, sepruda said:

I've only just noticed you writing about the temperature in post 3 @Hiob I've applied the thermal paste and installed the CPU cooler myself, so might be a possibility that I did something wrong. Is there a way to track what the highest temperature was before shutdown? I had the cabinet open for my last round of testing, which lasted longer than before, so I guess a cooling problem is still a possibility..

I will try to run the stress tests you recommended.

log.ev 68 kB · 0 download

 

Can you monitor the CPU temps whilst playing DCS? MSI Afterburner would be a good tool for that. To be clear: 90C isn't out of spec for a Ryzen, but it shouldn't hit such high temps in DCS (well, perhaps momentarily while loading but not for a prolonged time) - since DCS utilizes only a few cores (only one heavily) the power consumption and therefore heat of the CPU shoould not be in the extremes.

If you want to provoke a heat-issue (don't worry the CPU protects itself by shutting down in the worst case) you could run CinebenchR23 (free download) on all cores on a loop. With HWinfo (free tool) you can monitor the temps at the same time because both apps run in a window. If your CPU hits 90+ C on cinebench, that's perfectly normal depending on the cooling. But it shouldn't shut down. However if there is a serious cooling issue (no thermal paste as an extreme example), the sudden heat build up could lead to a safety shutdown.

[sepruda]

I used MSI Kombustor to run a stress test, and after I came back to the PC in about an hours time, it was shut off again. 

So does indeed seem to be a cooling issue of either the CPU or GPU. 

I have monitored the temperatures before while playing DCS, however not just before the shutdown and also using a different tool. 

I guess there are different paths for me now: Try to provoke the heat issue like you described, or check the installation of the CPU cooler and the thermal paste. 

I bought a Noctua NH-D15 CPU cooler, which was almost to big for my MB, it's pushing slightly towards the graphics card, don't know if that could be an issue.

[Hiob]

The Noctua is certainly a top tier cooler - provided it is installed properly. It shouldn't be under load/stress once mounted.

If I spot it correctly, you have two additional fans on the top of your case. They probably do more harm than good. Put as many as you can fit to the front of your case as intakes and your back fan as outlet. The airstream of your GPU fans should point to the outlet.

You want positive pressure in your case (more intake fans than outlet) and an airflow that is as little disturbed as possible. Having two airstreams perpendicular to each other can be counter productive.

 

Edit:

If the CPU has heat issues with an open side panel, I would bet on a poor contact between cooling plate and CPU-heat spreader. (Or you have forgotten to remove the protective plastic piece from the cooler)

 

Edit2:

With an open side panel, the GPU should be safe from a heat stroke (also GPUs problems usually cause CTD and not a shut down).

Edited November 14, 2022 by Hiob

[sepruda]

Correct, I installed two additional fans on top, though I only did that after the shutdowns  

I have three fans in front (intake), two in top and one in the back (outtake). I did have to disable one of the top fans though since it was hitting something on the Noctua  

I did remove the plastic piece on the cooler, but I will try to remove the cooler and install it again, making sure it contacts properly with the CPU. I think I'll also follow your advice and remove the top fans. Though first I will try to monitor the CPU heating in DCS again, wish there was some software that recorded the highest temperature since the time might vary quite a bit.

[Hiob]

8 minutes ago, sepruda said:

Correct, I installed two additional fans on top, though I only did that after the shutdowns  

I have three fans in front (intake), two in top and one in the back (outtake). I did have to disable one of the top fans though since it was hitting something on the Noctua  

I did remove the plastic piece on the cooler, but I will try to remove the cooler and install it again, making sure it contacts properly with the CPU. I think I'll also follow your advice and remove the top fans. Though first I will try to monitor the CPU heating in DCS again, wish there was some software that recorded the highest temperature since the time might vary quite a bit.

There is software to record temps over time. HWinfo e.g. or OCCT. It's all free software. The easiest way to examine the CPU heat would be running cinebench (also free) and hwinfo.

[sepruda]

So I ran Cinebench multi core test and hwinfo and CPU temp peaked at 63. 

Afterwards ran the MSI Kombuster stress test untill shutdown, latest CPU reading from hwiinfo says 64, max GPU is 63. 

Doesn't seem like a heating issue  

Of course it's possible that hwiinfo didn't record a spike just before shutdown..

 

Edited November 14, 2022 by sepruda

[Hiob]

3 hours ago, sepruda said:

So I ran Cinebench multi core test and hwinfo and CPU temp peaked at 63. 

Afterwards ran the MSI Kombuster stress test untill shutdown, latest CPU reading from hwiinfo says 64, max GPU is 63. 

Doesn't seem like a heating issue  

Of course it's possible that hwiinfo didn't record a spike just before shutdown..

 

 

Neither one of those temps is even close to being critical.

Please Download a tool named GPU-Z. Looks like this

 

There you can see all possible values of your GPU. Powerdraw, temps... and you can log them to a file. Perhaps that can give you some insights....

Edited November 14, 2022 by Hiob

[Mars Exulte]

I concur. Doesn't sound like general overheat. You've probably got some sort of hardware failure happening, IMO. I'd start trying to isolate hardware items to determine what specifically is messing up either with targeted stress tests, hardware swaps, etc.

Edited November 15, 2022 by Mars Exulte

[LucShep]

16 hours ago, sepruda said:

Hello, 

I recently upgraded my system (only to improve my DCS experience!), and unfortunately now my PC completely shuts down (without rebotting) after about 30 minutes of time flying. It doesn't matter what I do, just having the plane sit cold and dark for 30 minutes will lead to the same result. 

I first thought it was a heating issue, my CPU was peaking at about 90 celsius (even though it should be able to take that). So I installed an additional fan for the CPU and put in extra cabinet fans. That didn't change anything though. 

Then I tried reinstalling DCS, I tried going back a version, I've updated my Bios and GPU drivers. This only happens for me when playing DCS. 

I've read similar threads, and tried to follow the suggestions there (like this one) - the conclusions of both of those were the PSU that was faulty. 

That of course might be the case for me as well, but I upgraded my PSU in august (80+ gold certified), and it was working fine before I upgraded my CPU. 

Any help would be much appreciated!

My specs: 

TUF GAMING B450M-PLUS II MB
AMD Ryzen 9 5950X
Noctua NH-D15 CPU cooler
4 x Corsair Vengeance 8GB DDR4-3200 RAM
Asus GeForce® RTX 3080 10GB TUF
Corsair RM750 750W 80+ Gold PSU
Intel 665P 1TB M.2 NVMe SSD
Corsair Carbide 275R Tempered Glass Black VG Edition Cabinet
Windows 11

The way I see it, there are three redflags just looking at your systems specs.
I could be wrong but, reading your description and looking at those specs, it makes all the sense to me that at least one of these, if not a combination, is causing your problem:
 

1. B450 motherboard with Ryzen 5950X 
   The ASUS TUF GAMING B450M-PLUS II is, from memory an 8x46A VRM, which means it's a really bad combination with that Ryzen 5950X.
   For your to have an idea, it seems even the older Ryzen 3900X was making the VRM temps pretty darn toasty on that motherboard, so not recommended... and that's a CPU that seems to pull similar power as yours does (identical TDP)....
   That motherboard is simply not prepared nor meant to run that Ryzen 5950X. At some point that VRM hits really high temps that system is not prepared to endure, causing BSoDs and shutdowns (until sometime later it eventually "Kaputs").
    
2. Corsair RM750 750W 80+ Gold PSU
   While in theory you should be able to run that hardware with that PSU (which is a good one, make no mistake), you have a power hungry CPU, and if added with (ocasional) transient spikes from that also power hungry RTX3080 (which I'm sure will be running quite hard with DCS as it is these days), then you'll run at the PSU limits, even if just for brief moments, and probably trigger the OCP on that 750W PSU - causing CTDs and shutdowns (until sometime later it eventually "Kaputs")..
   That CPU+GPU combination should be running on a good 850W+ PSU, not less. 
    
3. 4 x Corsair Vengeance 8GB DDR4-3200 RAM
   It should run fine but, sometimes, stability is not there without some small tweaking.
   Running four sticks of RAM can require a 0.05v increase (if it's set at 1.35v, increasing to 1.40v then) in the DRAM VOLTAGE, in the BIOS respective setting.
   Either doing that, or decreasing the speed of the memory to the imediate degree below (for example, yours are 3200, turning down the speed to 3000) will ensure far better stability with the XMP/DOCP based settings of that RAM kit.
    

With all this said... easiest solution, in my opinion, is to sell that Ryzen 5950X and get a Ryzen 5600X.
"Whaaaat??"  Yes!!
Swapping for a Ryzen 5600X you will likely solve two of your possible problems in one go, the 1) and 2) points that I mentioned above (motherboard VRM issues and PSU limit), instead of spending ($$$) on a better motherboard and stronger PSU in an atempt to solve issues.

Your motherboard will run the Ryzen 5600X beautifully. At same time it'll also aliviate a bit that 750W PSU, giving it more margin.
And no, it's not really a downgrade as it's a great gaming processor - quite frankly, I think it's same identical performance in any and all games (also in DCS, I suspect).
If you decide so... heck, maybe consider opening an offer to fellow DCS'ers, some kind of deal where you can be given a fair sum of money AND a Ryzen 5600X, for your 5950X?

Regarding the point 3) that I mentioned above (RAM instability), it could or not be related, but you should try it anyway, regardless of CPU in use.
 

Edited November 15, 2022 by LucShep

[sepruda]

Guys, thank you for taking the time to try to help me, it's really frustrating not having my dose of DCS  

@LucShepThank you for your thorough analysis. I know I'm pushing my MB with that processor, but I'd rather upgrade my MB than downgrade my CPU  Maybe it's possible to see the VRM temps in one of those programs I am running (?). The 750w should be enough, my dealer assures me, but I'm also skeptical that that is really the case. I will try Hiobs suggestion  "MSI Afterburner you could try to limit the Power draw of your GPU to - let's say - 200-250W", and see if that solves it.". 

Next step will be turning down the speed of the RAM as you suggest. 

I will try to run GPU-Z as you suggest Hiob. 

I don't mind buying a 850w PSU (even though I just upgraded to the 750w  ) - but would like some kind of verification that power is really the issue. 

 

[LucShep]

3 hours ago, sepruda said:

Guys, thank you for taking the time to try to help me, it's really frustrating not having my dose of DCS  

@LucShepThank you for your thorough analysis. I know I'm pushing my MB with that processor, but I'd rather upgrade my MB than downgrade my CPU  Maybe it's possible to see the VRM temps in one of those programs I am running (?). The 750w should be enough, my dealer assures me, but I'm also skeptical that that is really the case. I will try Hiobs suggestion  "MSI Afterburner you could try to limit the Power draw of your GPU to - let's say - 200-250W", and see if that solves it.". 

Next step will be turning down the speed of the RAM as you suggest. 

I will try to run GPU-Z as you suggest Hiob. 

I don't mind buying a 850w PSU (even though I just upgraded to the 750w  ) - but would like some kind of verification that power is really the issue. 

 

Ok, understood.
FWIW, I completely disagree with your dealer there.

If you're buying a higher wattage PSU, definitely get 1000W instead of 850W, prices now got too close between them to not get the bigger wattage one.

There are some good quality and very reasonably priced 1000W 80+Gold PSUs worth looking for. 
For example, and among others, check the Corsair RMX 1000w, EVGA SuperNOVA GT (or G6) 1000w, Seasonic Prime (Gold) GX1000 and SuperFlower Leadex (Gold) 1000w.

For the motherboard, there are good mid range B550 and X570 motherboards with which that Ryzen 9 5950X will work great, and not outrageously expensive.
I'm a sucker for MSI, so recommend the MSI B550 Tomahawk and MSI X570 Tomahawk, but the Gigabyte B550 Aorus Pro and Gigabyte X570 Aorus Pro are also good choices.  
 

Edited November 15, 2022 by LucShep

[Hiob]

1 hour ago, sepruda said:

Guys, thank you for taking the time to try to help me, it's really frustrating not having my dose of DCS  

@LucShepThank you for your thorough analysis. I know I'm pushing my MB with that processor, but I'd rather upgrade my MB than downgrade my CPU  Maybe it's possible to see the VRM temps in one of those programs I am running (?). The 750w should be enough, my dealer assures me, but I'm also skeptical that that is really the case. I will try Hiobs suggestion  "MSI Afterburner you could try to limit the Power draw of your GPU to - let's say - 200-250W", and see if that solves it.". 

Next step will be turning down the speed of the RAM as you suggest. 

I will try to run GPU-Z as you suggest Hiob. 

I don't mind buying a 850w PSU (even though I just upgraded to the 750w  ) - but would like some kind of verification that power is really the issue. 

 

You can monitor the VRM temps with HWinfo e.g. (and any other hardware monitoring software), provided that the Mainboard has a sensor for that. @LucShep is right. The power stages on your MB aren't the beefiest ones, but that shouldn't be an issue on a low load scenario like DCS in my opinion. You should however make sure, that you have the latest firmware for your MB. Also, you could reset your BIOS to the failsafe values, which will also deactivate the XMP on your memory. You won't get the highest performance, but it will run stable.
Raising the Voltage slightly as suggested should also increase stability.

You could also deactivate one CCD in the bios (depending on the bios), which will make your 5950X an 5800X effectively.