Monitor turning off in games - issue narrowed down to monitor, gpu, ram, mobo or psu!!

gupott · 7 May 2021 at 09:24

Hi there,
i have had an issue over the last month or so that I thought was related to my move from windows 10 to Linux mint whereby my monitor would switch itself off saying 'monitor going to sleep'- when playing games, largely football manager, prey, dark souls 3. I spent ages trying to fix, and then when using Win10 (i have a dual boot machine) it happened then also, so its not just Linux related.
My specs

Intel Core i7-9700k
Samsung 970 Pro 1TB NVMe
Gigabyte Z390 Aorus Ultra Intel Z390 DD4 (running f10h bios)
Gigabyte Aorus RGB 2x8BG DDR4 PC4-25600C16 3200MHz
Corsair RMx 750 80 plus Gold Modular PSU
Corsair Hyrdro H 115iPRO RGH Liquid cooler
HP2475w monitor
Palit Geforce Jetstream 970.

Since then, I have been exploring the idea that this is a hardware issue rather than software. I should I also say, this kit has been running well for the past two years, so nothing new introduced.

I have tried different cables (DVI, HDMI and displayport), and the issue happens with all. Other than this issue the monitor has been trouble free for a decade. I don't think its the monitor.

From a ram point of view, I have pulled it, and tried in a different dual combination. I have run windows memory diagnostic tool and also memtest and it always passes. I have tried the XMP settings on and off, manually adding, running lower speeds, timings etc. With the graphics card in, I always get the problem. So probably not memory related but I am not sure.

The graphics card is ancient and I have been trying to buy a 3080 for ages but not managed it for obvious reasons, hence playing the less demanding games. I don't have another graphics card to swap out for it unfortunately. It is not overclocked. I have pulled it, removed the stock cooler and re-applied thermal compound. The tempretures always look fine. Currently I am using the integrated graphics, and I have never had the issue when doing so. Perhaps the GPU is the most likely culprit? A visual inspection shows nothing obvious on the card.

Mobo - seems ok, no other errors. Not sure what to check. I tried running overclocked for a while without the graphics card to see what would happen. I ran prime95 for hours with no issue. This was the case with no overclock and overclocked. So I am thinking not motherboard related for now.

The PSU is also tricky to check, I get no other shut-downs, I have looked at the voltage in the bios and there are no great variations just minor ones I have observed in all psu's across the years. I am thinking its not PSU related for now.

I have been able to run football manager and lots of other things with the integrated graphics (with the GPU pulled out) and I have had no issue. I think a key point is, I have not observed a 'crash' when not using the GPU. The opposite view could be the machine is not being stressed enough I suppose.

So, I am a bit stuck, something is not happy. Can anyone suggest anything to narrow the issue down? I think I have gone a bit 'blind' by trying so many different things.

Thanks for reading, apologies for the long post! Thanks.

Joxeon · 7 May 2021 at 09:54

Have you checked the monitors OSD settings to make sure a sleep timing is not activated?

Temp.sgk · 7 May 2021 at 11:02

It seems to be your GPU considering you tested with integrated graphics. Some questions what's your gpu? Is it actually a crash? Can't you just turn your monitor back on?

gupott · 7 May 2021 at 12:21

Thanks for the replies.

Joxeon said:
Have you checked the monitors OSD settings to make sure a sleep timing is not activated?

Yes, I think so. I have turned off all references to screen savers in the OS and on the monitor. Plus all references to batteries and power saving that I can find.

Temp.sgk said:
It seems to be your GPU considering you tested with integrated graphics. Some questions what's your gpu? Is it actually a crash? Can't you just turn your monitor back on?

No, I can't turn the monitor back on. I have tried on/off, pulling cables and putting them back in. Changing inputs also.

When it happens, it usually just looks like you have powered the PC down. But occasionally it does look more like a crash, with a small visual disturbance as the monitor turns to black and gives the message. The PC is generally still going though, as you can hear the music.

And sorry, I realise I didnt actually put the GPU, I will go back and add. Its an Palit Geforce Jetstream 970.

mickyflinn · 7 May 2021 at 12:31

Did you try older drivers asuming thet were up.to date ?

Can you test your gpu in another system.

Temp.sgk · 7 May 2021 at 12:41

Try downclocking your GPU using MSI afterburner or some other tool and see if the crashes stop. Also there should be someway to more accurately identify the crash. In windows 10 you can type view reliability history and it lists what is crashing. I think your crash in windows 10 is a Kernel-power event ID 41.

gupott · 7 May 2021 at 12:47

mickyflinn said:
Did you try older drivers asuming thet were up.to date ?

Can you test your gpu in another system.

I had the latest drivers for nvidia for both Linux Mint and Windows. I would need to confirn for windows, but for Linux it was 460.56-0. I don't have another PC I can test the card in unfortunately. Just laptops etc.

Temp.sgk said:
Try downclocking your GPU using MSI afterburner or some other tool and see if the crashes stop. Also there should be someway to more accurately identify the crash. In windows 10 you can type view reliability history and it lists what is crashing. I think your crash in windows 10 is a Kernel-power event ID 41.

I will try the down clocking over the weekend. Are you thinking core and memory?

As for the crashes, I did look, will try again too. I tried a number of kernels in Linux, currently on

5.8.0-50

Temp.sgk · 7 May 2021 at 12:49

Yeah core and memory, start with core first. I don't know about Linux so I can't help you with that.

Quartz · 7 May 2021 at 12:58

rosscouk said:
I have been able to run football manager and lots of other things with the integrated graphics (with the GPU pulled out) and I have had no issue.

This does point to the GPU being the issue.

rosscouk said:
I had the latest drivers for nvidia for both Linux Mint and Windows.

Rather than the latest drivers, try going back to the Nvidia driver you used before it started crashing.

Temp.sgk said:
Try downclocking your GPU using MSI afterburner or some other tool and see if the crashes stop.

This is also good advice.

gupott · 7 May 2021 at 13:50

Quartz said:
Rather than the latest drivers, try going back to the Nvidia driver you used before it started crashing.
.

There are three Nvidia drivers and one open source driver on Linux Mint, I had the same issues with all the Nvidia drivers and with the open source one, I can't get games to run. I will have a look at windows.

gupott · 7 May 2021 at 13:51

Quartz said:
This does point to the GPU being the issue.

Yeah, however I came across a thread, on here actually, where someone had similar issues and was using the same ram as me, so I really wanted to rule it out as the source of my issue. It got me thinking that perhaps it wasnt the GPU but I could be conflating the two issues...

ninjag · 7 May 2021 at 16:21

Recently my monitor would keep switching off and then on randomly. Turned out to be my DP cable which was either faulty or wasn't a high enough spec for my 4k ultra wide settings. I found the original cable that came with the monitor and it fixed it.

Just a thought and I know you've mentioned trying different cables but it might be worth considering the cables again if the settings when testing with integrated graphics (less demand on the cable?) is lower than when using the graphics card.

Or, have you got a TV you could plug in to the graphics card using HDMI and thus eliminate the monitor itself as being the problem?

Quartz · 7 May 2021 at 16:57

He's running 1200p at 60 Hz so cable quality is unlikely to be an issue.

gupott · 8 May 2021 at 09:26

I have tried lots off different cables, maybe 5 now thinking about it. Right, time to try the underclocking of the card...

mickyflinn · 8 May 2021 at 09:54

Can you borrow a gpu from somone or test it in another pc its the only real way to check if its faulty.

gupott · 8 May 2021 at 13:51

mickyflinn said:
Can you borrow a gpu from somone or test it in another pc its the only real way to check if its faulty.

Yes, true. And no, no-one around me and family too far away.

I did a bit more testing. So, currently on Linux. I updated to the latest drivers, 465.27. I then played dark souls 3, the first time, I managed 25 mins before a crash, then 12, then 3 minutes. The PC is fairly cool, but could it be tempreture related on the GPU?? The fans are spinning. Anyway, I have now downloaded 'GreenWithEnvy' and have added two new profiles. The fan profile is now more strict, it comes on earlier in terms of temps and spins faster. I have also underclocked the GPU by -150Mhz (around 10%), and -350Mhz on the video memory, again 10%. I will try again in a bit and report back.

Tanin'iver · 8 May 2021 at 13:55

See if you can make the problem happen when running something other than a fullscreen game - Heaven or Valley benchmarks running in windowed mode would be fine, just make sure vsync off etc... so they are pushing your GPU to 100%. It certainly seems likely to be the GPU.

You mentioned you removed the stock cooler and re-applied thermal compound, but what about the other parts of the card?

I had serious stability issues with my old GTX 1080 Ti about 6 months ago, it wouldn't run games without crashing unless I set the power limit down to 80% (it wasn't overclocked). I eventually took it apart and replaced the thermal compound, but also the thermal pads for the RAM and VRMs. It has been fine back at 100% power limit since. I think it's likely the VRMs were overheating even though the GPU core temperature was fine, the pads had degraded over time and disintegrated when I came to remove them.

gupott · 8 May 2021 at 14:18

Tanin'iver said:
See if you can make the problem happen when running something other than a fullscreen game - Heaven or Valley benchmarks running in windowed mode would be fine, just make sure vsync off etc... so they are pushing your GPU to 100%. It certainly seems likely to be the GPU.

You mentioned you removed the stock cooler and re-applied thermal compound, but what about the other parts of the card?

I had serious stability issues with my old GTX 1080 Ti about 6 months ago, it wouldn't run games without crashing unless I set the power limit down to 80% (it wasn't overclocked). I eventually took it apart and replaced the thermal compound, but also the thermal pads for the RAM and VRMs. It has been fine back at 100% power limit since. I think it's likely the VRMs were overheating even though the GPU core temperature was fine, the pads had degraded over time and disintegrated when I came to remove them.

Yes, I took the stock cooler off, and put it back on with fresh compound. There does not seem to be any cooling on the RAM or VRMs, well, that I could see.

I have also reduced the power now, down to 162w, although I am not sure I am understanding it correctly on GWE. I am not sure if I have just reduced the power limit.

I will have a look at heaven and valley now.

Temp.sgk · 8 May 2021 at 22:22

Have the downclocking work? Maybe go even further in downclocking if it still crashes? Actually thinking about it, I think your GPU is going to die very soon, and you will smell smoke.

gupott · 9 May 2021 at 10:00

Morning!

rosscouk said:
Yes, I took the stock cooler off, and put it back on with fresh compound. There does not seem to be any cooling on the RAM or VRMs, well, that I could see.

I had another look, and there is a heatsink on what I assume is the VRMs. It had a heat pad accross the whole heatsink which I peeled off and reapplied with some I had laying around. I also fancied that one end of the heatsink was not tight down. Anyway, I also re-did the themal paste on the main processor too. I then played dark souls 3 for around 25 minutes. I thought it was all solved, then it crashed. So back to the drawing board. I actually had a way more agressive cooling profile and it was underclocked too, so annoying it crashed. This was in Linux.

Tanin'iver said:
See if you can make the problem happen when running something other than a fullscreen game - Heaven or Valley benchmarks running in windowed mode would be fine, just make sure vsync off etc... so they are pushing your GPU to 100%. It certainly seems likely to be the GPU.

I decided I would carry on testing in windows as I could use MSI afterburner, and I set about trying to get the machine to crash without running a full screen game, so I looped the Unigine Heaven benchmark and it dutifully crashed after a few minutes. I am running heaven on 'extreme' settings. I then downclocked the processor by 300 mhz and the memory by 500 mhz, and it will run looped heaven with these settings, running now for about 10 mins. Tempretures actually look fine, 77c tops under load.

I am not sure what to aim for now though. I have not increased the core voltage of the card, is it worth doing that?

Edit - update, I was having crash after crash after crash, I lowered the core by 500Mhz and the memory clock by 500Mhz also. This was not having much effect. Anyway, I added 10Mv and I have just played Dark souls 3 for 20 mins or so on windows. I need more time to check, but perhaps adding more volts has increased stability?? No idea if I can do this in Linux, but I can with Afterburner in windows. More testing...