Question Instability, GPU output cutting off.

AntaresSQ01

Reputable
Oct 10, 2016
81
1
4,645
1
Hi,
I have some issues that I believe to be PSU related. I know my PSU is "low tier" but it hasn't had any issues. I think this is more likely due to the fact that my system is pushing it's designed wattage limits. As the parts come in at ~500W under full load.
The actual issue is that whilst gaming, Occasionally the screen goes to 1 solid colour, usually pale brown or dark blue, sometimes black (but it really can be any colour) for a couple seconds and then recovers and continues to work as it should. However this loss of output has some adverse effects on different games, League of Legends recovers and you can continue playing, Escape from Tarkov, you can still hear and move around but your desktop is shown (see through game window) and Warframe just straight up crashes (which is super annoying when you lose all progress in a mission). Another funny effect is that if it's late at night when I have window's "Night light" on it turns off after recovery.

I'll try to get footage of it but it's incredibly hard as it's pretty random.

However some games do it more often than others. Which is weird as i suspected this to be an issue due to the PSU being pushed to the limit in power, failing to provide power for the GPU when it suddenly requests more power and then recovery, however this happens about 1-2x a 30 minute game in League of Legends (which literally runs on Celeron laptops at 60 fps), Escape from Tarkov gets it about 1x every 1-2 hours of being in the game (it has surprisingly high graphics and CPU usage even in the menu as i assume the background is 3d processed), this is also by far the most taxing game out of the 3, as for Warframe which again, runs at max settings on a GTX 650 at 60 fps almost, however this game has relatively low CPU usage and yet this is the game which causes this recovery the most often, pretty much guaranteed 1x every 10 minutes of gameplay and as this is the only game that crashes to desktop, it's near unplayable.

The thing is my system changed remarkably little over the past couple years and I didn't use to have this issue. As another note this issue sometimes even happens whilst just browsing in Chrome, and there are games when it virtually never happens like Forza Horizon 3 I could play for 5hrs straight and not have it happen once, which is also a relatively intensive game. I have a lot of convenience goodies in my PC that could up the wattage such as a Sound card, a total of 4 drives, a wifi card, a couple gaming peripherals, a USB bluetooth receiver and a couple fans. That said, I've tried gaming with my PSU power limit set to 60% (still provides over 60 fps in warframe) but to no avail as it still gets the issue. Though I'm not 100% sure it actually worked, but EVGA's Precision XOC did say my GPU power usage was only 60% (usually it ramps up to 105-110% even on default). I've tried OC-ing, UC-ing but it doesn't matter if i'm running at 1700 MHz on the card and 5.0 GHz on the CPU or 1000MHz on the card and 3.5GHz on the cpu. Still happens.

Either way, I've tried to do a lot of things to narrow down my search, and I do suspect it's the PSU which i'm intending to replace soon but I want to make sure, this is the 3rd time i'm posting related to this issue as the last times I got no response apart from a mod requesting a correction o_O

Any help is much appreciated!

Specs:
CPU: i7 7700K @ 4.20 GHZ
Motherboard: MSI Z270-A Pro
Ram: 2x8GB Crucial Ballistix Sport LT DDR4 @2666MHZ
SSD/HDD: Samsung 850 EVO 250GB SATA
Samsung 960 EVO 500GB SATA
Samsung HD204UI 2TB
GPU: EVGA GTX 980 SC ACX 2.0
PSU: Corsair CX600M
Chassis: Corsair Graphite Series 230T
OS: Windows 10 Pro x64 Build 1903
 

Starcruiser

Honorable
May 18, 2013
897
28
11,390
213
That's a bit more info than I needed, which is probably part of why you didn't get a response before.

The CX-M series isn't a top tier but it's certainly not bottom tier either, more like in the middle. It should be fine barring any defects.
I would say it seems more like a video card or monitor issue than the PSU. The monitor itself either stopped detecting an input and decided to sleep until input was detected again or it reset itself. I'll assume you've done the obvious and updated your graphics drivers. Are you able to test another input port on the monitor? If not at least make sure nothing is loose, and unplug and clean connectors.

The best way to tell if it's a GPU issue is to try another GPU. Maybe borrow one from a friend? If you can't do that get a monitoring tool that will record GPU usages and system voltages. If GPU use drops suddenly when the screen cuts, it's probably the GPU. If there's a spike or other fluctuation in voltage, it's more likely the motherboard or power supply.
 

AntaresSQ01

Reputable
Oct 10, 2016
81
1
4,645
1
That's a bit more info than I needed, which is probably part of why you didn't get a response before.

The CX-M series isn't a top tier but it's certainly not bottom tier either, more like in the middle. It should be fine barring any defects.
I would say it seems more like a video card or monitor issue than the PSU. The monitor itself either stopped detecting an input and decided to sleep until input was detected again or it reset itself. I'll assume you've done the obvious and updated your graphics drivers. Are you able to test another input port on the monitor? If not at least make sure nothing is loose, and unplug and clean connectors.

The best way to tell if it's a GPU issue is to try another GPU. Maybe borrow one from a friend? If you can't do that get a monitoring tool that will record GPU usages and system voltages. If GPU use drops suddenly when the screen cuts, it's probably the GPU. If there's a spike or other fluctuation in voltage, it's more likely the motherboard or power supply.
I wrote my question differently on purpose for the very reason that i would get a response.
Either way, yes I've done a DDU clean driver install, I'm monitored GPU usage but never across a recovery. I'll try the other port on my monitor but sadly this is the only GPU I have on hand and it only has 1 HDMI output. My motherboard only has VGA, DVI and Display port too. When I wiggle the HDMI cable it sometimes does something similar as the recovery however it's not quite the same and it would make no sense for that to be the problem as the cable normally doesn't move at all and it requires a lot of wiggling for it to do that,

Additional information that might be leading somewhere though: When it does the recovery, the lights on my GPU flash momentarily as the screen recovers, I'll check the GPU usage but this could still be a PSU related issue. The PSU is about 5 years old and been through a couple systems. It might be a GPU issue but it seems too random yet too specific. As i said, straight up load doesn't seem to matter and it can happen at any time, though some games seem to entice it more than others. I'll be back with my findings
 

Starcruiser

Honorable
May 18, 2013
897
28
11,390
213
Since it's an older PSU, you have to account for it losing up to 15-25% of it's capacity after 3 years. After 5 years it's more like 20-30%. It would be time to replace it anyway, especially if it's been used daily.

The lights on the GPU flashing could indicate that it's resetting itself, though it doesn't help pinpoint the cause. It does help rule out the monitor though.
 

AntaresSQ01

Reputable
Oct 10, 2016
81
1
4,645
1
Since it's an older PSU, you have to account for it losing up to 15-25% of it's capacity after 3 years. After 5 years it's more like 20-30%. It would be time to replace it anyway, especially if it's been used daily.

The lights on the GPU flashing could indicate that it's resetting itself, though it doesn't help pinpoint the cause. It does help rule out the monitor though.
Hmm I've never heard about PSUs losing power over time, but my system should be pushing close to it's theoretical limit even with 80+

Here are some graphs shortly after the crash, though i'm not sure if it helps since this game crashes to the desktop so GPU usage obviously falls once you're out of the game.

 

Starcruiser

Honorable
May 18, 2013
897
28
11,390
213
The power loss over time is called capacitor aging and it's well documented here and elsewhere. If you use it 5-7 days a week, 8 hours or so on average, you'd be looking at the numbers I gave.

The voltages from the power supply look stable and in spec. The only one that drops is the GPU. It's interesting that the gpu would lower its clock THEN spike and crash to nothing before returning to normal. That shouldn't be happening. I'm calling this issue as the GPU.
 

AntaresSQ01

Reputable
Oct 10, 2016
81
1
4,645
1
The power loss over time is called capacitor aging and it's well documented here and elsewhere. If you use it 5-7 days a week, 8 hours or so on average, you'd be looking at the numbers I gave.

The voltages from the power supply look stable and in spec. The only one that drops is the GPU. It's interesting that the gpu would lower its clock THEN spike and crash to nothing before returning to normal. That shouldn't be happening. I'm calling this issue as the GPU.
The spike happened as it opened the chrome tab which does actually spike gpu usage. That graph is over a very long time, from the 1st dip to the end is probably 30 seconds + you can see the GPU stopped being utilised well before the clock dip. I think the spike after the initial dip is irrelevant to the issue but I'll keep an eye on it.
 

ASK THE COMMUNITY