GPU Usage Drops From 99% To 0%?

TCobra

Commendable
Dec 21, 2016
101
0
1,710
Hello there!
I'm having a really annoying issue with my graphics card. Randomly when I'm playing games (Witcher 3, Rainbow Six Seige, etc), my GPU usage drops to 0% for a couple of seconds, and then jumps back to 99% after a few seconds. This causes the game to freeze at 0 fps for a few seconds, until the GPU resets. I've heard that this can be caused by the GPU not receiving enough power. Is my PSU good enough to handle my specs? If so, what else can cause this issue?

Here's info on my PC:

SPECS:
NVIDIA Geforce GTX 1080 (MSI Gaming X version)

Intel Core i7-6700K Skylake at 4.2GHz

ASUS Z170 PRO GAMING, Socket-1151 (Motherboard)

HyperX Fury DDR4 2400MHz 16GB (Memory)

EVGA Gold Rated 650W Power Supply

For storage I've got a 250GB SSD for windows and programs, 1TB Seagate HDD, and a 2TB WD Black HDD for games.

TEMPS
My GPU is idling at 59-60c (This is because the fans are not spinning while idle, so they are quite high.)
My GPU hits a maximum of 75c when under full load, but is normally at 71-74c.

My CPU is idling at 19-25c, and under load it's between 45-65c.

OVERCLOCK:
My card is overclocked, but I have put the overclock quite far down because I thought this was part of the issue. I guess at some point ill have to test it without the overclock on.

The card is quite new, I've had it for maybe 2-3 months at most.

NEED HELP!
Thanks!

I saw a post about someone with the same issue as me. Turns out he had a broken ram stick. I tried running the ram diagnostic, but it says my RAM is fine. Any other suggestions are greatly appreciated!

I've tried to turn XMP profile on for my RAM, and nothing changed. I've also tried to set the power management settings to "Prefer maximum power" in the nvidia control panel, and the problem persists. It occurs when i have been playing for about an hour.
 


Yes that should be the starting point.

\My son's MSI Gaming X ( 2 days old) started out with these settings:

http://www.guru3d.com/articles_pages/msi_geforce_gtx_1070_gaming_x_review,30.html


Core Voltage : +100%
Power Limit : 120 %
Temp. limit : 92C
Core Clock : + 105
Memory Clock : +700 MHz (=9400 MHz effective data-rate)

And we got the familiar "stopped and restarted" message. I backed off memory OC by 50 and hasn't repeated ... will inch up a bit once he stops playing for 5 minutes :)


 

TCobra

Commendable
Dec 21, 2016
101
0
1,710


Hey!
I will try to test it without any overclock on.

I found that my card was pretty stable at +139 added core clock, and +850 added memory clock. But since heaven benchmark is quite OC friendly, I've put that down to +119 of added core clock and +750 added memory clock.

What is the "stopped and restarted" message you're talking about though? I dont get any message, it just drops my GPU usage to 0% for a few seconds, and then goes back to normal.

Ill post back results after playing for a bit without overclocking :)

 

TCobra

Commendable
Dec 21, 2016
101
0
1,710


Ah i see!
I've now played for about an hour, with my overclock put down to +100 on the core clock and +500 on the memory clock - Hasnt happened yet, so hopefully it wont. But ill test different games and post back
 
You can pass the flying donut (Furmark), all the Unigine and 3D mark tests just fine ... and then:

Metro 2033 has some spots that may knock ya down a bit... and then fine for rest of game.
If I get a call on a build that the OC is unstable, 96% of the time it's a Battlefield title.... so prolly shud start there :)
 

TCobra

Commendable
Dec 21, 2016
101
0
1,710


Still happened when putting the OC that low. (Was playing witcher 3). Now I\ve reset it to stock, no OC. Will be reporting back if it happens again. If it does then atleast we know its not the overclock
 

amtseung

Distinguished
Remember that GPUBoost 3.0 will do its own "overclocking". Changing the offset just offsets the entire curve.

I would try disabling all the power savings things in bios (C states, EIST, forget what else for the CPU, ASPM for the GPU) and use the Nvidia Control Panel to make the card favor highest performance for single monitor (it's under the "prefer maximum power" setting), and see if it's still wildly dipping like it was before.

Are there any differences in GPU behavior between a synthetic load like Unigine's benchmarks and actual games? Usually, a GPU underperforming on this scale is indicative of something else in the system being not quite right. Memory leak? Stray useless windows services hogging RAM and CPU runtime? Large quantity of RAM being listed as "hardware reserved"? Has it been a while since you've trimmed your SSD's and/or defragged your HDD's? Have you tried maxing out the voltage, temp target, and power limit sliders while leaving the other sliders alone?

60C idle is kind of high. How's your case airflow? Are you running multiple monitors?
 

TCobra

Commendable
Dec 21, 2016
101
0
1,710


Hey! I'm currently testing games without any OC on the card or the CPU, both are reset to stock. But i thought i'd just answer your questions:

GPUBoost 3.0?
I don't have GPUBoost 3.0, I'm just using afterburner to OC.

POWER SAVING SETTINGS?
I've disabled all the power saving settings I know of, Windows is set to "performance" instead of balanced in power settings, and in the Nvidia control panel I set the graphics card to "Prefer maximum power". I don't think my motherboard has any power saving settings by default since its meant for gaming/high performance, but ill look into the settings you listed :)

DIFFERENCE IN GPU BEHAVIOR?
I don't see any difference in behavior when playing and benchmarking. When benchmarking its using 99% of the GPU, and when gaming it uses 99% too. The only difference is that when I've been gaming for about an hour, the GPU usages drops to 0% and then back up to 99% after a few seconds.

MEMORY PROBLEMS?
I've done the memory diagnostic in windows, and it says my memory is fine, so don't think it has anything to do with memory. I also enabled an XMP profile for my memory, as someone with a similar problem got it solved that way. It didn't help me though.

TASKS RUNNING WHILE GAMING?
When gaming I don't have any CPU or RAM heavy tasks running. Usually its just a google chrome windows, and Discord (An app almost like TeamSpeak).

HARDDRIVES?
I just recently defragged my hard drive, but i also got a new hard drive for Christmas, one that was significantly better than the other one. Most of the games I'm testing is on the brand new hard drive, so I'm fairly certain the HDD is not the issue.

MAX POWER SLIDER WITHOUT TOUCHING OTHER SLIDERS?
I will try to max the power slider and temp target without touching the core clock and memory clock, and see if that helps! Thanks for the suggestion.

GPU TEMP?
It's true that the GPU runs hot while idle, but as I stated its because the fans doesn't spin while its idle

CASE AIRFLOW?
Case airflow is possible the best you can get. I've got the Corsair Carbide Air 740, which is made for good airflow. I've got two front fans running at 950 rpm each, two top fans running at 500 rpm each, and one back fan running at 500rpm. So airflow shouldn't be a problem :)

UPDATE: Forgot to mention: I am indeed running 2 monitors. They are 2 BenQ monitors 27", and they are both running 144hz. One of them is a newer model while the other is a bit older.
 

TCobra

Commendable
Dec 21, 2016
101
0
1,710


Oh! Thought it was a program. Should I do anything with GPUBoost 3.0? And what does "flashing the GPU" mean? Sorry if thats a noob question
 


Flashing is not something you do unless absolutely necessary. If you mess up a flash, the card is bricked and can't be recovered easily. Just know that Nvidia GPU's basically automatically OC.
 

TCobra

Commendable
Dec 21, 2016
101
0
1,710


Alright. I've been playing with the card at stock speeds for quite a while now, and it hasnt happened. Ill test further before i call this solved though. Thanks for suggesting though!

 

TCobra

Commendable
Dec 21, 2016
101
0
1,710
Ok, Update:
It still happens, even though i've completely reset my CPU and GPU to stock - no OC.
Further suggestions are greatly appreciated, really wanna find a fix for this!
 

amtseung

Distinguished
GPUBoost 3.0 cannot be disabled without, as stated above, nearly bricking the graphics card. It's there when you want it, it's there when you don't. It cannot be turned off.
Do the same dips (after an hour) happen when you're running a synthetic benchmark? Try run Unigine heaven or valley, and just leave it on for a long time, and see if that too, will have its dips.
XMP usually creates more problems than it fixes. It has a nasty habit of screwing with bios settings without telling you, which in some cases, create more instability.
Google Chrome is both CPU and RAM heavy. Actually, every web browser is. Especially Chrome.
That explains why your idle temps are high. It has nothing to do with the fan being off. The fan running or not is based on temps, not load.

 

TCobra

Commendable
Dec 21, 2016
101
0
1,710


Hi, thanks for the answer!
I'll try to run a benchmark for an hour and look at the graph.
In regards to google chrome, when i have that open it only uses 0.1% of my cpu, and 100MB of ram. I really doubt that is gonna affect my gaming performance, but ill try and close when i run the benchmark :)
 

TCobra

Commendable
Dec 21, 2016
101
0
1,710


I have now been running Heaven Benchmark for about an hour and 10 minutes. I monitored the GPU trough MSI Afterburner and here are the results:
https://gyazo.com/51833f101e1aed94919da6f243bba3d8

The GPU usage is the bottom graph. As you can see there are dips, but those are from whenever the benchmark is changing scene. It never dropped to 0% like it does in games. :/
 

TCobra

Commendable
Dec 21, 2016
101
0
1,710
Update:
I've heard from another guy having the exact same issue. He had Advanced SystemCare installed, and theres a built in function that cleans out your memory when it reaches 50-60%. Thats what caused his stutters and the GPU going to 0%.
I have ASC too, and i've just turned this function off too, so hopefully thats what was causing it. Will update later when i know if its fixed or not.
 

TCobra

Commendable
Dec 21, 2016
101
0
1,710
This issue was resolved by turning off the "Auto Clean Ram" feature in Advanced System Care. So for anyone reading along, with the same problem: If you have advanced system care, Press the expand icon in the upper left corner, press quick settings, and under "performance" turn off "Auto RAM Clean".
 

kol12

Honorable
Jan 26, 2015
2,109
0
11,810
Hope a thread revive is OK. I suffer from this is issue of intermittent 1-3 sec freeze + GPU usage dropping to 0%. I am at a total loss, I don't have Advanced System Care, what could be the cause for me?