Graphics Card massively losing performance.

[SeanGoku]

Hi guys. I've noticed my graphics card dropping to umplayable Clock rates lately for no apperant reason.
I play League of Legends and never had this problem in the past. I've noticed it happen during 4 games so far and for it to drop far enough to make that game unplayable is pretty harsh..

So, what I can say so far is:
- I had a slight overclock that was runing stable for probably years now.
- Even with the overclock I can let FurMark or Valley Benchmark run for at least 20 Minutes without anything happening.
- During these stress tests the graphics card can reached a max of 92 degrees celcius and didn't crash or make any problems in general.
- You cant see it on the screenshots, because I was too dumb to expand them enough. But the voltage drops together with both the clocks.
- Nevertheless I toned down the overclock in both the GPU Clock and Memory Clock just to be sure. (Didn't solve the problem.)
- The only thing I changed on my pc lately was that im now using a G410 Keyboard instead of a G19.
- I've tried updating my graphics card drivers (even tried a complete new install) and tried updating my mainboard drivers (witch were already "up to date").
- Just now while posting this it happened again. I'm not even in gaming. I have Skype, Teamspeak, Firefox and MSI Afterburner running.

The last game I played I let Afterburner run to observe the hardware stats and I made a screenshot each time a drop happened. Here are the 2 screenshots:

First Drop:
http://i.imgur.com/j8kwPQ9.png
(about 5 minutes into the game)

Second Drop:
http://i.imgur.com/LV73d1V.png
(not sure, butr probably 10 - 15 minutes into the game)

Drop during this posting:
http://i.imgur.com/SBh7vPR.png
(not even overclocked at this point, I had it reset back to standard values.)

[Mr.Meerkat]

1. What GPU do you have?
2. 92C is a tad too hot IMO...85C is the highest I would allow my GPU to go.
3. Maybe try putting the power limit higher?
4. Just out of interest, what PSU do you have?

[SeanGoku]

Hi there Mr.Meerkat. And thanks for your quick reply.

Yeah, it never actually reaches anything close to 92 during gaming. Especially not during a League of Legends game. 40ish, maybe 50ish was the max I've witnessed.

I have a NVIDIA GTX 650 Ti from EVGA.
I tried putting the power up, but it didn't seem to have an effect.
The PSU I have is a Xilence 420 Watt.

If you need any more info let me know. And thank you for your time and help in advnace. :)

[Mr.Meerkat]

That PSU isn't the greatest but if it was working before fine then there's no reason for it to not work anymore :/

Have you tried reinstalling your drivers?
Maybe up the voltage of the core by a few mV may help?
When the dips happen, do memory and core clocks dip with the GPU usage?

[Mahadir Iqbal]

The gpu is suffering high temperature.cool it first.then make sure its in the top pciE16 slot (the first one) for maximum performance.hope this helped you.

Follow me on twitter @mahu_rdx

[SeanGoku]

Yeah I reinstalled the graphics card drivers from scratch.
I tried increasing the Core Voltage. First by 12, then 25 then 50 (max). Nothing happened.

The Gpu Core usually ran at 900ish. Around 1k with the overclock. Currently after it dropped it's down to 300ish sometimes jumping up to 500ish.

During the game the GPU Usage actually went up, as if the 300ish mark, was the max the GPU could offer. Before that it wasn't maxed out.

This is what my MSI looks like.
http://i.imgur.com/9hii5ul.png

[Mr.Meerkat]

Hmmmm...have you tried using older drivers?

Anyway, I'm not too sure as the most common reason for clock speed issues is driver issues but...doesn't seem like the issue here :/
Do you perhaps have a second PSU you can try or even a second computer to rule out a GPU or a PSU problem? :D

[SeanGoku]

Na, I don't have any components around to trade anything out. :(

Gonna search the web if I can fiond an older driver for now. In any case thank you for your help.

[Haydenp10]

I know this may seem a little silly but, make sure you have all the cables from the GPU plugged into the PSU, my 970 only have 1 of the 2 cables actually plugged into the PSU resulting in lower performance.

[Mr.Meerkat]

...the double post bug can't still be happening with a forum this stripped down...right? :/

[SeanGoku]

Thanks for the idea, but my PSU isn't modular and the only one cable (a sixpin) is still firmly in place.

[SeanGoku]

The servers are still temporarily not available to me, even now. It's probably still due to the giveaway. :P Winning that sure would solve some of my problems. xD

Edit: Also, can I delete my own posts? I should have posted this message as an edit to the old one. ^_^

Edit2: I tried installing a driver from july. Everything was fine back then. Still the problem persists.

[Tsx]

Clean your gpu fan cooler and if you can renew the cooling thermal gel. As one of the others said over 90 degrees is to hot. Have you check windows energy settings? Set it to performance.

[SeanGoku]

Hi there Tsx and thank you for your reply.

I dont havy any cooling gel i could apply. But I did clean it up the best I could.

And yes, I set the windows settings to performance back when I went and overclocked it. Also just to be sure I checked it again, it hasn't changed.

 

Yesterday I played 3 more games just to see if it would still occure, but it didn't. So maybe the older driver (as that is the only change I made since coming here to the forum) did have some effect.

 

I'm not convinced it's solved yet thow. As the problem started it wasn't persistant in every game either.

 

Also as mentioned, the drops were occuring during league games while the temperature was between 40 and 50 degrees.

Sure, even for stress tests 90 is too high. But that was just to say, that even at these extremes it was still running at max clock speeds. But during a puny little league game it was droping down to a third of it's capabilities.

 

Anyway. If the problem doesn't occur anymore over the next few days I'll let you guys know. Who knows, maybe it actually was the driver.

 

But in any case. Thank you all for your time and efford.

[SpaceGhostC2C]

19 hours ago, SeanGoku said:

Hi guys. I've noticed my graphics card dropping to umplayable Clock rates lately for no apperant reason.
I play League of Legends and never had this problem in the past. I've noticed it happen during 4 games so far and for it to drop far enough to make that game unplayable is pretty harsh..

So, what I can say so far is:
- I had a slight overclock that was runing stable for probably years now.
- Even with the overclock I can let FurMark or Valley Benchmark run for at least 20 Minutes without anything happening.
- During these stress tests the graphics card can reached a max of 92 degrees celcius and didn't crash or make any problems in general.
- You cant see it on the screenshots, because I was too dumb to expand them enough. But the voltage drops together with both the clocks.
- Nevertheless I toned down the overclock in both the GPU Clock and Memory Clock just to be sure. (Didn't solve the problem.)
- The only thing I changed on my pc lately was that im now using a G410 Keyboard instead of a G19.
- I've tried updating my graphics card drivers (even tried a complete new install) and tried updating my mainboard drivers (witch were already "up to date").
 

What happens with CPU usage during those dips? Since your GPU temps were moderate, it could be CPU temps causing the CPU to lag behind, or maybe some background process hogging the CPU. Windows update, maybe? I've heard Win 10 isn't very polite when it comes to timing its update installs. Since it stopped happening, it could have been that. In any case, I would suggest adding CPU usage to Afterburner's plots, so you can check that as well if it happens again.

 

19 hours ago, SeanGoku said:

- Just now while posting this it happened again. I'm not even in gaming. I have Skype, Teamspeak, Firefox and MSI Afterburner running.
 

That is to be expected, since the GPU was essentially idling. Only the dips while gaming are a concern.

[SeanGoku]

Hi there SpaceGhostC2C.

I didn't see the CPU at the time the dips occured. You might have a point. And your right about the idling ofc. I was just up the wall at that point not even thinking straight anymore.

Anyway, I played another 3 games today and the problem didn't occur anymore.

I'm gonna just assume it was either a windows update that fked over my pc (thow  I never had such problems before, even during updates, as League of Legends uses close to no resources) or the latest update of the driver, that caused the problems every once in a while.

 

Thanks again to all who came here to help me.

 

Have a nice day.