Graphics GTX 1080 boost clock stuck at low speeds after PC has been idle few days.

Hello. I have an some problem with my system/gpu/mobo I dont know which exactly.

At some point after having my system powered on, it often happens while the system is idle few days, my GTX 1080 becomes permanently throttled. The graphics clock is halved from a maximum of 1990-2000mhz (factory OC) to 1285mhz . When this occurs, it doesn't matter how much I load my card, be it games or other demanding applications, it does not ever go past the above caps. The only solution I have had up until now was to reboot the system .GPU-Z reports the "perfcap" reason as PWR

And I have found many topics with the same issue on internet, its one of them even with Gtx 970/980/1070/1080:

https://forums.geforce.com/default/...970-clock-rate-throttling-and-temporary-fix-/

Just worried because I found topic about similiar issue here: https://linustechtips.com/main/topic/387651-gtx-980-ti-occationally-gets-stuck-at-idle-clock-speed/

And reason was not stable oc. So maybe I have not stable factory OC? Also they had driver crashes from the link above, but me not, I have 0 driver crashes. And normally the event log will show the drive crash report which in my case it doesn't show.

Maybe this is different situation because my card must be idle few days and I dont have driver crash.

Any opinions ? Please help me, thank you.

PC: 6700K No Oc

16gb DDR4 Kingston Hyperx

750W Corsair RM

Asus Z170-P

SSD Crucial Bx100 256GB

Gigabyte Gtx 1080 Xtreme

 

Windows 10?

Have u tried using other GPU monitor software?

 

Win 10 yes

Other gpu monitor doesnt matter, because i had low fps. So gpuz was reporting fine.

 

I'll be honest it's a while since I left my PC to Idle for a few days, most likely a power save function of some kind. Pcie management on mobo has settings for idle periods that can be disabled for example.

That's where I'd start, PWR means the card has hit the power limit and is been declocked to save the gpu. As far as I'm aware a 30 second reboot will resolve this.

 

Weird that on my previous card 980 Ti Xtreme dont had even once that issue. I left computer few times for long time in idle and card was not throttle . Maybe something with drivers on current 1080 gtx? Because many peoples on geforce.com forum have this issue.

And normally the event log will show the drive crash report which in my case it doesn't show. I dont have driver crashes. 0.

 

Had this issue with a few Nvidia cards. 560TI, 670, 780 and once on my 1080. A reset would resolve it and later drivers would stop it happening.

Edit: I don't see how it can be overclock related if the machine is idle and doesn't increase beyond idle clocks.

 

Can you explain more brighter? "dit: I don't see how it can be overclock related if the machine is idle and doesn't increase beyond idle clocks. "

You mean its not possible that is caused by factory overclock?

 

I'm not saying it's completely impossible, but in your case if all you did was leave the computer on without giving the GPU a reason to reach its full speed, then it has no way of deciding whether the clockspeed is stable or not.

If I leave my computer on for too long, the GPU stops reporting any Power Usage data to Rivatuner Statistics Server (Which in turn gives me graphs on my keyboard's display). This is one of the factors in deciding which clock speeds can be used, it is resolved by resetting the computer and I am confident that a driver update will resolve this permanently. Remember it is early days.

 

So it happened just once in my situation. Is this reason to worry? Also you have Gtx 1080 and this happened to you just once too ? ( after idling pc long time , core clock throttled from 1950mhz to 1285 mhz in 3d and stay until restart )

 

It just happened once for me so far. I'm not worried and I don't think you should be. If it keeps happening after the next 2 driver updates then contact Nvidia driver feedback (or do so now if you like).

 

Last question. Somebody said this from link above :"Are you getting driver crashes at all?? I see it's being capped at 34.4% TDP, with PerformanceCap Reason of PWR.
That usually happens due to a driver crash if it doesn't recover properly, will get capped at low power/clocks."

But me not, I have 0 driver crashes. And normally the event log will show the drive crash report which in my case it doesn't show. So its not related to driver crash. And of course you dont had driver crash during this?

 

I have no driver crashes. This is why I think our situation is different to that one.

 

Yes so when your pc was idle some time and then card throttle gpu core in 3d, you dont had driver crash? Me not .

 

I've not had a driver crash since I installed this card.

 

So that throttling after some time resting in idle is propably driver issue not hardware? GPU-Z reports the "perfcap" reason as PWR. You had the same reason in GPUZ ? Look here is my screen:





ps:
What was your boost clock and what clock was after throttle?

 

Yes I expect it is a driver issue that will be resolved.

I didn't check what the reason was at the time, I just reset it and have never had the problem since. Is it happening to you a lot?

I noticed that at times when Rivatuner Statistics server reported 0% for my power usage, GPUz still showed a measurement. This could mean that there is more than one way to measure it, but only one of them is reporting as it should.

If one of the ways isn't reporting to GPU boost properly then the card might resort to safer clockspeeds until it gets the right information.

 

It happened once just So dont worry and dont rma card?

 

Exactly

 

Last thing. Somebody said to me this:" The driver will not necessarily throw you a TDR ("Display driver stopped responding") error, either. The overclock can fail silently after X errors."

I am worried a little that can be card issue,even if driver not necessary throw a TDR error.

What you think?


http://techreport.com/forums/viewtopic.php?f=4&t=116028



"Improper overclocking causes these symptoms.

My card usually locks into a non-performance power state after too many hardware faults due to overclocking. Like others have said, a reboot will solve it, but double-check the stability of your overclocking.

The driver will not necessarily throw you a TDR ("Display driver stopped responding") error, either. The overclock can fail silently after X errors."

 

It expect it can happen with a bad overclock, but that doesn't mean it only happens because of that.

I've seen it happen for a variety of reasons and at the moment I believe there is a problem with the driver, especially after leaving it on for too long.