Why on earth would they set the card to downclock in the low 60s (although not always consistently)? Does anyone know a way around this aside from flashing a different bios? I really don't feel like bricking the card or voiding the warranty, but this is pretty ridiculous and causing some gaming pain. As a sidenote I'm using EVGA Precision X with a minor overclock and kboost which downclocks below standard core clock speeds at these temps (on an pretty aggressive fan profile with ACX 2.0 cooler).

It's downclocked from standard clocks and no overvoltage - not getting a temp issue other than it's so low to be downclocking. Arma 3, GTA V, Witcher 3... lots of games actually. 1920x1080 @ 144hz. Reinstalled drivers many, many times with clean install and driver wiping. Cooler is clean as a whistle (new) and really not the problem, the problem is the insanely low temp that it's downclocking at - likely a bios issue (or "safeguard"), but as I said above I don't want to mess with the bios and mess up my warranty or brick the card.
Current plan when I get a day off work will be to try out the new driver nvidia just posted and possibly ditch EVGA Precision as I've been hearing bad things. Any other suggestions are welcome.

Depends on the game/application. Try setting NV control panel to "prefer max performance". Or you can enable K-Boost in EVGA Precision (But that will lock to 3D speeds)

I'm pretty sure all Maxwell cards are designed to start throttling at around 63C. That's why water cooling is so important on NV Cards for maximum performance on the stock BIOS. Replacing a cooler is less risky than bricking your card from flashing it.

astrallite
I'm pretty sure all Maxwell cards are designed to start throttling at around 63C. That's why water cooling is so important on NV Cards for maximum performance on the stock BIOS. Replacing a cooler is less risky than bricking your card from flashing it.

I was afraid this was the case. I plan on getting the Corsair closed loop water shroud and reusing an old radiator when it comes out, but 63 is still just ridiculous for throttling. I get wanting to keep RMAs down but crippling the card isn't the answer. You shouldn't have to modify, buy extra parts, and risk bricking your card to reach (or even get in the ballpark of) the cards potential. Going fan cooled in a very well ventilated case was acceptable before, I don't see why it shouldn't be now.
 
It'd be nice to see if EVGA or other companies could offer a supported BIOS to flash to correct what I find to be an issue with these cards. I'm also wondering if this throttling is vendor specific and EVGAs is lower than others. Does anyone from EVGA have any knowledge on this?

Vlada011
Yes 63C is ridiculous for throttling when we talk about cards with 250W power consumption.
That's crazy even if we talk about GTX980... I think that is not true, TITAN X work on 83C.
Same is with GTX980Ti, on 84C card reduce clock.
84C is not bad... NVIDIA want to prevent damage of card on longer period when temperatures pass 87-88C is not good.
Always should stay below 85C with any graphic and without their reducing clock.
 

Where did you get those numbers? I agree 80s is fine, but I'm throttling in the low 60s to below stock clocks.

Vlada011
Forget on 60-70C throttling, on TechPowerUp you can read review just exactly for EVGA GTX980Ti, SC ACX 2.0
model and they say 84C is limit but in games card work on much lower temperatures. 
If you don't believe to me you can read ...
 
 
http://www.techpowerup.co...980_Ti_SC_Plus/34.html
 
TPU pull 1290MHz base clock from their sample... I would be more than satisfied with such overclocking capabilities...
Even 1250MHz is nice number if you can play games on 1200MHz. For me that is excellent, for other people I don't know...
OK GIGABYTE 980Ti G1 go on 1310MHz with 1400MHz boost but that's custom design and that card need to be compared with 980Ti Classified not SC. 
SC is only EVGA gift for reference design and cost only 20$ more than reference but again is excellent card, much better than reference and worth investing 20$ more.

It does say GPU Boost 2.0 set 84C as a limit for the clocks. I am saying my clocks are being downed on Kboost from stock clocks with an aggressive fan profile when reaching the low 60s and thus causing issues in 3d applications. I'm just trying to figure out how to fix this without going against EVGA warranty or risking bricking my card. I'm still interested if EVGA has any knowledge of throttling being altered or tweaked on their end - or if they know why this is happening at stock. Also, as stated above, when I have some free time from work I'll start testing to see if I can find a solution on my own via software and settings.
 
As far as "being happy", I don't expect to be happy with something that doesn't consistently perform at (and/or above) advertised levels in more than adequate conditions. At the enthusiast level you pay a very high price for the ability to have the newest and best technology. I could understand having to learn to deal with something or being content with imperfections say I had taken a handout or bought an extremely cheap low end product, but this is the creme of the crop, at creme of the crop pricing, and as such most people at this level want all they are sold, ask for, and more. I don't think I'm being unreasonable wanting my card to perform at its best, and honestly I'm not going to relent until I get to that point. It's just not in my nature to settle and I don't advise other people to either. So, while I appreciate your input, I unfortunately don't agree with your logic and will continue to work on improving my experience with the card.

The bottom line is that no modern Graphics Card (nVidia or Radeon) should be throttling because of temp targets with the GPU at 60C.  If kboost is actually working, it should stay at max frequency all the way to 84C. 
 
I tested mine for 24 hours before installing the waterblock and it didn't throttle because of temp.  Performance Cap reason was always Power or Voltage.
 
What did tech support say?

Shelledfade

DRY_ICE
Power management has been set to prefer max performance and kboost has been used during all these instances I noted. My old EVGA gtx 780 would start to downclock (heavy OC) in the 70s so I don't understand if this is an intentional decrease in trigger temps for newer cards or not. I'll try to run more testing changing different variables and software when I get a day off of work in a few days.

Same for me. My card: Reference EVGA Gtx 980 Bios: 84.04.31.00.80

Card down throttles from 1506 to 1493 at 68c. It ain't much but its kind of annoying just to see that its doing that lol.

It would be nice if we could turn off this down throttling feature completely. Is there a way to do that without messing with something in the BIOS? Like this poster also said, I experience this same slight down throttle with kboost and prefer maximum performance on.

You would think kboost would eliminate down throttling completely but oddly it doesn't. I agree with the above poster, we should be able to control how our cards throttle. It shouldn't be doing that based on temp. 68c is a very good OC temp. This down throttle is unnecessary and annoying.

And you're absolutely positive its not throttling because of power limits?

Kramps

Shelledfade

DRY_ICE
Power management has been set to prefer max performance and kboost has been used during all these instances I noted. My old EVGA gtx 780 would start to downclock (heavy OC) in the 70s so I don't understand if this is an intentional decrease in trigger temps for newer cards or not. I'll try to run more testing changing different variables and software when I get a day off of work in a few days.

Same for me. My card: Reference EVGA Gtx 980 Bios: 84.04.31.00.80

Card down throttles from 1506 to 1493 at 68c. It ain't much but its kind of annoying just to see that its doing that lol.

It would be nice if we could turn off this down throttling feature completely. Is there a way to do that without messing with something in the BIOS? Like this poster also said, I experience this same slight down throttle with kboost and prefer maximum performance on.

You would think kboost would eliminate down throttling completely but oddly it doesn't. I agree with the above poster, we should be able to control how our cards throttle. It shouldn't be doing that based on temp. 68c is a very good OC temp. This down throttle is unnecessary and annoying.

And you're absolutely positive its not throttling because of power limits?

I don't believe so but I'm not exactly sure I understand what you're asking me. This down throttle happens at any mhz value between 1350 to 1506 core mhz. There's always a slight down throttle no matter what the OC is set to when temps get around 65c-68c. For example: It down throttles from say 1450 to 1439'sh or something around here. Like it does it no matter what like its some built-in mechanic or something. Sorry I kind of suck at explaining detailed things like this but I hope you get what I'm trying to say here.

My power is set to 125% maximum value.

Duarian
Mine do the same thing.
 
Throttle from 1455 to 1442/1443 and don't move again until they get to 75+ degrees C. Not a big difference but it is a little annoying.

Pretty much as this guy explains it for himself, for my card as well. Although if I push an OC like 1506, it won't drop to 1443, just a little bit lower to 1493. It's weird and annoying.

K-boost and prefer maximum performance, and setting power limit to 125% doesn't fix this problem.

I have no idea why your card isn't throttling then. Maybe there is something F'd up with 84.04.31.00.80 bios and reference 980. These newer drivers don't work with my card at all either, crash constantly. Last stable driver was 347.88 for me.

Kramps

Shelledfade

I don't believe so but I'm not exactly sure I understand what you're asking me.  Sorry I kind of suck at explaining detailed things like this but I hope you get what I'm trying to say here.
My power is set to 125% maximum value.

 
I think you are explaining it good.  What I mean is that there is 3 reasons that a gpu will throttle; Temperature, Power and Voltage.  Or maybe 4 things.  Resources I think is one also.  And of course if the full power is not needed, like browsing the internet but kboost should eliminate that.
 
Open up GPUZ and go to the second tab and watch the Performance Cap (throttle) reason.  I believe the resource screen in Precision X also details the performance cap reason.  Watch it as the GPU heats up, it will tell you the reason but since, even at stock, the cap is 84C, I'm thinking its not going to show temperature as the reason for the throttle. 
 

I should have been checking this but I guess ignored it because it was at stock clocks. It appears the GPU almost immediately hits PWR cap at max TDP when I enter a stress test - with kboost on or off and with power target set to 100% or max 110%. VDDC also drops at these times. I find it kind of strange that the card would max out power usage almost immediately and start downclocking like that? I don't have time to play around with it I need to head to work soon, but does anyone have suggestions?

DRY_ICE, can we please have some NUMBERS? "Downclocking" is not defined. What is the clock speed before the "downclock" and what is the clock speed after the "downclock"? The base clock for your card (assuming I am choosing the correct card based on your vague description) is 1102 MHz. Is it "downclocking" below 1102 MHz? If not, then it is not technically downclocking at all. The boost clock above the 1102 MHz base clock is variable and adjusts based on current conditions. GPU boost is sort of a "free factory overclock"; but it isn't constant. If the card's clock isn't dropping below 1102 MHz, it isn't technically downclocking or throttling; it would be working the way it is supposed to. Just like how Intel Turbo Boost boosts higher or lower depending on how many cores are loaded, CPU temperature, voltage, and power limit settings.

For what it's worth, Kboost is pretty much a waste of electricity and a source of extra needless heat for most users. Kboost doesn't cause your card to boost higher or stay boosted higher. GPU Boost is built into the card's BIOS and cannot be bypassed by such simple means. All Kboost does is keep your card from clocking down when the card is idle. Why would you want your card to sit at 3D performance clocks (1102+ MHz) while the card is idle?

sahafiec
to design a card with zero fan speed until 60C and to trottle it down at 63C right after the fans started spinning, that's just crazy.

Good thing they aren't designed that way...

@Kramps
but too many user live exactly that nightmare.
 
@ty_ger07
according to your definition cards never trottle as they never go under the base clock while under load. right?
throttle is when your card can't maintain it's speed and performance under load until a given temperature is reached.
yes, GPU Boost is variable but meant to increase the performance meaningfully within a range.
it's not meant to decrease it just because it's "allowed to be variable".
 
the guys are indeed confused to pay > 650$ for cards which don't do what they are advertised for.
I can clearly understand them.