The GTX 980 Ti owners thread.

tyler_jrb · 17 Oct 2015 at 19:33

AWPC said:
I appreciate that people are trying to help:

1: With the card out of the case on the motherboard on its own the internal GPU temps are the same. With or without a 120mm case fan being held next to the backplate where all the heat gathers.

2: 3840x2160 @ 60FPS is what the card cannot handle in some games (Arkham Knight yes Mad Max no) without going to 100% fan mode. 1920x1080 @60 or even 2560x1440 @60 the card can handle in silence & temps rarely go above 70C its well within thermal tolerances.

3: I have looked at several reviews where they took thermal images (toms hardware even took video) of the backplate region. Within 30 mins its gone red hot around 101.C with no cooling to it either on an open bench test rig so within a case it will go a little higher. This seems to be the cards weak spot it needs a backplate to prevent board warping if its not the marginal reference cooler otherwise you need a backplate model which traps the heat even when they drill loads of little slots in it like the MSI has the entire backplate drilled out into little slots (they would not do that unless they knew it was too hot would they!).

4: Precision X does not help if you set a target temp the card stutters a little when it dials back the power/volts to keep it within parameters.

This tells me its not the case its the jump from 1080/1440 to 4K basically maxes out the GPU & entire PC producing more heat. I am just going to keep monitoring temps in games & keep an eye on things for now see how it goes.

The backplate will be hot. That's residual heat from the VRMs not the core. The VRMs will get hot and are designed to take the heat. Those pictures will be from the reference titan x cooler. an aftermarket cooler will be cooler as they have better Heatsinks and overall cooling for VRMs. Titan x's can easily throttle at 92c with their reference cooler.

If someone else is getting 10-15c cooler than you with the same card and same settings. Doesn't that tell you something?

IceWolf · 17 Oct 2015 at 19:50

With regards to case airflow, it is also helpful to check that the fans are running at full speeds: often the case fans are set in the BIOS to adjust their speed according to the CPU temperature, so if your CPU is happy the poor GPU will be baking. Check under power options in BIOS.

Also it has been shown that good airflow in a closed case can give LOWER temperatures than an open test bench, since the case fans circulate the air better than convection currents from the heat alone. So just because it gets hot out the case doesn't mean it can't be better inside with good airflow. It sounds like you have found a happy compromise between performance and temps now anyway.

Dirk Diggler · 17 Oct 2015 at 20:10

IceWolf said:
With regards to case airflow, it is also helpful to check that the fans are running at full speeds: often the case fans are set in the BIOS to adjust their speed according to the CPU temperature, so if your CPU is happy the poor GPU will be baking. Check under power options in BIOS.

Also it has been shown that good airflow in a closed case can give LOWER temperatures than an open test bench, since the case fans circulate the air better than convection currents from the heat alone. So just because it gets hot out the case doesn't mean it can't be better inside with good airflow. It sounds like you have found a happy compromise between performance and temps now anyway.

You make a really good point there. I have Speedfan configured to raise my case fans to 50% when the GPU hits 65C. If it was left to the BIOS, they would be shifting next to nothing because the CPU is under water and reaches around 35C during the heaviest of gaming sessions.

This might well be his problem. I'd also be interested in the ambient in the room, as it may also be a factor.

Purgatory · 17 Oct 2015 at 21:17

arc@css said:
+1

Welcome to the dark side arc

Nice purchase on your 980Ti and really hope you are enjoying your upgrade mate (Just realized you went green and got a cracking upgrade from your previous GPU).

tyler_jrb · 17 Oct 2015 at 21:19

Hmm maybe not. Seems like im having to disable and re-enable SLI everytime I boot for it to downclock on idle. boots into windows at 38% power. requires disable and re-enable of SLI to drop back down to 8%. This is going to take some figuring out...

Dirk Diggler · 17 Oct 2015 at 21:28

tyler_jrb said:
Hmm maybe not. Seems like im having to disable and re-enable SLI everytime I boot for it to downclock on idle. boots into windows at 38% power. requires disable and re-enable of SLI to drop back down to 8%. This is going to take some figuring out...

Have a look at start up programs in the task manager and try a selective start up?

Joe! · 17 Oct 2015 at 21:36

jc0 said:
[...]

Arrived today but disappointed/surprised I'm only able to achieve a stable OC on the GPU of +75MHz (1440MHz). Anyone else think I'm being short changed there? It's running very cool at 47~ dog under load.

Very nice purchase - Make sure you register the card with EVGA and claim your "free" (you have to pay postage) shroud

As for the overclock, remember that a reference 980Ti is supposed to run at 1075MHz, so you're really getting a 365MHz boost, which is a lot. You won't notice much, if anything boosting up to what's average & stable for a 980Ti (1450-1500) compared to 1440MHz, so enjoy your card as it is.

tyler_jrb · 17 Oct 2015 at 21:55

String said:
Have a look at start up programs in the task manager and try a selective start up?

ill have a fiddle tomorrow. Seems if I disable and re-enable SLI. it almost locks my GPUs to a maximum of 45% power usage and they run massively low fps. re-start the system, still 38% power at idle but they run fine now at load right up to 80-90% power. This is wierd. Recently replaced motherboard due to a fault. Thinking re-installing windows might be a possibility.

jc0 · 17 Oct 2015 at 22:01

Thanks Joe! After reading rumours that pascal might not be be released on a high end card until possibly q3 next year, I decided to sell my pair of 670s while they're still worth something.

Here's a pic of the installed 980 Ti for anyone who's interested...

jc0 · 17 Oct 2015 at 22:06

Sorry, forgot to ask... Am I right in thinking that the voltage on these cards cannot be be tweaked past 1.212V?

Intra-Agnostic · 17 Oct 2015 at 22:42

Just waiting on a cablemod kit

Joe! · 17 Oct 2015 at 22:52

Mine have all gone up to 1.243v (+87mV in Afterburner)

jc0 · 17 Oct 2015 at 23:09

Ah thanks Joe! I hadn't been putting it all the way up to 87mV. That might help me squeeze a bit more out the core!

I did apply for the evga shroud btw...

bimbleuk · 18 Oct 2015 at 05:22

jc0 said:
Arrived today but disappointed/surprised I'm only able to achieve a stable OC on the GPU of +75MHz (1440MHz). Anyone else think I'm being short changed there? It's running very cool at 47~ dog under load.

As already mentioned in reality you're already getting a large increase but I see many cards happily running around the 1450MHz range. Mine is running at 1450MHz (+125 on a reference SC based Hybrid). I can run it at 1475 but that's 5Hz less than the unstable speeds so I backed down for longevity. It may improve a little bit after a good burn in period.

jc0 · 18 Oct 2015 at 09:38

While running Heaven benchmark, I'm seeing the GPU clock speed fluctuate (and voltage with it) quite a lot. It will sit at 1455MHz for a bit then go up to 1468, then back down again and so on. There's no throttling going as it never gets above 44 degrees Celsius.

Is this normal?

IceWolf · 18 Oct 2015 at 10:01

Could be reaching the power limit rather than the temp limit. Or maybe you have adaptive power set in Nvidia control panel. Use GPU-Z in the background to track all the values so you can analyse them after maybe

Joe! · 18 Oct 2015 at 10:12

13MHz points to GPU Boost 2.0 - That's the amount it changes your clock speed when it gets above 62-ish C

Are your power limit settings as well as temperature settings set to the maximum in Afterburner?

jc0 · 18 Oct 2015 at 10:36

Thanks for the responses guys. I have maximum performance set for power management in NVidia Control Panel.

I've attached screens of my Precision settings and monitoring when running Heaven. Sorry it's not that clear in the pics but you can see the gnu clock speed fluctuating about. The temp never gets higher than 47 degrees so I really can't figure this out.

Joe! · 18 Oct 2015 at 10:40

As IceWolf said, download GPU-Z (https://www.techpowerup.com/gpuz/) and see what is capping your performance in the 'Sensor' tab whilst running heaven.

Also that screenshot is tiny! :eek:

jc0 · 18 Oct 2015 at 10:45

Joe! said:
As IceWolf said, download GPU-Z (https://www.techpowerup.com/gpuz/) and see what is capping your performance in the 'Sensor' tab whilst running heaven.

Also that screenshot is tiny!

Sorry Joe, I'd post the image thumbs instead of the real things...I've corrected them now!

I have GPU-Z installed but doesn't the monitoring in that just show the same stuff as is available in Precision?