GTX 980 Ti temp issues

the_bears

Distinguished
Dec 6, 2013
215
0
18,680
This will be a long thread, and while I will try to not post useless info I'd appreciate it if anyone willing to help would be able to read everything to understand my situation to know what I'm dealing with.

A few weeks ago, I've got my MSI GTX 980Ti Gaming (https://www.msi.com/Graphics-card/GTX-980-Ti-GAMING-6G.html) card and immediately once I got it, I fired up the games that I played to see how it performs @ 1440p. The card is excellent, it provides very high FPS (60+) in many games, boost itself to 1350 MHz when any game is opened, is quiet etc... But ofcourse there is a problem, and that problem are temperatures.

When the fan is at automatic mode (no custom fan curve), the temps of the cards were at 54-57C, when a game is opened and played for awhile the temps get very fast to the temp limit (83C). I posted this in another forum and was recommended to improve my cooling and reapply TIM. I do both things by buying additional 3 fans (all intake 2x front, 1x bottom) and disassembling my GPU and reapplying the TIM (X method was used here). After all of that was done, I can say that the temps while they did drop didn't drop by much at both idle and load (3-5C).

After this I still didn't like the 80C temps and have gone and set my own custom profile (https://imgur.com/C49Z4fO). Now the temps are at 37-40C at idle and 70-77C at load, and also the worst part about this is that the fans are spinning 95-100% at those temps and it's still in high 70's ?!?! Why ? .

This is the temp issue which I'm talking about, why would it require almost 100% of fan speed just to keep the temps at high 70's ? I have to mention that I tried to OC my card, and managed to go up to +115 on the core speed in Afterburner, but the temps then rise to 80-84C with fans still spinning at 100%.

Also me taking the side panel off doesn't change temps at all.

I hope anyone might know how to help me solve this issue, as this is the only thing that is stopping me from gaming normally since I'm too much worried about this.

Any questions on any info which I've missed, please ask.
Also my ASIC is 77.5% , I'm on stock cooling (Twin Frozr) and stock BIOS.

My specs:

CPU: i5 2500K @ 4.5 GHz
GPU: MSI GTX 980 Ti Gaming Edition
Motherboard: AsRock Z77 Extreme4
RAM: Patriot 8GB (2x4GB) 1600 MHz
CASE: Fractal Design Define R5
SSD: Samsung 850 EVO 250GB
HDD: Samsung Spinpoint 1 TB HD103SJ
PSU: SuperFlower Leadex Gold 650W

Thanks.
 
Solution

Jwpanz

Honorable
I don’t think there’s much you can do to flat out solve this. The issue seems to be the cooler just not cooling well enough. This could be MSI’s fault of the product of Nvidia giving MSI a lower tier chip.

In all honesty, temps in the 70’s shouldn’t worry you. If you don’t like those temps then you could opt for a liquid cooling solution. NZXT has an AIO you could use with a mounting bracket or you could go with a full on custom loop. EK just came out with an affordable aluminum liquid cooling system.
 

TheFluffyDog

Honorable
Oct 22, 2013
469
0
10,960
First things, although the x-method can work fine for GPU's, because the entire silicon is exposed, you need to cover every square mm. Unless there was enough excess in the x you laid down, its possible you have a few uncovered spots. I like the pea and x methods for CPU but i do spread myself for GPU (unless you plan on using enough to ooze out the sides). Further to that, if your not going for "MAXIMUM OC" more thermal paste is better than not enough granted its not electrically conductive.

And after that, i honestly think there isnt an issue. the 980ti does produce a fair amount of heat, and if your room is more liek 25C or 77F (rather than a typical 20-22C for testing) being in the mid to upper 70's to low 80's isnt abnormal. Do you know your typical room temperature?
 

TheFluffyDog

Honorable
Oct 22, 2013
469
0
10,960
I have all of my computer gear in the basement, which is only barely finished so its not 100% insulate. My temps down there in the summer usually dont go above 20C and in the winter it could be <15C. this results in quite a bit of fun as the kraken and AIO system described above can hold my 980ti OC to 1575Mhz @ 28-32C the entire time it runs. I my bedroom upstairs however, even in the winter, a gaming session with the door closed can easily being the room temp above 75F (right around 25C) which would severely impact PC cooling.
 

the_bears

Distinguished
Dec 6, 2013
215
0
18,680

The temps aren't what worries me too much when in the 70's, more like why is my fan spinning at max while temps are high.


Well the throttling starts to happen at 84C, so I'd rather keep it to 80C max.



Hmm, I might try to remove the paste and reapply it by spreading all around the chip (like MSI did). Unfortunately I don't have a thermometer in my house so I don't know the exact ambient temperature, but if I would to guess it would be something like 27C.

 

TheFluffyDog

Honorable
Oct 22, 2013
469
0
10,960


yeah, more is better for someone who doesnt build and rebuild upgrade and mod regularly. AS for 27C room temp, thats 80F so if you think your room is atleast around or above the 25C mark (77F) than your temps are normal. very few 980ti's if any in a 25+C room cooled on air will see temps much lower than the upper 70C's
 
1. First off let's clarify some misinformation.... temps > 84C are not "fine" because the card throttles above 84C

2. In TPus tests, the MSI Gaming X 980 Ti produced the following results.

MSI GTX 980 Ti Gaming = 50°C @ Idle / 74°C @ Peak gaming load / 75°C @ Peak gaming load w/ OC

At 9C below the trolling point, it's not the card unless something is wrong w/ that particular card. Your efforts to date seem to rule that out.

3. What is the weather like at your location ? Your idle and load temps are 4 - 7C above what should be expected.... or 7 - 13F ... If you have ambient air temps of 80 - 85+, what you are seeing is normal

4. When you installed the TIM ....

a) Did you fully cover the GPU ?
b) Did you dully cover all the VRM and memory chips (both sides of pads / where ) ?
c) What about backplate side ?

I use Shin Etsu 751 TIM on CPUs as it's in the top 3 of TIMs, has no curing or capacitance issues like AS5 and it's < $4. However, we use Gelid Extreme on GPUs VRMs, Memory because a) you need a lot more and it takes much more time to apply to all those surfaces and b) you get a lot more.

Shin Etsu loses its workability after being exposed to air for as long a sit takes to apply to all those surfaces, It also comes with an application tool which serves well on GPU surfaces.

5. We don't recommend CLC coolers on nVidia GPOUs as unless you have ambient air or case cooling issues, you will generally remain far below the throttling point and there's nothing to be gained but bragging rights to low temps whether you are 40C or 75C. The problem is many of the aftermarket CLC solutions make VRM / Memory cooling worse.

6. That card can pull 281 watts under peak gaming loads.
https://www.techpowerup.com/reviews/MSI/GTX_980_Ti_Gaming/28.html

w/ Overclocking you can increase the power limit by 20%
http://www.guru3d.com/articles_pages/msi_geforce_gtx_980_ti_gaming_oc_review,35.html

That's 337 watts for GPU / 140 for CPU, 35 MoBo, 40 watts everything else, say 550 watts of heat.

With 120mm fans @ 1200 rpm max, you'd want 1 case fan for each 50 - 75 watts ... that's 7 fans
With 140mm fans @ 1200 rpm max, you'd want 1 case fan for each 75 - 100 watts ... that's 5 fans

Whaddya got ?
 

the_bears

Distinguished
Dec 6, 2013
215
0
18,680


Sorry for the late reply.

3. Weather right now is around 21C (20, 22,19 etc...), also the ambient might be I guess around 27C but there is no thermometer here.

4.
a) yes, everything "clicked" while I was trying to match the card onto the heatsink
b) this is one thing I forgot to mention when I disassembled my GPU before, thermal pads were stickied at the heatsink
, and weren't like this http://www.overclockers.com/wp-content/uploads/2015/06/cardnohs.jpg?w=600, but I don't think there was any issue because they fitted normally when I put it all back.
c) I never removed the backplate, only the 4 screws for heatsink.

Also I used my Arctic MX-4 TIM

5. No liquid cooling here.
 
2. Again the main question here is ... why are you results so different from published test results ?

3. No candy thermometers upsstairs in utentil drawer ? No thermostat ? Got $2 ?
https://www.walmart.com/ip/AcuRite-8-Thermometer-00322/16888912

4. Not sure how clicking fits in but when removing heatsinks, backplates, etc.... all mating surfaces must be cleaned w/ isoprophyly alcohol. Any damaged distorted thermal pads must be replaced w/ TIM applied both sides.

5. I was addressing a comment above where CLC was recommended.

6. This is the big one.

7. Case configuration (inlet versus outlet locations) may result in dead spots.

It would appear that you can fit:

Top = 3 x 120 / 3 x 140
Front = 3 x 120 / 2 x 140
Rear = 1 x 120 / 1 x 140
Bottom 2 x 120 / 1 x 140

And from your post, I'd gather that the bolded items are installed. This paints a picture of all the fresh air coming in the front and arcing up to go out the rear and top. Id suggest removing the lower HD cage and putting two of these on the bottom blowing fresh air up into the card intakes.... w/ 6 fans you'd be closer to that 7 fan rule of thumb

 

the_bears

Distinguished
Dec 6, 2013
215
0
18,680


2. Like I said , I don't know and thats the whole point of this thread. I have no idea why my temps are so high on my GPU. They are around 50-55C at idle, and 72-79C on load and that's with the custom fan curve (which is spinning at 100%)

3. Nothing right now, also it's pointless to link Walmart since I don't live in the US.

4. What I meant by "click" is after I was done with cleaning out the thermal paste from both the heatsink and GPU die, everything fit in as it was before. Also the thermal pads weren't damaged, just that they stickied to the heatsink instead of remaining on the GPU.

6. Calculating the PSU required for my setup with this website: https://outervision.com/power-supply-calculator
recommends me that the system draws 550W.

7. As for case configuration:

2x 140mm front intake fans
1x 140mm bottom intake fan
1x 140mm rear exhaust fan.

Both HDD cages are removed, and the cable management is good. Also removing the side case panel doesn't change the temps.
 
2. That's the information that confirms what you are seeing is not normal. If it was normal, there would be no solutions. After having confirmed that it is not normal, we can then move on to the subsequent numbered items which are possible solutions. Always betetr to confirm you have a problem

3. Where you live is irrelevant, tho I would have pointed to a source in your country had you provided that information. But tho it was suggested several times that you determine ambient air temp, you just answered that you didn't have one. Recognize that, wherever you live, a) it's being suggested that you go out and get one so we can continue to diagnose your problem b) the link shows you what to look for and c) you can see that they don't cost much ... not at Walmart nor anywhere else in the world.

4. OK, understood, most folks will associate "click" with hearing a noise when assembled in place.

5. Of course they were stickied... that's intentional. The goo not only holds them in place but aids in thermal transfer. Once you take it apart, the pads should be replaced .... and TIM applied for best results. When you buy a water block, you take off the old pads, clean the surfaces with alcohol and apply the new pads that came in the box w/ the water block.

6. PSU size is irrelevant. Sizing for PSUs is based upon power draw and headroom of 25 - 50% is added to handle transients and keep actual load a reasonable distance. A system that draws 550 watts should use a 650 to 750 watt PSU.

Here is Guru3D's power supply recommendation:

GTX 980 Ti - On your average system the card requires you to have a 600 Watts power supply unit.
GTX 980 Ti SLI - On your average system the cards require you to have a 900 Watts power supply unit as minimum.

If you are going to overclock your GPU or processor, then we do recommend you purchase something with some more stamina.

With heat loads, we don't need that extra 25 - 50%. The calculator also gives you the same number for any card ... which is not real. So if we are going to bother doing calculations, isn't it best that we do them with real number based upon the specific card in question. Since providing the link and the numbers didn't get the point across let's try a picture:

980 Ti = 238 watts
MSI 980 Ti Gaming = 281 watts
MSI Ti Gaming Overclocked = 337 watts w/ 20% increase in power limit

That's 100 watts difference between what outervision is using and what is appropriate for you card (assuming you want ot OC eventually)

7. As indicated, the target number of 140mm fans for 550 watts is 5.5 fans @ 1200 rpm. You have (4) 1000 pm fans. Recognize that 4 is < 5.5 and that 1000 is < 1200 rpm




 

the_bears

Distinguished
Dec 6, 2013
215
0
18,680


2. I'm honestly confused about this one. All other users and reviews I see have way lower temps and fan speed, but recently I decided to ask MSI about this and here are their replies:
https://forum-en.msi.com/index.php?topic=292000.0

3. Ok, I got the thermometer and for a few hours of how I left it even while gaming the temps seem to be 24C (but if I keep my door and windows shut and game for quite a while, the room heats up to 25C)

6. I wrote in the OP that I use SuperFlower Leadex Gold 650W, so I hope it's good as it's Tier 1 here.

7. Forgot to mention, the fans spin at 1400 RPM.
 

TheFluffyDog

Honorable
Oct 22, 2013
469
0
10,960
So i havent commented in a while, but if you actually measured a room temp of 24C and after the re-installing the heasink with thermal paste you are seeing a temp of 75-80C thats mpretty much normal. 24 and 25C air temperature compared to 20-22C (typical room temp) can make a really, really big difference because the rate of heat transfer is dependent on T difference. For instance, the card will produce a certain amount of heat in watts, and the heat sink needs to absorb it and transfer it to the air. The temperature of the heat sink will always be colder than the chip (except for right next to the chip). SO the fins that make up approx 50% of that heatsink are not at the temp of the GPU, they are probably 10C cooler near the edges and probably 5C cooler 50% between the edge of chip. so its the difference between the fin temperature and the air temperature that determines the rate of heat tranfser. And since the difference between fin and air temp is always smaller than the difference between gpu and air temp, a 2-4C difference (as in your case) could be close to a 1/8 difference or 13%. 13% of the benchmark numbers (probably taken at 20-22C) is right around 7C

So your 4C difference in room temperature could require a near 7C core temp increase to actually get the proper number of watts removed from the card.

BAsically, if your room is 2-4C warmer, it will raise you card temps by much more than 2-4C
 

the_bears

Distinguished
Dec 6, 2013
215
0
18,680


Ouch, now after a while of gaming (without any windows or doors open), the temps rose up to 26C. So this looks normal ?
 

TheFluffyDog

Honorable
Oct 22, 2013
469
0
10,960


yeah, i used to game with my shirt off in my room because i had the same issue. Small room, high powered computer. I moved all my gaming and entertainment equipment to the basement, and its quieter, cooler and i have more space. I also had a pretty powerful sound system, so a room temperature of 30C could happen o.o

Anyway, if you measured your room getting to 26C, thats your problem hands down. thats extremely hot to try and compare to a benchmark or review online. The good sites like anandtech, toms etc. will all run benchmarks with room temps in the 20-22C range. at its worst your room is almost 6C hotter, which from the perspective of air cooling with a radiator and heatpipes like GPU's use, is a very large difference. One of the benefits of liquid cooling is the ability to maintain greater linearity with increases in ambient temperature, where as air cooling commonly runs into thermal runaway in hot environments.

*Disclaimer* i understand that the math i posted above isn't a real formula, and nor did i take the time to actually calculate the change in the relative heat transfers, just trying to show that a small change is a bigger deal for air cooling
 
Solution