GTX 970 questions

[droopyRO]

So, i have this GTX 970 Phantom from Gainward. Temps at full load reach 79-80ºC i decided to replace the thermal paste with something aftermarket i had laying around namely Arctic Silver 5 and Noctua NH1.
At first it looked like in the picture, when i saw that i thought that was the problem, so i cleaned that mess and replaced it with Noctua NH1 after about a week the temps were the same. Then i cleaned that too and used AS5 and waited 10 days for the burn in, still get 80ºC.
What i did next is set the fan to 100% and play, i get 77ºC. I know that due to Boost 2.0 it changes clock speed based on temperature and power target but it looks like i hit a wall and dont know if that is the cooler of the card sensors or something else.
What is going on ? problem for me is that if the fan speeds upwards of 60% it becomes loud for my tastes.

 

[st2000]

thinking of wrong sensors or not smooth surface of radiator
thinking of 2nd variant cause there should be some difference with different TIs

 

[RCoon]

Your temps are accurate compared to reviews

http://www.eteknix.com/gainward-phantom-geforce-gtx-970-4gb-graphics-card-review/18/

"The Gainward GTX 970 Phantom sacrifices temperatures to reduce noise: something I prefer. Despite this we still manage to stay south of 80 degrees celsius and in Unigine Heaven it didn’t even reach the thermal throttling point of 79/80."

The phantom has a smaller heatsink area than most other 970 aluminium stacks, and only two fans. It's designed for quiet operation, not good temps.

 

[Cybrnook2002]

70-80 is normal for the 970

 

[droopyRO]

70-80 is normal for the 970

It's designed for quiet operation, not good temps.

I know the temps are normal, that was not my question, but i expected a difference in load temperatures something like 3-4ºC lower while replacing the standard TIM, nothing happens. Idle it is completely silent but if the fans rev up above 60% or about 1500 rpm they become audible. Thanks.

 

[Cybrnook2002]

The piece that you may be missing, is if the Heat-sink is saturated already, changing TIM will not "lower" your temps (as the transfer of heat from your GPU to the HSF is NOT the issue), it's the HSF being able to dissipate the heat faster than the rate the heat is being transferred to it.

It's not always a guarantee that changing TIM will lower your temps. This will only help when the transfer of the heat is less than adequate. Sounds like the heat is getting into your HSF just fine .

 

[rtwjunkie]

TIM for the most part is TIM, and if they applied it correctly at the factory, then indeed you would see little to no improvement by changing out the TIM.

 

[vega22]

despite the massive overflow the contact patch in the pic looks pretty good tbh.

you may find it was already ideal and matching the temps is as good as you can get from normal tim.

you could try better, diamond based, tim which should give a slight improvement or even the metal based tim which are the best you can use.

 

[Ruyki]

Factory did a good job. It's a misconception that excess TIM around the die has a significant adverse effect on temperatures. As long as there is a thin layer between the contact surfaces, the excess TIM does not really matter.

 

[Cybrnook2002]

As long as it's NON-conductive TIM (Not like AS5), you don't want to short out the resistors on the top of the GPU etc.... (Why I now use MX-4)

 

[newtekie1]

I know the temps are normal, that was not my question, but i expected a difference in load temperatures something like 3-4ºC lower while replacing the standard TIM, nothing happens. Idle it is completely silent but if the fans rev up above 60% or about 1500 rpm they become audible. Thanks.

TIM isn't the problem, Gainward used a shit cooler design. The finstack is extremely small, so there isn't a lot of surface area to dissipate heat from. The card is just poorly designed.

 

[BUCK NASTY]

Don't think anyone has asked, but what are your ambient temps in that room?

 

[Nordic]

I wonder if your gains were minimal in temperature reduction, and you card automatically put it towards reduced noise with the fan curve. Did you notice if your card got quieter?

TIM for the most part is TIM, and if they applied it correctly at the factory, then indeed you would see little to no improvement by changing out the TIM.

In the last 5 years I have replaced the TIM on at least 5 cards, and only one had it right. My first 7970 saw an improvement of 10c.

 

[rtwjunkie]

I wonder if your gains were minimal in temperature reduction, and you card automatically put it towards reduced noise with the fan curve. Did you notice if your card got quieter?


In the last 5 years I have replaced the TIM on at least 5 cards, and only one had it right. My first 7970 saw an improvement of 10c.

No doubt! But the OP had no difference in temps, suggesting it was fine to begin with.

 

[droopyRO]

In other cards the temps went down sometimes by 10ºC probably due to older TIM. I will look for an aftermarket cooler then. Thanks all.

. Did you notice if your card got quieter?

RPM is the same about 1800-1900.

suggesting it was fine to begin with.

Either that or Gainward and me both suck equally at applying TIM

 

[droopyRO]

A bit of an update, so i got this Arctic Cooling Xtreme III(i regret now i didnt pay 10$ more for Xtreme IV) but installation problems aside the difference is huge. From 80ºC load stock and 85ºC load overclocked it went to 62ºC with overclock. Seeing the Xtreme III is compatible with the GTX980Ti it`s a keeper.

 

[swaaye]

As long as it's NON-conductive TIM (Not like AS5), you don't want to short out the resistors on the top of the GPU etc.... (Why I now use MX-4)

AS5 isn't conductive. It is said to be slightly capacitive, which may or may not cause problems if you slop it everywhere. But it will not short anything out.