Hot GPU

[KE2012]

Okay, so I've been on here a few times to try and get to the bottom of my hot GPU with little luck, so hopefully 3rd time is the charm. 

 

When gaming or under any heavy load my GPU, an EVGA GTX 980 SC, reaches the low - mid 70's on water. (Idle is high 20's) My GPU block is a XSPC GTX 980 980TI block with  some kryonaut as TIM. I have cleaned out the block not so long ago, it had 0 debris blocking it up. 

I have 2 rads, a 360x60mm Alphacool Nexxos UT60 and a XSPC 360x30mm EX360, with  3 Corsair SP 120 fans running  and 3 no namers on the thicker rad which are actually surprisingly good. Pump / Fan speed seems to make little difference to my temps. 

 

CPU temps are great under heavy load so I know heat is getting out. 

 

(Picture of loop is in spoiler)

Spoiler

 

Changing the ball valve to force the water to go through both rads doesn't make any difference to temps. 

 

So, any ideas as to what is causing such high temps?

 

Thanks for reading and thanks in advance to any who reply!   

 

 

[NicekillBG]

You think that is hot

wait let me see that again

yup you are mental 

[JakeNTech]

 

are the pumps working properly? are the radiators bloked? Try redoing the thermal paste. make sure that the case has good air flow too as that will help cooling as well, are your ambient temps low? Is your card overclocked?

[W-L]

5 minutes ago, KE2012 said:

-SNIP-

Why do you have two lines coming off of the GPU that will redirect the waterflow if you have the ball valve open. 

[KE2012]

53 minutes ago, NicekillBG said:

You think that is hot

wait let me see that again

yup you are mental 

Stock air cooler performed much better.... 

53 minutes ago, W-L said:

Why do you have two lines coming off of the GPU that will redirect the waterflow if you have the ball valve open. 

It's so that if I remove the 2nd rad for what ever reason (Like testing temps with and without it) I can still use for the computer. 

 

53 minutes ago, fixitnow said:

 

are the pumps working properly? are the radiators bloked? Try redoing the thermal paste. make sure that the case has good air flow too as that will help cooling as well, are your ambient temps low? Is your card overclocked?

Pump yeah its definitely pushing the water around, ive done the thermal paste about 100 times, no exaggeration, cleaned both rads numerous times too, ambient temps are around 15DC, only a small overclock applied but removing it barely effects the temperatures. 

[W-L]

3 minutes ago, KE2012 said:

Stock air cooler performed much better.... 

It's so that if I remove the 2nd rad for what ever reason (Like testing temps with and without it) I can still use for the computer. 

Where do those two lines go to the bottom rad? 

Try turning the pump up a little and see if you get any different due to the long tubing runs you have. 

[KE2012]

1 minute ago, W-L said:

Where do those two lines go to the bottom rad? 

Try turning the pump up a little and see if you get any different due to the long tubing runs you have. 

The two lines go off camera slightly but they go straight from VGA to rad 2, which then goes straight into rad 1. I thought the pump speed might be at issue too but it really doesn't make any real difference, and you can hear and feel it power up, i have no doubt it's really pushing the water around when set to max. 

[W-L]

2 minutes ago, KE2012 said:

The two lines go off camera slightly but they go straight from VGA to rad 2, which then goes straight into rad 1. I thought the pump speed might be at issue too but it really doesn't make any real difference, and you can hear and feel it power up, i have no doubt it's really pushing the water around when set to max. 

What pump and res are you using just to eliminate some variables, only thing I can really think of is as long as you keep that valve closed is a lack of mounting pressure since you do have good idle temps. 

[KE2012]

5 minutes ago, W-L said:

What pump and res are you using just to eliminate some variables, only thing I can really think of is as long as you keep that valve closed is a lack of mounting pressure since you do have good idle temps. 

Pump is an   EK-XTOP Revo D5 PWM and the res is a Bitspower ;Z-Multi 100mm Water Tank.   

[W-L]

8 minutes ago, KE2012 said:

Pump is an   EK-XTOP Revo D5 PWM and the res is a Bitspower ;Z-Multi 100mm Water Tank.   

Hmm that shouldn't be any problem as long as you have the PWM line connected, just to do a test since you have quick disconnects remove the second rad form the loop but keep the valve to see how that affects things or if there is trapped air in the GPU block. 

[KE2012]

17 minutes ago, W-L said:

Hmm that shouldn't be any problem as long as you have the PWM line connected, just to do a test since you have quick disconnects remove the second rad form the loop but keep the valve to see how that affects things or if there is trapped air in the GPU block. 

Okay, after a 10 min Fire strike Stress test this is the result, As you can see the temps had not even started to stabilize when I ended it.  

Spoiler

 

I'll chuck in the other rad in a moment or two when temps have cooled a little and upload its graph too. (Also side note, remember to turn down pump speed before disconnecting QDC haha)

 

[KE2012]

@W-L Sorry for taking so long to get the results with the 2nd rad, life just happened.

 

Spoiler

This was with fans at full, top lid of case removed  (And the cheapo fans overpowered with lab power supply) Also pump at full. 

[W-L]

15 minutes ago, KE2012 said:

@W-L Sorry for taking so long to get the results with the 2nd rad, life just happened.

This was with fans at full, top lid of case removed  (And the cheapo fans overpowered with lab power supply) Also pump at full. 

That's a little warm but not too bad from the looks of things, is everything overclocked in the runs?  

 

Still a little stumped at why things aren't a bit cooler than they are.

[KE2012]

24 minutes ago, W-L said:

That's a little warm but not too bad from the looks of things, is everything overclocked in the runs?  

 

Still a little stumped at why things aren't a bit cooler than they are.

Those temps are with the overclocks on both CPU and GPU removed. 

[W-L]

3 minutes ago, KE2012 said:

Those temps are with the overclocks removed. 

Yeah that would be on the hot side of things then, personally with a gtx970 and a UT60 and 240mm XT45 it doesn't get that hot when overclocked. 

 

Hmm kind of at a loss here then at what it could be really... 

[KE2012]

5 minutes ago, W-L said:

Yeah that would be on the hot side of things then, personally with a gtx970 and a UT60 and 240mm XT45 it doesn't get that hot when overclocked. 

 

Hmm kind of at a loss here then at what it could be really... 

I'm really starting to think it could be the block. I'm half tempted to get a fine emery cloth and rub the contact areas with the GPU, see if it removes any oxides or some other muck I can't see.

[W-L]

1 minute ago, KE2012 said:

I'm really starting to think it could be the block. I'm half tempted to get a fine emery cloth and rub the contact areas with the GPU, see if it removes any oxides or some other muck I can't see.

When you took it off before does it look like there was good spread of the thermal paste on the die and even contact around the thermal pads, a lack of mounting pressure can cause higher load temps, the GPU design should work with that block without issues though being a reference designed PCB.

[KE2012]

5 minutes ago, W-L said:

When you took it off before does it look like there was good spread of the thermal paste on the die and even contact around the thermal pads, a lack of mounting pressure can cause higher load temps, the GPU design should work with that block without issues though being a reference designed PCB.

Yeah it has a good spread. I'll probably take it off in a tad (Tomorrow or day after maybe) to rub those areas so can send a picture then or something. I have thought it could be a lack of pressure so I have a plastic washer to decrease the distance the screw can go in on the 4 die screws, pulling it onto the die more, this had no noticeable effect. 

[W-L]

2 minutes ago, KE2012 said:

Yeah it has a good spread. I'll probably take it off in a tad (Tomorrow or day after maybe) to rub those areas so can send a picture then or something. I have thought it could be a lack of pressure so I have a plastic washer to decrease the distance the screw can go in on the 4 die screws, pulling it onto the die more, this had no noticeable effect. 

Just as a second option try to contact XSPC just to confirm that the block is compatible with that model of GPU and if they have some general data on what it should be. 

[KE2012]

3 minutes ago, W-L said:

Just as a second option try to contact XSPC just to confirm that the block is compatible with that model of GPU and if they have some general data on what it should be. 

Yeah I tried and long story short I've sent a complaint to Trading Standards UK because of how bad / unprofessional their support was. 

[KE2012]

@W-L Just thought I would give you an update. I got a fine emery cloth and give my contact areas a rub, since doing that I've dropped to the low 50's in temperatures. So im guessing my block had oxides or muck or something else preventing efficient heat transfer.