New watercooled build - GPU-temperatures seems really high

[Nokiron]

Hi guys, I have some issues with my watercooled build. I just finished it a couple of days ago.
I hope you have some feedback or input on the build. I have scratched my head with this for too long so that I may have tunnelvision.

• NZXT H630
• 4960X 4.5Ghz - 1.25V EKWB Evo Supremacy X99
• Asus Rampage IV Black Edition
• EVGA 980 Ti Classified - EKWB 780 Classified

Together with an Alphacool ST30 at the top and an Alphacool 360 Monsta in the front. Both of these radiators are running a push configuration with a total of six Scythe Gentle Typhoon 1450 rpm.
The pump is a D5 set to 3. The loop goes Res/Pump -> GPU -> ST30 -> CPU -> Monsta -> Res/Pump

My ambient temperature is in the 20-25C range. The processor stays relatively cool, with temps around 60-65C which was what I expected. But the graphics card on the other hand, gets really hot. Just above 80C during heavy gaming/load at 1456/8000.
I have double checked the fitting of the block and the fit seems good.

I can supply a picture of the build if needed.

 

[OneMoar]

I would say your gpu block isn't mounted correctly or the paste is dried
or something is plugging the flow though the block up

 

[Finners]

Air in the block or a bad mount I would say. Temperatures are defiantly to high ATM.

I picture might help if you have made an error that a second pair of eyes might see.

 

[Nokiron]

I would say your gpu block isn't mounted correctly or the paste is dried
or something is plugging the flow though the block up

Im going to remount the block this coming week and see if ill get better results. I can't think of anything else that could cause it, its a clear GPU-block so I cant see any obstruction.

After some though it might be a combination of the block not sitting correctly and the thermal paste not being very good. I used to have good results with IC Diamond 7, but it isnt very soft.

 

[OneMoar]

Im going to remount the block this coming week and see if ill get better results. I can't think of anything else that could cause it, its a clear GPU-block so I cant see any obstruction.

After some though it might be a combination of the block not sitting correctly and the thermal paste not being very good. I used to have good results with IC Diamond 7, but it isnt very soft.

IC diamond is the worst possible product you can use
order some MX4 and use that

 

[Nokiron]

IC diamond is the worst possible product you can use
order some MX4 and use that

Oh, how come? It's a free sample but it gets pretty decent results from reviews
http://overclocking.guide/thermal-p...d-with-air-cooling-and-liquid-nitrogen-ln2/6/

I have a lot of Noctua NT-H1 just lying around so ill probably use that.

 

[OneMoar]

Oh, how come? It's a free sample but it gets pretty decent results from reviews
http://overclocking.guide/thermal-p...d-with-air-cooling-and-liquid-nitrogen-ln2/6/

I have a lot of Noctua NT-H1 just lying around so ill probably use that.

http://remixedcat.blogspot.com/2013/07/innovation-cooling-ceo-throws-tantrum.html
http://www.hardwarebbq.com/ic-diamo...e-hand-thermal-pastes-damaging-cpu-gpu-cores/

 

[Nokiron]

http://remixedcat.blogspot.com/2013/07/innovation-cooling-ceo-throws-tantrum.html
http://www.hardwarebbq.com/ic-diamo...e-hand-thermal-pastes-damaging-cpu-gpu-cores/

Welp, glad I got it for free and didn't pay for it.

On the other hand, i'm not sure it would be the cause of my problem though. If anything it probably hasnt spread properly.

 

[Ebo]

IC7 diamond is a damn good paste, end of story.

Last year when I built my rig, I went for the Polimatech PK III instead, otherwise I still would have used IC7.

 

[Norton]

I have a lot of Noctua NT-H1 just lying around so ill probably use that.

That Noctua paste works very well- it's as good or better than MX4 imho.

 

[OneMoar]

That Noctua paste works very well- it's as good or better than MX4 imho.

agree but that ic7 stuff is terrible pure marketing hype and zero understanding of what paste does
thermal paste is intended above all else to fill in the imperfections between the IHS and the cooler and ideally provide a air-tight seal

it would be like putting sand in-between to plates of glass all it does is create gaps and scratch the glass up where as if you take those same to plates of glass and put oil between them they will vacuum seal and the oil will fill any surface imperfections
the ideal paste is something that is as fluid as possible and thermally conductive its why liquid metal pastes work so well its not the conductivity of the martial but how well it bonds

 

[Nokiron]

Ill post back with my results after a change of thermal paste and a reseat!

 

[Nokiron]

Well, the reseat worked. No idea why though, temps seems to have really dropped. Does this seem more realistic?

CPU at 60-65C at 4.5Ghz, no surprise there.

GPU has spiked to 67C once during Unigine Valley after ~45 minutes. Otherwise it stays at 55-60C

 

[RejZoR]

I'm running my i7 920 up to 80°C (coolant usually goes up to 50°C). I can't stand fan roaring. It has 5+ years of service behind. Why are people so obsessed with low temperatures?

 

[micropage7]

Well, the reseat worked. No idea why though, temps seems to have really dropped. Does this seem more realistic?

CPU at 60-65C at 4.5Ghz, no surprise there.

GPU has spiked to 67C once during Unigine Valley after ~45 minutes. Otherwise it stays at 55-60C

looks good, just test touch the heatsink, does it pretty hot?

or just run benchmark for several times then you got the average temp, btw thermal paste need curing time before it can show its full performance, just wait and see

 

[micropage7]

I'm running my i7 920 up to 80°C (coolant usually goes up to 50°C). I can't stand fan roaring. It has 5+ years of service behind. Why are people so obsessed with low temperatures?

maybe coz lower temp, said its better for your hardware
and its said over generations

 

[Nokiron]

I'm running my i7 920 up to 80°C (coolant usually goes up to 50°C). I can't stand fan roaring. It has 5+ years of service behind. Why are people so obsessed with low temperatures?

Mostly because it shouldn't be as high as it was with my setup

I like to see my components run cool and quiet.

looks good, just test touch the heatsink, does it pretty hot?

or just run benchmark for several times then you got the average temp, btw thermal paste need curing time before it can show its full performance, just wait and see

I will check when I return home from work later today!

 

[peche]

IC diamond is the worst possible product you can use
order some MX4 and use that

this ...!
+1 on Arctic MX4

agree but that ic7 stuff is terrible pure marketing hype and zero understanding of what paste does
thermal paste is intended above all else to fill in the imperfections between the IHS and the cooler and ideally provide a air-tight seal

it would be like putting sand in-between to plates of glass all it does is create gaps and scratch the glass up where as if you take those same to plates of glass and put oil between them they will vacuum seal and the oil will fill any surface imperfections
the ideal paste is something that is as fluid as possible and thermally conductive its why liquid metal pastes work so well its not the conductivity of the martial but how well it bonds

indeed a point here.... well explained,

I'm running my i7 920 up to 80°C (coolant usually goes up to 50°C). I can't stand fan roaring. It has 5+ years of service behind. Why are people so obsessed with low temperatures?

well i guess is because some people here is pretty enthusiast... i cannot said more...i love low temps...
by the way how do you know your coolant temp? you system specs says that you are running on a antec AIO,

Together with an Alphacool ST30 at the top and an Alphacool 360 Monsta in the front. Both of these radiators are running a push configuration with a total of six Scythe Gentle Typhoon 1450 rpm.
The pump is a D5 set to 3. The loop goes Res/Pump -> GPU -> ST30 -> CPU -> Monsta -> Res/Pump

take a watch ... my two cents...
flow direction or route: Res/Pump -> CPU -> GPU-> ST30 -> Monsta -> Res/Pump

 

[RejZoR]

The program for Antec 920 is running fans based on coolant temperature (it has a thermal sensor inside). OEM for Antec H2O 920 is Asetek.

 

[peche]

The program for Antec 920 is running fans based on coolant temperature (it has a thermal sensor inside). OEM for Antec H2O 920 is Asetek.

interesting !

 

[Bad Bad Bear]

I'm running my i7 920 up to 80°C (coolant usually goes up to 50°C). I can't stand fan roaring. It has 5+ years of service behind. Why are people so obsessed with low temperatures?

I don't understand why you'd make a completely facetious / redundant remark, posed as a question on a watercooling and advice thread ?

 

[tabascosauz]

@Nokiron I've had lots of good temps with NT-H1, not so much with MX-4. NT-H1 usually gets shat on by people who don't understand how paste spreads and the pea method, saying that "it's like spreading play-doh", but I find that it works better than MX-4. Not that I wouldn't use MX-4 for less-important PCs (I'm looking to get the 20g syringe for mass applications), but I'd use NT-H1 in such an important PC.

Also, maybe a little more paste would be good? I only put a small pea on my CPUs but on Tahiti I do a small X in the middle. My NT-H1 blobs usually covers an area far greater than the die but still not reaching the edge of the IHS; I think this is insufficient for GPUs, especially large ones (Tahiti/Hawaii/GK110/GM200).

 

[RejZoR]

I don't understand why you'd make a completely facetious / redundant remark, posed as a question on a watercooling and advice thread ?

And why are you questioning it? Water cooling is about balance and I've explained it from my example of a quad core which is rather hot one. You can either stick 2x 15k RPM Deltas on it and go blind (because the air flow will scoup your eyeballs out of your head) or you go with silence, something watercooling was initially used for.

As for the temperaqtures, usually things at the end of the loop are hotter which means there is something wrong with the graphic card part. I don't think it's flow itself, otherwise CPU would suffer as well. I'd inspect the GPU cold plate and everything surrounding that...

 

[Nokiron]

this ...!

take a watch ... my two cents...
flow direction or route: Res/Pump -> CPU -> GPU-> ST30 -> Monsta -> Res/Pump

Unfortunately, i don't really have the opportunity to do that. This was the simplest route I could take.
But I also seem to have found another culprit. From what I can tell, three Gentle Typhoons at 1450rpm isnt really enough to push the air through the Monsta. I added two NF-F12 (900 rpm) as a trial configured as pull (can't fit a third) and the temperatures dropped significantly. At full load my 980 Ti doesnt go above 55C, same with the CPU. So around a 5-10C drop in temperatures by just adding the fans.

@Nokiron I've had lots of good temps with NT-H1, not so much with MX-4. NT-H1 usually gets shat on by people who don't understand how paste spreads and the pea method, saying that "it's like spreading play-doh", but I find that it works better than MX-4. Not that I wouldn't use MX-4 for less-important PCs (I'm looking to get the 20g syringe for mass applications), but I'd use NT-H1 in such an important PC.

Also, maybe a little more paste would be good? I only put a small pea on my CPUs but on Tahiti I do a small X in the middle. My NT-H1 blobs usually covers an area far greater than the die but still not reaching the edge of the IHS; I think this is insufficient for GPUs, especially large ones (Tahiti/Hawaii/GK110/GM200).

I actually tried the X on my GPU this time aswell. Did a testmount and it seemed to have spread well.

 

[GreiverBlade]

well the 980ti is hotter than the 980 ... my poseidon with a simile-waterblock (if we can call a 6mm U-copperpipe a waterchannel) and 2 90mm fan idle @ 29° and load most of the time around 46°
same goes for the R9 290 who was previously in that loop, 30ish idle and never seen more than 48° load

tho ... i had my gpu loop separated from the cpu loop .... so for me a spike at 67° is not that bad.

as for past strangely enough i got the opposite, not so good result with Noctua NT-H1(and i found it overpriced and underperforming, well a constant with Noctua ... overpriced, not underperforming), better result with MX-4 (almost 6chf less than the NT-H1) and best result with my current Gelid GC-Extreme( around 3-4chf less than the NT-H1 but it perform a tad better than the MX-4 so the 3 to 2 chf are justified ) i think i gonna order some Thermal Grizzly Kryonaut to test ... next, cost a lot more but the syringe is 1.5ml (which equale to 5.55g ) and if it perform well ...

the block application is primordial also, but i think you already did check it, as for the path i can't see why you can't do Res/Pump -> CPU -> GPU-> ST30 -> Monsta -> Res/Pump (tho for me it's more res->pump->rad->gpu-block->res, but has i already pointed it i have a separate dual loop )

And why are you questioning it? Water cooling is about balance and I've explained it from my example of a quad core which is rather hot one. You can either stick 2x 15k RPM Deltas on it and go blind (because the air flow will scoup your eyeballs out of your head) or you go with silence, something watercooling was initially used for.

As for the temperaqtures, usually things at the end of the loop are hotter which means there is something wrong with the graphic card part. I don't think it's flow itself, otherwise CPU would suffer as well. I'd inspect the GPU cold plate and everything surrounding that...

well your example explained nothing, it was a statement with a question out of place "why people are so obsessed with low temp"
also neat info: your aio has around 15-30ml of Propylene/ethylene glycol in it as a coolant, not really comparable to a custom loop or even a Diy aio like a Swiftech H220-X or a Triton (with water based coolant)

i always found that pics funny (well after i went for a custom loop and a Triton)