Newbie help

[Flowroro]

Hi, i just put together my loop with a 240+360 rad, with cpu and gpu blocks. However in unigine valley/in games im getting 62+ degrees c on my gpu? pretty sure i put in all the screws correctly, i tried to retighten them to the point where i nearly stripped a screw, used an acceptable amount of thermal paste, put thermal pads in the right place, every tube is connected from out port to in port on each part of the loop, could be air bubbles in rad but i dont think that makes a 10+ degree difference? ive flipped and tossed my pc around so no huge bubbles left, waterblock for gpu looks like it has a lot of mini bubbles, would that disappear over time? i also dont think that it could make a 10+ degree difference. dont want to really drain the whole loop as i didnt make a way to do that =/ could it just be an unlucky airpocket in thermal paste or something else? any tips appreciated. cpu is an 8700k, changed from aio to loop, basically same temperatures on cpu as on aio, gpu is basically the same as air cooled rn. not sure whats going on. my flow setup is pump/res > 360 front rad > cpu block > 240 top rad > gpu > pump/res. Idle temps are good, cpu at 32c gpu at 28c/ ambient is around 14c

[wildgg]

what GPU?

 

also a list of specs will help us, to help you out faster.

[For Science!]

 

7 minutes ago, Flowroro said:

 

Also good to know if you have a temp sensor is the fluid temperature. Important to know if the fluid temperature is consistent with the GPU temp (i.e. loop cooling capacity issue) or if there is a huge difference between the fluid temp and the GPU core, this may point towards bad installation of the block. With a 240+360 I would anticipate it is the latter, but more info on radiator thickness, GPU, blah blah, would be helpful.

[chen57]

14c ambient and cpu is at 32c... thats a little high. i have a 360 and a thick 280 my room temp is 20-24 and my 8700k idle at 28-30. i should delid soon, cant wait to see the results

[Flowroro]

ekwb 240 + 360 rads, spc60 pump. Both rads 27-28mm thick, slim series, fitting all this in a corsair 570x was a bad idea.

8700k delidded running 5.0ghz @ 1.37v

rtx 2070 no oc yet, gpu iddles 27-28

cpu idles 30-32

 

under load: cpu 55-65c, gpu 62-65c. both have waterblocks, im going to reseat the gpu tomorrow and see how it goes if theres no other option. May have also cut the thermal pads a tad too wide but as long as it doesnt add thickness it should be fine i hope? ill cut it by a few mm anyway just to be safe. will update after reseat

[For Science!]

11 minutes ago, Flowroro said:

 

A picture of your thermal paste spread on the GPU die may be indicative, since it is a direct die contact you usually benefit from going a bit overboard with the thermal paste. I prefer to either spread it manually for thick compounds like kryonaut, or a generous X and + in the old EKWB manual style. Provided your compound is non-conductive there is little disadvantage of too much paste for GPU waterblock except for wasting the compound itself.

 

Here are a few pics of my successful installations of GPU waterblocks.

 

Kryonaut, pre-mount

 

MX-4, Pre Mount

 

MX-4, Post Mount (Die Side)

 

MX-4, Post Mount (Block Side)

[Flowroro]

Is it possible that I messed up the tubing direction? theres always this huge bubble in my gpu to pump tube that i have to shake it to get out. im pretty sure its all correct.

[Flowroro]

10 hours ago, For Science! said:

A picture of your thermal paste spread on the GPU die may be indicative, since it is a direct die contact you usually benefit from going a bit overboard with the thermal paste. I prefer to either spread it manually for thick compounds like kryonaut, or a generous X and + in the old EKWB manual style. Provided your compound is non-conductive there is little disadvantage of too much paste for GPU waterblock except for wasting the compound itself.

 

Here are a few pics of my successful installations of GPU waterblocks.

 

Kryonaut, pre-mount

 

MX-4, Pre Mount

 

MX-4, Post Mount (Die Side)

 

MX-4, Post Mount (Block Side)

so i took off the block, i applied a generous amount of paste, the stock one that comes with the block from ek, rescrewed everything, cut the thermal pads smaller in width so it wouldnt add any kind of leverage, screwed it all in hand tight, no improvement in temps.

[Flowroro]

This is how my loops goes, did I do it wrong? gpu temps sitting at 59-60c on unigine valley, cpu sitting at 50-59c on unigine at the same time.

[Flowroro]

12 hours ago, wildthing said:

what GPU?

 

also a list of specs will help us, to help you out faster.

 

12 hours ago, For Science! said:

 

Also good to know if you have a temp sensor is the fluid temperature. Important to know if the fluid temperature is consistent with the GPU temp (i.e. loop cooling capacity issue) or if there is a huge difference between the fluid temp and the GPU core, this may point towards bad installation of the block. With a 240+360 I would anticipate it is the latter, but more info on radiator thickness, GPU, blah blah, would be helpful.

after leaving on cinebench and valley for a while, i touch my rads and they arent even warm/hot. when i used my evga clc 280 on my cpu it got super hot, wondering if theres a flow issue in my loop?

[Flowroro]

issues/potential issues right now:

1. i couldve put too thick of a thermal pad increasing leverage and making poor contact.

2. cpu increased in temps to 65-70 while gaming for less than an hour

3. gpu still sitting at 60c and now the whole system is louder than when my gpu was on air and cpu had an aio.

4. cpu temps were better on an aio, a 240+360 rad yielding these results is ridiculous. gpu is basically aircooled right now with a spare 240mm radiator doing nothing

 

so in conclusion im gonna have to try and use the vram thickness thermal pads to replace some hoping it fixes the issue, or i couldve simply created a conflicting flow according to my picture, im not sure, block is visually inspected to be compatible with my gpu pcb on ekwb site, im absolutely lost, why are my temps soooo bad? i just threw a chunk of money down the drain for worse performance lol.pump is set to max, not sure if liquid is even moving, from what i can see pinky nail sized bubbles are trapped in the gpu block and not coming out no matter how hard i shake it. whenever i put my pc flat down though the liquid from the gpu bottom side will lose liquid, and when i turn it on again there is there strange space where its not really flowing, like water flowing on the tubing surface but hollow in the middle until i shake it somewhat again and it fills up. any thoughts? getting a headache here.

[For Science!]

2 hours ago, Flowroro said:

 

Loop looks decent. Can't immediately see anything catastrophic.  You should try a CPU only stress test and report back the cpu and gpu temps (this will give me an idea about fluid temp).

 

Barring this, how are your fans setup in terms of fan curve. Are they actuallt set to speed up as the water heats up? Otherwise presumably they are tied to the CPU temp.

[Flowroro]

48 minutes ago, For Science! said:

Loop looks decent. Can't immediately see anything catastrophic.  You should try a CPU only stress test and report back the cpu and gpu temps (this will give me an idea about fluid temp).

 

Barring this, how are your fans setup in terms of fan curve. Are they actuallt set to speed up as the water heats up? Otherwise presumably they are tied to the CPU temp.

theyre set to 100% currently, and auto optimized in bios before. fans are connected 3 into 1 fan header. s is it a bad idea to dip a thermometer rinsed with distilled water first to check fluid temp? started prime95 small fft cpu starts at 58-58c and gpu starts at 28. (I reset all overclocks and am just running base to compare temps as well). I havent run it for long but gpu looks like its slowing gaining 1c every 5mins-10 mins or so?

 

Edit: its been 10 mins running prime 95 now, cpu went up to 61c, gpu staying at 29.

[For Science!]

1 minute ago, Flowroro said:

theyre set to 100% currently, and auto optimized in bios before. fans are connected 3 into 1 fan header. s is it a bad idea to dip a thermometer rinsed with distilled water first to check fluid temp? started prime95 small fft cpu starts at 58-58c and gpu starts at 28. (I reset all overclocks and am just running base to compare temps as well). I havent run it for long but gpu looks like its slowing gaining 1c every 5mins-10 mins or so?

This is fine, I am therefore going to assume your idle water temperature is 28 degrees. Let it climb until the CPU/GPU temp is stablized. Given you have probably less than 1 liter of fluid, it should take between 30-60 minutes. 

 

Also, what was your GPU cooler model (i.e. what is it a founders edition? a Strix? a FTW3).

[Flowroro]

Just now, For Science! said:

This is fine, I am therefore going to assume your idle water temperature is 28 degrees. Let it climb until the CPU/GPU temp is stablized. Given you have probably less than 1 liter of fluid, it should take between 30-60 minutes. 

 

Also, what was your GPU cooler model (i.e. what is it a founders edition? a Strix? a FTW3).

it was an asus dual oc. ive used about 600ml of coolant mixture i think

[Flowroro]

39 minutes ago, For Science! said:

This is fine, I am therefore going to assume your idle water temperature is 28 degrees. Let it climb until the CPU/GPU temp is stablized. Given you have probably less than 1 liter of fluid, it should take between 30-60 minutes. 

 

Also, what was your GPU cooler model (i.e. what is it a founders edition? a Strix? a FTW3).

been 50 mins and cpu stayed at 60-61c, gpu 30-31c.

[Flowroro]

how is it possible that people with just a 240 rad cooling a cpu and gtx 1080 gpu can get way lower temps than me with an extra 360 rad. somethings wrong, if my tubing/flow looks fine, then it could only be the gpu block, it was getting good contact on the block too, my last option is to remove the thermal pads off the highest component and reseat it once more, or block is the problem, or pump is the problem, what else could it be?

[For Science!]

9 minutes ago, Flowroro said:

-

Can you confirm the model of your GPU waterblock?

[Flowroro]

.

[Flowroro]

14 minutes ago, For Science! said:

Can you confirm the model of your GPU waterblock?

EK-AC GeForce RTX

https://www.ekfluidgaming.com/ek-ac-geforce-rtx

 

was i supposed to put 1mm thick thermal pads on the chokes? could that be causing the issue?

[For Science!]

Just now, Flowroro said:

 

0.5 mm for the memory, 1 mm for everything else.

https://www.ekfluidgaming.com/EK-IM/EK-IM-3831109817049.pdf

 

While this may not directly impact core temperatures, it is pretty critical that your power delivery is correctly cooled.

[Flowroro]

13 minutes ago, For Science! said:

0.5 mm for the memory, 1 mm for everything else.

https://www.ekfluidgaming.com/EK-IM/EK-IM-3831109817049.pdf

 

While this may not directly impact core temperatures, it is pretty critical that your power delivery is correctly cooled.

yea ive followed each instruction and re-read it two/three times. used the right screws, used the right thickness of thermal pads, it only came with precut 0.5 for vram and 1mm uncut for others. what other steps should i try next to resolve this issue

[For Science!]

4 minutes ago, Flowroro said:

yea ive followed each instruction and re-read it two/three times. used the right screws, used the right thickness of thermal pads, it only came with precut 0.5 for vram and 1mm uncut for others. what other steps should i try next to resolve this issue

One picture of your GPU PCB + Pads would be the only other thing I can look at to give my opinion (i.e. double check all the pads are where they are suppose to be).

 

Other than that, I think beyond changing the pump for a D5 (which I do not think will change the temperatures). This may just be the limit of the aluminium system. It is clear to me that your loop is functional to a degree since  neither your CPU or GPU are anywhere close to thermal throttling. Indicating that there is some transfer between the heat generating compound and block to fluid. During a CPU only load, you fluid temperatures are in the low 30's (in this case, the GPU temp acts like a thermal sensor) indicating that the cooling capacity of the loop is also fine.

[Flowroro]

3 minutes ago, For Science! said:

One picture of your GPU PCB + Pads would be the only other thing I can look at to give my opinion (i.e. double check all the pads are where they are suppose to be).

 

Other than that, I think beyond changing the pump for a D5 (which I do not think will change the temperatures). This may just be the limit of the aluminium system. It is clear to me that your loop is functional to a degree since  neither your CPU or GPU are anywhere close to thermal throttling. Indicating that there is some transfer between the heat generating compound and block to fluid. During a CPU only load, you fluid temperatures are in the low 30's (in this case, the GPU temp acts like a thermal sensor) indicating that the cooling capacity of the loop is also fine.

that sounds valid however how are all the big youtubers who use just a single 240 rad for gpu and cpu get significantly lower temps though? yes they have different gpus and cpus but their heat output would nearly be the same, even with an extra 360 rad i shouldnt get higher temps than a single 240 cooling two units right?

[For Science!]

32 minutes ago, Flowroro said:

-

I have never felt the need to match some big youtubers as I have found some of them to consistently have suspiciously low temperatures in the context of showing off their builds and have stopped consuming their content.

 

But yes, more radiator space would normally be associated with approaching closer towards the ambient temperature, which in your case is particularly cold (I hope you wear a jumper in your room). CPU temperatures of 20-30 degrees (and also higher, for high OCs) above the fluid temperature (which we determined to be about 28-30 degrees) during load is not particularly unusual in my opinion, For GPU this is more often like 15-25 in my experience and so yes, your GPU temperatures do seem a little bit high given that it is also a 2070, a relatively efficient and cool card. Granted, I am not well versed in aluminium kits and so there may be some penatly incurred for not being copper based, but I understand this to be a couple degrees at most.

 

One should also remember that these RTX cards come with very beefy heatsinks (2.5 slots in the case of the Dual OC). Heat dissipation is pretty much proportional to mass and surface area and so it is entirely possible that your 360 mm radiator in mass terms is equivalent to the air cooler that you ripped off. Which could be why the temperatures are similar.

 

Also one thing to note is whether the clocks/power draw are actually the same between your comparison of air and water, my cards hold a much steadier clock when watercooled when compared to air. Lastly, it could be just down to silicon quality. In a previous build I did I watercooled 4x 1080Ti's in the same loop, same blocks, same installation method and one card was 5 degrees hotter than the rest consistently (blue line).