Extremely high load temperatures with custom loop

[Philz69]

Hey guys!

I have recently finished my first watercooling loop and have the CPU and GPU running in parralel, however, even though im having >30c idle temps, i managed to get up to 95c on my CPU in prime 95!

I normally would've just remounted the cooler, but its out of question right now since i'd have to drain the loop and that is out of question, at least for this week! Could it be the pump failing to drive enough flow?

Is it possible that i have an air bubble in the cpu block? Is my CPU block so not restrictive it bassicaly receives no flow? 

I need your help guys!

[JebKerman]

don't use prime 95, use something like intelburntest, or Aida64. but what are your tech specs and water cooling specs (nvm i looked it up)

 

Ok, then it shouldn't be a lack of radiator or anything directly to do with the loop (I assume the pump is maxed out).

[Arcanekitten]

Wow that ain't right 

i would look if the flow is correct (speed and stuff) and the pump is on its max pwm

also don't use prime95 atleast not the more recent versions but aida64

[Br3tt96]

Hey guys!

I have recently finished my first watercooling loop and have the CPU and GPU running in parralel, however, even though im having >30c idle temps, i managed to get up to 95c on my CPU in prime 95!

I normally would've just remounted the cooler, but its out of question right now since i'd have to drain the loop and that is out of question, at least for this week! Could it be the pump failing to drive enough flow?

Is it possible that i have an air bubble in the cpu block? Is my CPU block so not restrictive it bassicaly receives no flow?

I need your help guys!12170616_1038433959522318_478706653_n.jpg

Your signature says world's fastest 4790k. Define world's?

[FilipSebik]

Your signature says world's fastest 4790k. Define world's?

Linus had 5GHz 4790K, GTX 980 and he is supposed to have fastest 4790K in world like you (not br3tt96)

[Br3tt96]

Linus had 5GHz 4790K, GTX 980 and he is supposed to have fastest 4790K in world like you (not br3tt96)

I could get 5ghz, but I'm not gonna get a fucking sponsor to get a phase cooler like he did. So in a way Linus' 4790k isn't Evan fair, and they're plenty of 4790ks that are faster than mine and his.

[Philz69]

I could get 5ghz, but I'm not gonna get a fucking sponsor to get a phase cooler like he did. So in a way Linus' 4790k isn't Evan fair, and they're plenty of 4790ks that are faster than mine and his.

You didnt read it correctly, im not saying that i have the world's fastest 4790k, im not saying i have the world fastest 780s, im saying i have the world fastest combination of both!

Anyway, its not even a record anymore, even though the previous score was like 1 year old, a guy decided to beat me one day after i achieved #1 ><

[Philz69]

don't use prime 95, use something like intelburntest, or Aida64. but what are your tech specs and water cooling specs (nvm i looked it up)

 

Ok, then it shouldn't be a lack of radiator or anything directly to do with the loop (I assume the pump is maxed out).

That did not answer my question at all its not normal to reach temps of 85c in 3dmark either! 

 

Wow that ain't right 

i would look if the flow is correct (speed and stuff) and the pump is on its max pwm

also don't use prime95 atleast not the more recent versions but aida64

The flow isnt really seeable sadly, but the pump is maxed out! ( Also temps have nothing to do with me using prime95, i know it produces unreal temps, but it still shouldnt reach that much and reaches 85c peaks in 3dmark anyway!

 

I thought it could be a bubble lodged in the top of the cpu block, could that affect temps that much?

[W-L]

That did not answer my question at all its not normal to reach temps of 85c in 3dmark either! 

The flow isnt really seeable sadly, but the pump is maxed out! ( Also temps have nothing to do with me using prime95, i know it produces unreal temps, but it still shouldnt reach that much and reaches 85c peaks in 3dmark anyway!

I thought it could be a bubble lodged in the top of the cpu block, could that affect temps that much?

 

If you just assembled the loop give it a few days to bleed and shake it around to make sure no bubbles are trapped in the black that can heavily affect temps. Having everything in parallel isn't a problem when the blocks have very similar restriction but said prime 95 gives unrealistic load testing so Adia64 or Inel XTU is a better option.

[Philz69]

If you just assembled the loop give it a few days to bleed and shake it around to make sure no bubbles are trapped in the black that can heavily affect temps. Having everything in parallel isn't a problem when the blocks have very similar restriction but said prime 95 gives unrealistic load testing so Adia64 or Inel XTU is a better option.

Im going to turn the case and make the bubbles ( if there is any ) follow the path to the res!

[Philz69]

Can confirm there is bubbles, but i really dont seem to be able to get them out, i moved the case all arround, but when i started the pump, then went right back into the cpu block! ( saw a couple of them go into it ) 

EDIT: Been dancing with the case for a few hours and i seem to have gotten quite a bit of bubbles out!

[Dango]

Why parallel you CPU and GPU block?

[Philz69]

Why parallel you CPU and GPU block?

Because style?

[airdeano]

you have got monster air trapped in that system. with the res mounted lower

than the CPU (highest point) the trapped air will collect in the CPU and never

bleed out. you'll have to lay the PC on its back (with a full reservoir to prevent

cavitation) to allow the air to escape and replenish with coolant. and then on

its i/o to make the res the highest point.

[Philz69]

you have got monster air trapped in that system. with the res mounted lower

than the CPU (highest point) the trapped air will collect in the CPU and never

bleed out. you'll have to lay the PC on its back (with a full reservoir to prevent

cavitation) to allow the air to escape and replenish with coolant. and then on

its i/o to make the res the highest point.

Thats exactly what i've been doing, no matter what i do, the bubble need to go through the pump to get to the res, so its REALLY hard to get them through, most of the time, the pump pushes the bubble back in, then i get them out the other way, then i restart! I'd need to have a port on the top of the cpu block to actually bleed it correctly

[OddsCrazyStuff]

I would check out this thread: http://www.overclock.net/t/1499542/help-with-water-cooling-full-parallel-loop

But the main point is that parallel requires much more thoughts on blocks then serial, since if 1 block got more restriction, it suffers.

[Dango]

Because style?

When you parallel your CPU and GPU blocks which have different restriction. The water will go where have the lest resistance pass. That is probably why you are having high temp under load.

[Philz69]

When you parallel your CPU and GPU blocks which have different restriction. The water will go where have the lest resistance pass. That is probably why you are having high temp under load.

 

I am conscious of that and mentionned it in the first post as an option, however, i doubt that, even if it recieved 1/4th of the flow, it wouldnt be enough to cool it properly! But im curious as to how a cpu block, wich is much smaller and has less cooling area so probably less fins, could be more restrictive then the gpus?

[OddsCrazyStuff]

I am conscious of that and mentionned it in the first post as an option, however, i doubt that, even if it recieved 1/4th of the flow, it wouldnt be enough to cool it properly! But im curious as to how a cpu block, wich is much smaller and has less cooling area so probably less fins, could be more restrictive then the gpus?

Its as simple as this: the smaller the surface, the less time the heat have to transfer.

So restriction matters a lot more on the CPU then the GPU.

[Philz69]

Its as simple as this: the smaller the surface, the less time the heat have to transfer.

So restriction matters a lot more on the CPU then the GPU.

I red through all of the post you linked, seems like im out of luck, i will try remounting next time i drain, if that doesnt help, im going to look into adding another d5 in series!

[0ld_Chicken]

I red through all of the post you linked, seems like im out of luck, i will try remounting next time i drain, if that doesnt help, im going to look into adding another d5 in series!

did you disassemble the cpu block before installing? There is a piece in the ek supremacy (i forget the name of it, not the jet plate) that can be removed to increase the flow. I saw the owner of Mayhems coolant talking about it in reference to increasing flow rate for aurora. He said removing it will increase cpu temps by no more than a few °c (if it's already optimal temps) but in your case if it helps balance the flow then you would see better temps i think. I'm at work but when i get home I'll figure out exactly what it is in my install manual and link you more info

[BrightCandle]

You have two GPUs in the loop, what are their temperatures? Also what is the temperature of the water?

 

Given these other data points it should become clear whether its not enough heat being removed from the loop (the water is hot), a problem with flow over the CPUs (the water is normal and the GPUs are OK) a general heat problem (everything is hot)?

 

More data and then the cause will present itself.

[Philz69]

You have two GPUs in the loop, what are their temperatures? Also what is the temperature of the water?

Given these other data points it should become clear whether its not enough heat being removed from the loop (the water is hot), a problem with flow over the CPUs (the water is normal and the GPUs are OK) a general heat problem (everything is hot)?

More data and then the cause will present itself.

45c MAX on the gpus at 1.35v

I have no water temp sensor though!

The gpu temperatures dont really seem to rise when i put the cpu to load! That could mean either the temp doesnt get into the water ( mount problem ) or the water as so low flow the quanitity of heated water vs not heated water isnt affecting the gpus! ( That was a really stupid statement, they dont change at all, even in a serial path, was a bit disctracted. ignore this) 

[W-L]

45c MAX on the gpus at 1.35v

I have no water temp sensor though!

The gpu temperatures dont really seem to rise when i put the cpu to load! That could mean either the temp doesnt get into the water ( mount problem ) or the water as so low flow the quanitity of heated water vs not heated water isnt affecting the gpus!

 

The GPU temps shouldn't really rise if you put the CPU to load, going on the basis if everything is well balanced and flow is fairly even since the fluid moves very quickly throughout the loop and not to mention being in parallel. Are you still getting fairly high idle and load temps on the CPUeven after getting rid of the bubbles stuck in the CPU block?

[Philz69]

The GPU temps shouldn't really rise if you put the CPU to load, going on the basis if everything is well balanced and flow is fairly even since the fluid moves very quickly throughout the loop and not to mention being in parallel. Are you still getting fairly high idle and load temps on the CPUeven after getting rid of the bubbles stuck in the CPU block?

I never was getting high idle temps! I only have been getting hih load temps, im at 4.8ghz right now and getting 60ish in aida64, its still pretty high for a system with a 420mm and 360mm radiator