Custom loop i7-7700k Thermal Issues

Toggle sidebar Toggle sidebar
M

matyounatan

Reputable
Feb 21, 2014
21
0
4,510
Hello,

Ever since I built my new gaming rig I've been having thermal issues with my processor. I'm running a 1080Ti EK waterblock and a Z270E Monoblock in a custom loop with the radiator mounted at the top of my case.

Picture of the monoblock: http://

My issue is that when playing games like Elder Scrolls Online or running synthetic benchmarks, after about 10-20 minutes my CPU temps can rise all the way to 85C+!! One time hitting 90C! I know these are not normal temperatures especially for watercooling and having no overclock.

This doesn't make sense because at idle my temps stay around 31-33C but the moment I open an application like Adobe Premiere they spike to ~56C and then when I'm gaming they hit around 80-85C

Things I've tried:
Flush the loop and clean the block with distilled water
Reapplied the thermal paste
Set my voltage manually to 1.200

And I still hit ridiculously high temperatures, I'm at a loss. Any insight would be helpful.
 
Sort by date Sort by votes
Radiator should be at least a 280mm (140mm X 2 fans) or larger.

rule of thumb
stock cpu to mild OC 240mm of radiator surface area
mild OC to high OC 360mm of radiator surface area

Adding extra blocks ( motherboard or GPU ) each block add 120mm of surface area.

In your case a CPU + GPU you should have at least 360mm of radiator surface area.
 
Upvote 0 Downvote


360 rad top mounted



I used a cooler master T2 to test my system for 5 minutes before I installed the blocks on it. I didn't have windows or anything but idle temps in bios under air cooling (if I remember correctly, it's been about 3 weeks since then) seemed fine to me.

I'm also running heaven benchmark right now, after 13 minutes GPU temps seem to be sitting at 44-45C under full load. So my GPU waterblock works. Also note that my CPU tempts spiked to 73 while running a GPU benchmark (and having chrome open, but really that shouldn't spike it to 73).

Motherboard temps are always around 30-35C for me.
 
Upvote 0 Downvote
Also what type of radiator do you have? Is it a high fin density or low fin density? Your cooling could be affected by a combination of wrong fan on the wrong type of radiator.

You should be running a static pressure fan on a high density fin radiator, and a high air flow fan for a low density fin radiator.
 
Upvote -1 Downvote
You are not the only one having this issue. Its all normal for the 7700K
Well most people are saying that the 7700k/ kaby lake cpus run hot due to poor thermal paste under the IHS. If you are willing to take the risk of delidding your cpu, you could lower the temps by 15-20c. Well The paste only plays a very small part... it's bad but not that bad. The issue is that the adhesive silicone used to keep the IHS in place is too thick and raises the IHS off the die too much. Without good contact you have higher temps. Delidding allows you to remove the adhesive completely and/or apply a very thin layer of your own which vastly improves contact between the two, especially for Kaby. But the temps that you are reporting are a little bit higher than the rest. And that's because not all CPU's are the same. And It is massive, but it's because stock Kaby's have a thicker layer of adhesive on the IHS causing poor contact between it and the die. Once you delid and remove that silicone, ditch the stock paste for liquid metal tim, then you're seeing these temp drops. Some people reporting as high as a 26° difference.

These are normal temperatures

80+C Hot (100% Load)
75+C Warm
70+C Warm (Heavy Load)
60+C Norm
50+C Norm (Medium Load)
40+C Norm
30+C Cool (Idle)
25+C Cool
 
Upvote 0 Downvote


I have an EK SE radiator, with 3 Corsair HD120 fans on them all in PWM using a silverstone 1-8 PWM fan hub.


The only reason this doesn't seem normal to me is because my temps go above and beyond 85 C during gaming which shouldn't be happening during gaming.

I would definitely de-lid my CPU but I don't want to void my warranty.

I'm afraid that one day my temps will creep to 100C without my knowledge during gaming (they've already reached about ~90-95C on the package, 93 on most individual cores one time)
 
Upvote 0 Downvote


right port is always marked as inlet in the manual so that's what I'm using.

Pump i'm using is a Laing DDC 3.2 PWM 18W Pump ( MCP35X )
 
Upvote 0 Downvote
Ok, wanted to make sure you had the proper inlet.

You might be a little light on that radiator with the SE. At least a PE would be better, or an XE ideally for your setup especially if you're overclocking. You should have enough pressure from that pump (unless its bad....) to keep the liquid flowing over your CPU quickly enough to get its heat out of there.

My concern, especially considering when only your GPU is under load how its making your CPU temps rocket, is your loop order. I'm guessing here but are you going: Pump -> GPU -> CPU -> Radiator -> Res? You may want to try going Pump -> CPU -> GPU -> Radiator -> Res. Keep in mind, best practice is to have your radiator before your res. Good chance you'll see a difference in CPU temps and while it may make your GPU temps climb it sounds like you've got a lot of room to go there with it only running in the 40s now.

I'd still look to upgrade your radiator to the PE or XE, or add another radiator in your loop (between GPU and CPU) if possible to see if this helps.

I built an EK loop recently with 4 x 1080 ti's (not a gaming rig...) and a 6800K (no OC). The app on this machine in particular ramps all four GPUs to 100% for weeks at a time. Loop goes: Pump -> GPUs (semi-parallel bridge) -> 280mm rad -> CPU -> 280mm rad -> Res. With all four GPUs at full load AND running Prime 95 on the CPU, GPU temps only hit 60c and CPU hovers around around 60c. Those radiators make all the difference, and my fans (ML140 pros) never went over 1200RPM. You've got good fans so I wouldn't sweat that aspect.

Yeah, you could have lost the silicon lottery but if at stock speeds you're still reaching up in to the 70s something's wrong....
 
Upvote 0 Downvote
Yeah I even unplugged the GPU completely and I still saw CPU temps reach the 75+ range easily.. I may grab a warranty claim on the CPU see if getting a new one would help but for now that's my last option.
 
Upvote 0 Downvote
UPDATE

I've contacted EKWB and I'm trying to diagnose it with them and maybe get a new monoblock..

What I've done to try and keep the temps from rising to 90c is the following (spoiler alert, they still do 🙁 )

- remove monoblock and reapply paste (again)
- reconfigure fans so they pull air into case and on the rad, then pull the air out from the back of the case (pc-o11 rog edition)

Aaaaaand the temps still reach 90C while gaming (fallout 4, battlefield 1)
In comparison I've seen my GPU hit 50C at max load.

I've also noticed that none of the liquid in my loop is cold (not even the liquid exiting the radiator) so my thoughts are that maybe my rad isnt thick enough (EK SE 360) but it is definitely long enough to cool only two components.
Or it could be my DDC pump (although I have it on full blast right now and the temps still rise to 90C)

Any thoughts would be appreciated.
 
Upvote 0 Downvote
The coolant will never be 'cold' in a loop, it's constantly moving and being heated by the gpu and cpu. Best case scenario the coolant can be ambient (room temp). Right now my ambient room temps are 31c (88F) meaning with the system off mine wouldn't be cold regardless. More like tepid bath water. Of course your ambient temps will probably vary. Once the loop has been running awhile the coolant will generally stabilize, x amount of heat from the cpu and gpu, x amount of heat dissipated from the radiator. It will be less warm than the components it's cooling but will still be warm.

Just to be certain, the coolant flows one way through the entire system right? Meaning it comes out of the gpu or cpu and into the other, the pump on one side and the radiator on the other? Just making sure that when you tied them together you didn't somehow end up with a gpu loop and splice in tubing for the cpu creating a secondary loop that runs off the same radiator/pump. Only speculating since there's no photos to go by. Trying to make sure you don't have a main loop cooling the gpu (since that seems to be working properly) and adding the cpu didn't create a weak/dead flow pattern.

Temp spikes within safe temp ranges aren't abnormal, the sensors are near the cores and they can jump temp instantly. By the time the heat makes it through the tim, the ihs, absorbed through the baseplate of the waterblock and absorbed into the water the cpu could have spiked from 30c to 60c and back to 35c before the cooling system really comes into play. It's not like a stove or car engine that works up to temperature and slowly the cooling system heats up with it. The coolant will heat up after awhile but the cpu is immediate, if the load jumps the temps are already there. More like a torch where the heat goes from off to on.
 
Upvote 0 Downvote


True that makes sense, I was worried that hot coolant meant thats why the temps were bad.

I was just playing battlefield 1 and my temps averaged around 88C, hitting 93C as max..

imgur link

Here is a picture of my loop. Im only worried because while playing fallout 4 my fans get REALLY loud, and periodically I check the current CPU temps using HWMonitor it always reports a new maximum CPU package temp of around 90-95C (especially when loading areas).
 
Upvote 0 Downvote
Well it's all definitely going one way. I agree with marko, I'd probably try to go cpu first then gpu. It could be you just don't have enough radiator. There's a lot of surface area on the gpu where the coolant is picking up heat from the gpu itself plus the vram. Now the coolest water in your loop in the reservoir has just picked up all the heat from the gpu and is now trying to absorb heat from the cpu. The cpu isn't getting cooled by whatever's in the reservoir, it's getting cooled with the gpu's 'waste water' (as in heat waste). Being a smaller block the cpu needs all the help it can get.

I don't blame you for not wanting to delid and void the warranty. It may still be an option but one I'd personally put off as a last resort. If you can maybe try leaving the gpu on the cooling loop you've got setup and try running an air cooler on the i7. If temps don't change then it could be the cpu itself, the tim under the ihs plate and a delid is in order. If temps do improve even slightly then it's time to look at the loop.

It could be not enough radiator surface/fin density for everything you're trying to cool. It could also be the water block and how it contacts the cpu. There's a balancing act between too much mounting pressure potentially bending the cpu (new models having thinner pcb substrates than 4th gen and earlier) and not enough pressure. Some water cooled setups in the past had to add washers to the mounting brackets to try and increase pressure between the water block and cpu for good thermal transfer.

Does the air coming out of the radiator feel rather warm? Is the air being pulled in through the top of the case through the radiator and vented into the case? If so it looks like warm air would be blowing right against that tube taking water back to the reservoir. Maybe I'm seeing it wrong from the angle of the pic. Is there a fan at the rear of the case exhausting the warm air (assuming the exhaust from the radiator is blowing into the case)?

You're right to be concerned, you're nearly at thermal throttle temps for the cpu. Especially when you're only gaming.
 
Upvote 0 Downvote
Question: on your loop, were it goes from the GPU out to the CPU in is it me or does the tube right were it bends right angle into the CPU block have a kink on the tube ? hard tube or soft ?
Here : http://imgur.com/a/j1iQh
Stupid Question : what liquid you using ?
Another Question: Can you 100% confirm the water is being pumped around the system ? like drain some water to make sure its still being pumped bk into the res >
I know these sound stupid but I have to ask
 
Upvote 0 Downvote
The three fans at the top are pulling air into the case. The subsequent air is definitely hot, but I don't think it's hot enough to make a big difference in the temperature of the water when it hits the hardtubing (at least not enough to push the CPU to 90-95.

The three rear fans are pulling air out of the case. I may consider adding another rad behind the three fans and reorganize the loop like

RES/PUMP>GPU>RAD1>CPU>RAD2>RES/PUMP

but thats another $150-200 (im in canada, shipping and US to CAD conversion is a bitch) for a new rad, fittings and tubing (id have to re do most of the original tubing).

My last option would be de-liding but it seems like its the only option as of now.

EKWB told me to remove the thermal pads on the inductor coils, and to leave the pads on the mofset chips in tact. I might try this sometime this week, but it takes me at least 2-3 hours to completely drain, take apart and fill my system (not to mention lifting this case, its heavy as shit lol) so if I can avoid removing the block for the fourth time then thats preferred haha.
 
Upvote 0 Downvote

I think its just the light reflection, before I installed the tubing there were no marks of kinking or anything like that. that bend is subpar but I doubt its enough to severely block flow.

Interesting observation though, I may order more tubing in a day or two and maybe if I rebend that tube itll fix my problem. If it doesnt well at this point failure is just another starting point...
 
Upvote 0 Downvote

yuou could get sopft tube to test it then see what happens, may not be needed to replace the hard tube if its no different with soft tube.
Can u get a top down pic of that bend ?
It my not restrict it much but then that depends how fast your pump is running
 
Upvote 0 Downvote
Additional food for thought, were each of the loop components thoroughly cleaned (especially the blocks and rad) before assembly? A clog somewhere may be preventing proper flow, compounded by water having to fight gravity all the way from the pump to the rad. If that gunk got caught up in the pump...

Speaking of pump, the MCP35X isn't that strong of a pump relative to many others on the market today. However, it *should* be able to handle a loop like yours competently.

When I was first assembling my loop, I did some experimentation, and managed to completely heatsoak an XSPC EX120 (slim 120 rad) with my mildly overclocked 4790k (4.6ghz, 1.3V). It took an EX240 just to properly cool the CPU alone. I see a lot of people recommending another 240mm of raddage for the 1080ti. If that's the case, like synphul said, maybe a slim 360 just isn't enough surface area, especially when you're adding in an even hotter CPU+VRM's. Combine that with the whole intel manufacturing/assembly process ordeal, geezus.
 
Upvote 0 Downvote


I rinsed every block with distilled water before installing (poured water in, shook it around a bit, dumped it out, repeated about 4-5 times)
same with the rad (that thing was dirty as hell).

The pump seems to be fine, every time I fill this system up the water shoots around really fast so i doubt theres something wrong with the DDC pump itself.

I also poured distilled water into the loop, let it run for about a day then emptied it, and then filled it with EK Clear fluid and now ive probably drained and filled the loop a total of 4 times. I don't see anything in the liquid when I pour it back in and nothing physical other than the water itself comes out of my drain.

The next time I remove my block I may take it apart to check within the channels themselves.
 
Upvote 0 Downvote
This is kinda crazy....

I've gotta ask a few questions throughout this.

I don't doubt you're a little light on the radiator FOR OVERCLOCKING. Thing is, your original post mentioned you're hitting upwards of 90c with no OC during synthetic benchmarks? First question: What benchmark tools are you using (and what version, especially for prime 95 if that's one of them) for CPU benching?

I can't imagine that 90c is normal, no matter where your CPU lands in the lottery, when under a synthetic stress tool like ROG realbench or Prime (v266), whether you've got a slim SE 360 radiator or not and the GPU is at idle. At a minimum that radiator should be fine to keep your CPU cool (when your GPU is idle) even with some overclock. Even if you run through the configurator on ekwb.com, punch in an X270 board and a 1080ti, it claims a 360 SE should cool OK with overclocking, and I tend to agree with the proper fans and a good pump.

Just for reference here, i just wrapped an X299 build with a 7800X clocked to 4.7. If you've read about these chips, they run HOT and believe me its true. However, even on a corsair H105 closed loop I had it at 4.6 and under prime 95 (266) and ROG realbench it would stay mid-low 70s. Prime 95 latest version would ramp it over 90c of course. I finished up a custom EK loop in the same build along with a single 1080 ti added to the loop and a D5 pump/res combo. On a 280mm slim Alphacool ST30 and an EK EX 120mm, its up to 4.7GHz and .04 more volts and its only in the high 60s/low 70s under those same stress tools. While I do have a radiator between my CPU and my GPU (pump - CPU- 280mm slim rad - GPU - 120mm XE rad - res), I don't have a whole lot more radiator than you and I'm highly OC'd on a hot CPU. Point is, at stock speeds you really shouldn't be hitting those temps.

IF your CPU block is making good contact, I've gotta think at this point there's one of two things causing this: (1) You don't have enough flow in your loop. Could be pump power, could be bad bends in the tubing, could be having so many 90 degree points in the loop. (2) Maybe the fans aren't ramping up when the CPU gets hot to try to help compensate? Its one thing to have enough surface area in radiators, its another to actually displace the heat off them with air flow. Your fans are strong enough, but are they running strong when heat picks up?

I'd still love to see your CPU be first in the loop right out of the res. I guess in your place I'd get straight to grabbing some soft tubing and make the loop order Pump -> CPU -> GPU -> Radiator -> Res. Also verify the fans are running at a good RPM (a good 1400-1500 or more when things are heating up). With this basic setup, at stock clock on that CPU, if you're hitting upwards of 90c I've got to imagine you've got a bad CPU water block or a bad pump. The fact that running just GPU benchmarks is making your CPU temp climb to the 70s when the CPU is idle concerns me about your flow rate. That said, if you're using something like Furmark to bench your GPU, that app will ramp up temps on one of your cores when running. That's normal. If your whole CPU is going mid 70s though its not normal.

At some point, in order to rule out EVERYTHING in your loop in regards to CPU temps, you're gonna have to put a "standard" cooler on your CPU and bench it. Whether that be a decent air cooler or some AIO liquid cooler (even a half-installed H75) you may want to segregate your CPU from the loop all together to try to get a better baseline on the CPU.
 
Upvote 0 Downvote
At this point I'm really stumped since the gpu seems to be cooled fine and the cpu is so hot. With everything flowing you'd think if it was purely a radiator issue the cpu would be slightly warmer as well as the gpu. Rather than a gpu with good temps and the cpu with poor temps on the same loop. Just speculating. The cost isn't wonderful but having individual loops would be ideal, you're right. At least it would help when troubleshooting.

Hopefully it can get sorted, the tubing looks pretty nice the way it's all set up.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom