CPU in Custom Loop bad temperatures

Toggle sidebar Toggle sidebar
P

Peter Pimmelmann

Reputable
Feb 3, 2016
12
0
4,510
I have the problem, that my cpu temps are not as good as they should be.

i have a 4790k oc at4.8 Ghz at 1.36 Volts, i use a ek supremacy evo and a 420mm radiator.
i also have a 980 ti with an aquacomputer waterblock in the loop, which never exceeds 44 degrees celisus.

the cpu temp are at 50-60 while gaming, idle at 45-40 and under full load 75 up to 100 after half an hour stress test.
i use arctic mx-4 thermal paste. it is applied correctly and the cpublock is mounted correctly. i dont know why the tams are so bad on the cpu, considering the radiator barely even gets hot during a stresstest or normal gaming.

i looked everywhere but couldnt find a solution. please help if you can. thanks.
 
Solution
1LiquidPC
I've heard of folks on here that may have that issue, but its internal in the CPU cores. The only way I can think of is process of elimination by what you've already tried. If multiple mounts/CPU coolers still give high readings its either bad IHS or TIM, or temp sensors.
Sort by date Sort by votes
Leaps-from-Shadows

Leaps-from-Shadows

Splendid
Jan 10, 2011
2,941
2
21,460
The "full load" -- is that Prime95 v28.7? That's an excessive stress test according to many ... but even so it shouldn't bring your CPU to anywhere near 100°C.

Are the fans controlled by a fan profile, or are they running full speed all of the time? If they have a fan profile, can you increase the speed at higher temps?

I have an i5-6600K at 4600MHz at 1.375v and an EVGA 980 Ti Hydro Copper at 1483MHz core 8000MHz memory in a custom loop with an EK-Supremacy EVO CPU block. Mine has two 360 radiators with the fans running at 100% at all times (they are low-speed fans though, so never get noisy). The GPU never exceeds 50°C and the CPU never exceeds 85°C even when running Prime95 v28.7 Small FFT test and a GPU-blasting program at the same time.
 
Upvote 0 Downvote
P

Peter Pimmelmann

Reputable
Feb 3, 2016
12
0
4,510
I tried it at full speed and i didn´t see a change in the cpu temps.

when i do a gpu stress test, it stays under 50 degrees and the radiator gets considerably warmer, but when i do one with the cpu the radiator gets barely warmer and the temp are still over 90 after 20 minutes.

but are these cpu temp normal? because some people have like 60 deg. max on air with a similar tdp
 
Upvote 0 Downvote
P

Peter Pimmelmann

Reputable
Feb 3, 2016
12
0
4,510
i am sure it´s mounted correctly, i even reseated it yesterday. i applied a 2 grains of rice sized thermal paste blob and didnt manually spread it. the reseating made no difference.

restriction-wise the loop barely even has restrictions. i had a filter and another cpu block in it at first, but now i have this high-flow ek blcok and no filter. my loop has very few restrictions now. i use a laing ddc 1t pump at full speed, and you can even see a little vortex in the reservoir because the liquid flows so quickly. it doesnt pick up water through it, so its not a problem.

also i purchased that ek cpu block about a week ago, so it doesnt have gunk in it.
 
Upvote 0 Downvote
Leaps-from-Shadows

Leaps-from-Shadows

Splendid
Jan 10, 2011
2,941
2
21,460
Those temps would be normal if you were using the stock cooler. Even the worst AIO cooler ever made would keep your CPU under those temps.

And I know that CPU block is good because it's currently keeping my overclocked and over-volted CPU cool. I've got a Liang DDC-1T Plus 3.2 pump, two XSPC EX360 Crossflow radiators, and the Hydro Copper GPU block on my EVGA 980 Ti. I went overboard on the radiators because I could, and it looks cool (pun intended) in my case.

You have a GPU and CPU in your loop. If you stress the GPU the temp goes up to a reasonable level and the radiator gets much warmer. That's good. But if you stress the CPU the temp rises to nearly dangerous levels and the radiator barely gets warmer. That's bad.

That's definitely pointing to a failure with the CPU block. Either it's not mounted correctly, or there's a problem within the CPU block itself. When I got mine, it had instructions with it and they stated that the default restrictor plate wasn't the optimal one. I left it with the default one anyway, because I'm not concerned with a one or two degree difference. Did you mess with the restrictor plates?
 
Upvote 0 Downvote
P

Peter Pimmelmann

Reputable
Feb 3, 2016
12
0
4,510
i swapped in the recommended one. it was really easy, just unscrew the 4 heatspreader screws, take the one plate out and put the thinner one in. but i had another cpu-block from phobya before, the uc-1 extreme, and the temps were much worse.
 
Upvote 0 Downvote
Leaps-from-Shadows

Leaps-from-Shadows

Splendid
Jan 10, 2011
2,941
2
21,460
"Much worse"!? You can't get much worse than what you have now.

And by the tests I've seen (Source - 4770K, 4.7GHz, 1.4v), the UC-1 Extreme should have been within a degree or two of the EK-Supremacy EVO at full load.

The heat is getting from the GPU to the liquid, as evidenced by the radiator getting warm when you stress the GPU.

The rest of the loop is functioning, otherwise the radiator wouldn't be affected by the GPU's heat.

But the heat isn't getting from the CPU to the liquid. If the temps are this bad with two different CPU blocks, maybe something is interfering with the mounting of the blocks.
 
Upvote 0 Downvote
P

Peter Pimmelmann

Reputable
Feb 3, 2016
12
0
4,510
well before it was like 20 degrees hotter, which was really disappointing. but i dont see anything interfere with the mounting of the ek block. it makes even contact everywhere on the cpu, i have all screws tightened as far as they go (without tools of course).
could it be the connection from the cpu chip to the cpu heatspreader?
 
Upvote 0 Downvote
1LiquidPC

1LiquidPC

Splendid
Oct 11, 2014
5,314
83
31,590
Can you post pics of your loop? Something is definitely wrong here. Is there some kind of backplate that's used behind your motherboard to attach to your block? When you remounted the cooler, what did the compound look like on the block before you wiped it clean? Did it appear to have even contact w/ your CPU?
 
Upvote 0 Downvote
P

Peter Pimmelmann

Reputable
Feb 3, 2016
12
0
4,510
the thermal paste was evenly applied and the block appeared to be seated correctly. yes, there is a backplate, there is a bit of rubber at the backplate that stands up bc there are a few pins, but i already checked and there was nothing interfering with the mounting of the block. its mounted evenly.


Pics: http://imgur.com/a/HaohE
 
Upvote 0 Downvote
Leaps-from-Shadows

Leaps-from-Shadows

Splendid
Jan 10, 2011
2,941
2
21,460
If it's not the mounting, then something else is wrong with the CPU block. There are two other things about the block that could be making a difference -- liquid flow direction and the orientation of the copper block plate.

The liquid should be going in the port nearest the center of the block and coming out the port nearest the edge of the block.

The copper block plate should be oriented so the channels are perpendicular to the hole in the jet/restrictor plate and the insert.
 
Upvote 0 Downvote
1LiquidPC

1LiquidPC

Splendid
Oct 11, 2014
5,314
83
31,590
I'm starting to think it may be your particular CPU. If your mounting and flow are good, and your radiator barely gets warmer on CPU only load, it could be a heat transfer problem. The contact between the TIM between your CPU die and inside of heatspreader may be mediocre. Was your CPU always hot, even stock? Also possible its a faulty temp sensor in your chip.
 
Upvote 0 Downvote
Leaps-from-Shadows

Leaps-from-Shadows

Splendid
Jan 10, 2011
2,941
2
21,460
Yeah, I'm thinking this is just "normal" for that chip. Assuming everything else is correct, there is nothing else to check.

The idle temps 10-15°C higher than they should be.
The gaming temps are normal.
The high-load temps are 20°C higher than they should be.

I knew the previous generation had bad TIM between the die and IHS, but Devil's Canyon was supposed to fix that. I didn't think temps would be that bad.

Oh well. If everything that can be fixed has been fixed, then we have to assume it's supposed to be that way. I don't like it, and I would never accept it if it was my system, but whatever...
 
Upvote 0 Downvote
1LiquidPC

1LiquidPC

Splendid
Oct 11, 2014
5,314
83
31,590
Yeah, not sure what else to recommend by this point. Your correct about Devil's Canyon, it was supposed to "fix" the heat problems from Haswell generation using better TIM. I also would not accept those temps either. I would try running at stock again to check temps, even install a temporary air cooler to check. If the results are the similar with these higher than normal idle/load temps, I'd try to RMA the CPU.
 
Upvote 0 Downvote
P

Peter Pimmelmann

Reputable
Feb 3, 2016
12
0
4,510
i dont recall the exact temperatures with the stock heatsink, but i konw i got like a high temperature warning once, although i didnt stress the cpu.

should i delid it? i dont really want to, but these temps are bugging me a lot.

Edit: i just did a stresstest at stock speeds and the temps hovered at just under 70, although i noticed that core #3 was almost 10 degrees colder (59-60) than cores #0 - #2 (67-69)
i didnt bother doing a test with the stock cooler, it´s just to much hassle
 
Upvote 0 Downvote
1LiquidPC

1LiquidPC

Splendid
Oct 11, 2014
5,314
83
31,590
I've heard of folks on here that may have that issue, but its internal in the CPU cores. The only way I can think of is process of elimination by what you've already tried. If multiple mounts/CPU coolers still give high readings its either bad IHS or TIM, or temp sensors.
 
Upvote 0 Downvote
Solution
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom