Jump to content
Corsair Community

bad temperatures h115i


danteafk
 Share

Recommended Posts

I got two of the h115i here, both are giving me those temperatures. I have removed and reseated the heatsink 100 times, reapplied TIM on difference ways, always the same result. profile is on performance, pump running on almost 3k.

 

I have excellent temperatures with nocuta dh14 and dh15, but I would like to use the h115 because of the aesthetics.

 

http://i.imgur.com/oaPnZjq.png

 

 

 

temps with notua d15

http://i.imgur.com/1EbUWld.png

Edited by danteafk
Link to comment
Share on other sites

The only thing that I can think of is something on the motherboard stopping the pump block from making full contact with the cpu. When you removed the cpu to reseat it, was the TIM evenly spread? At first I thought that there was a problem with the cpu (31 degrees between cores is ridiculous) but then I saw that it was only 12 degrees with the Noctua cooler.

The temp differences between the Noctua and the H115 are more or less as I would expect, except for the 92 degrees on core 0.

 

As you are getting the same result on both H115s, that suggests that either something is blocking them both or they have the same, very odd, problem.

 

One more thing has just occurred to me, I'm not that familiar with the h115 mounting system, but is it possible that one of the backplate mounting bolts or the pump screws is faulty/incorrect? Do these coolers still have two different lengths of bolts to mount the backplate? I presume that you used the same mountings for both coolers.

Edited by snapper69
Link to comment
Share on other sites

Check if you used the right screws to mount the backplate. There are screws for socket 2011 (one end is shorter) and 115x screws. Check the manual for using screws matching your socket. What is your fan speed?
Link to comment
Share on other sites

the fan is running at max, pump is at 3k, the radiator is barely getting hot.

 

I'm using the screws for the 115x socket, not the 2011 ones, the ones I'm using are equal long from both sides the 2011 ones are shorter from one site.

 

I tripple checked the manual to make sure everything is right, but I suspect too that the heatsink is not creating evenly pressure on the CPU? What can I do?

 

one thing to mention, even if I mount he bracket with the bits until I cannot screw it any longer, it's still a bit wonky and lose, it only gets fix after I put on the heatsink and mount that.

Link to comment
Share on other sites

the radiator is barely getting hot.
AS s Ray suggested, I would like to see the H115i Temp in Link (water temperature) and an approximate room temperature for comparison, but this goes along with what I suspect below -- you have a contact issue between cold plate and CPU and/or a TIM mess. Some of the heat will not transfer from the CPU to water when this is the case.

 

one thing to mention, even if I mount he bracket with the bits until I cannot screw it any longer, it's still a bit wonky and lose, it only gets fix after I put on the heatsink and mount that.

 

That part is normal and by design. However, I do believe there is some sort of contact issue as suggested above. It is normal to see some variance between individual core temperatures, but in your air tower data the grouping is fairly tight. In the H115i data, Core 1 is 8C over Core 4 at the minimum and 30+ at maximum. That is not a normal difference and any CPU defect would not change between coolers.

 

I'll admit, the backplate can be a little fiddly and I love my 2011-3 where I can screw it down in 5 seconds and be sure it's right. However, perhaps the TIM is the first thing to check. If you have been struggling to mount it and the paste keeps getting prematurely smeared prior to lockdown, that could account for the differences.

 

If not, then it becomes a little more difficult. Make sure the block is secure when fully tightened. If there is still a little bit of play or your CPU temps change when you press down on it, then you might be a candidate for some washers to help things along. You are not supposed to need them, but there are a few Asus boards that seem to be problematic in this regard. I am certainly open to other suggestions, but these are the two most common problems.

Link to comment
Share on other sites

i've removed the block again, cleaned everything, reapplied TIM. Now i've put on the washers that came with the package that are supposed for the radiator/fan>case mounting.

 

temps are better, still not optimal, but now I'm using CPU-z stress test because prime95 seems to kill my cpu with that heatsink

 

first picture is idle, second picture is load

 

idlecxoev.png loadnaq1h.png

 

the radiator is still barely hot, I suspect sill contact issues?

Edited by danteafk
Link to comment
Share on other sites

Well, the core temp grouping is better and more reflective of your past results. I'll have to fire up CPU-Z and see if I can make sense out of the load temps, however the Link shot does indicate the cooler is behaving as expected and you don't have a functional problem with the water flow.

 

The idle temps are high, but I suspect that is a result of the spiky nature of Windows 10, particularly when it comes to image management. There is a 14% load at that instant and it is likely that is responsible for the elevated temperatures. At 31.5C H115i temp (water temperature), your cores should bounce between 31-34 at idle, when the frequency steps down. This is assuming you are using C-states/EIST. If not, it will be a few degrees higher.

 

EDIT: I ran the CPU-Z multi-threaded test for all 12 of my cores. Unfortunately, we have different processors and I am not sure a 5930K 4.5@1.275v can be directly compared. However, my individual core temperatures were very steady in the 55-57C range with not much more than +-1 fluctuation at a given point during the load. If your individual core temperatures were considerably more deviant during the load phase, that might suggest the contact still isn't quite right.

Edited by c-attack
Link to comment
Share on other sites

  • 2 weeks later...

So I have put another pair of washers on the bracket, there are now 2 washers on each and applied LM on the DIE (not delided).

 

I assume the contact is stil not OK... should the water temp of the 115i be much higher on load or is that normal?

 

capturebpsjv.jpg

Link to comment
Share on other sites

Dont use Prime95 for a 4790k. It will over volt by loads and create lots on unwanted heat. Stick to either XTU or Aida64 for haswell chips.

What vcore are you running to get 4.6ghz ?

Also what other settings have you changed within the bios when overclocking ?

 

Not sure what thickness washers you are using, but I personally used some 10mm thick rubber gromits (From my car supplies in the garage) to space the backplate out slightly, the washers have some 'squish' in them when tightened up. Without the washers I could not get good contact between the CPU and the pump header.

 

I've just done a quick comparison for you (Im on a different motherboard and cooler though, Gigabyte Z97x and a H100i) I would expect the H115i to perform better than mine by a deg or two.

 

http://i.imgur.com/EtZQd9M.jpg

 

I suspect this is either a bad contact issue, or bios setting related - I notice on your pc specs its state you're running the 4790k @ 4.8. Have you previously been running that speed, and not reverted some of the settings back to stock etc ?

Edited by Kippsy
Link to comment
Share on other sites

Hmm interesting.

 

I'm using XTU yes. Vcore is 1.267 for 4.6Ghz. No changes in bios. 4.8 is only for benchmarking.

 

What are you idling temps? Can you read out the water temperature with the corsair link app?

 

I might get some rubber gromits too and test that out.

Link to comment
Share on other sites

Hmm interesting.

 

I'm using XTU yes. Vcore is 1.267 for 4.6Ghz. No changes in bios. 4.8 is only for benchmarking.

 

What are you idling temps? Can you read out the water temperature with the corsair link app?

 

I might get some rubber gromits too and test that out.

 

Current idle temps are 25 - 28 deg on each core

 

I dont have the Corsair link installed (As its utter garbage !!!) If you look on my screen shot, and to the left of the XTU window is a series of sensor readings (Aida64 OSD panal), towards the bottom in purple you will see the H100i temp readings etc (These are taken from the same 'place' CL4 reads from)

 

My ambient temps at the moment are roughly 20 - 25 deg. The H100i temp was 24 deg on cold boot, and sitting at 27 deg now whilst typing this (My CPU core voltage is 1.260 for 4.6 so very little difference between the 2)

 

If you do go for the rubber gromit type like I used, make sure they arn't the type that melt with hot temps !!! I would think that would make a mess of the motherboard.

 

 

EDIT - In response to your question about the water temps, Your water temp at the load where the same as mine, but the test where only run for less than a minute. The water temp will rise and fall slowly compared to CPU temps. So the longer the test runs, the hotter the coolant temperature will get

 

2nd EDIT - Quick idle screen shot for you - http://i.imgur.com/0tbWH43.jpg

Edited by Kippsy
Link to comment
Share on other sites

Remember idle temperature is significantly affected by room/case ambient temperature. Someone running a default voltages in a 30C room will have higher idle than another system with high Vcore in a 20C room.

 

In your XTU screen shot from post #9, it shows "2" active cores. The core temps suggest all 4 are active. Did you configure the test this way?

Link to comment
Share on other sites

  • 7 months later...
  • 6 months later...
 Share

×
×
  • Create New...