H115i not cooling as expected

I need some help with a H115i that has never really worked as expected.

I purchased it in august 2016 initially to cool a 6700k running on an Asrock Z170M Extreme4 motherboard. The CPU was overclocked to 4.5 ghz at 1.28V and run extremely hot, running prime95 would see temperatures in the low 90s, also the temperature wouldn't rise steadily, it would suddenly spike and then fluctuate a lot between high 60s and low 90s.

I tried re-seating it, made sure the bracket was in the correct orientation and even tried using washers to force it to make better contact with the CPU, nothing made much difference.

In the end I though that was just how the product worked and lived with it.

However, now i have just upgraded my system, my old 6700k is being cooled with a Swiftech H240-X and is experiencing none of the issues the H115i did. Temperatures build up slowly, and then hold at a consistent level, even with the voltage bumped to 1.38 temperature never go above me mid 80s while running prime95. So obviously there was a problem with the H115i.

So now I am trying to use the H115i to cool an 8700k on an Asus ROG Strix Z370-G motherboard, I am experiencing the exact same problems I had while using it with the 6700k, although they are worse, the temperature spikes to 100 almost instantly and remains there.

The water temperature is increasing so there is seeming contact with the CPU and the pump sounds like it is working, at least i can hear it very quietly doing something. I am at a loss as what the problem could be and would appreciate any help.

Erratic temperature spikes in a program that provides constant load (not all stress tests do) usually indicates a contact problem. Make sure you have the right standoffs and the backplate that came with the H115i. Some Asetek backplates are univeral, but not all.

What kind of coolant temp change do you while under load? This doesn't have to be a full bore test and I would recommend you drop Prime 95 until you figure out what is going on. The Linpack test within OCCT will be steady load but less extreme on the CPU. 8700K also has a few tricks to watch out for on Asus. What is your measured peak Vcore? Too many people think they have it set, but the reality is something different.

@c-attack I'm fairly certain 8700k is set to run at 1.22v @4.7ghz

I've set load line calibration to the lowest possible setting and I think I've turned off all the Asus auto overclock/voltage settings but i'm in work right now so i'll check when i get home, along with making sure I am using the right standoffs. However I am fairly certain I am, the thermal compound is being spread by the contact that is being made.

Just to clarify, when people are overclocking the 8700k, they are overclocking all cores, so in my example all cores should be running at 4.7ghz ?

Erratic temperature spikes in a program that provides constant load (not all stress tests do) usually indicates a contact problem. Make sure you have the right standoffs and the backplate that came with the H115i. Some Asetek backplates are univeral, but not all.

 

What kind of coolant temp change do you while under load? This doesn't have to be a full bore test and I would recommend you drop Prime 95 until you figure out what is going on. The Linpack test within OCCT will be steady load but less extreme on the CPU. 8700K also has a few tricks to watch out for on Asus. What is your measured peak Vcore? Too many people think they have it set, but the reality is something different.

Typically, yes, all cores are overclocked.

And check the actual vCore while under load. Just because you set it that way in the BIOS doesn't mean that's what's actually being delivered.

I turned off all automatic settings in the BIOS and set the CPU to a low frequency of 3.7ghz with a voltage of 1.1v

The H115i should easily be able to keep the CPU cool while running prime95, however it failed. The temperature was going between 86 - 91C

Image linking doesn't seem to work so you can see the results here https://imgur.com/a/UiqQh

My Idle temps are 40C with the occasional spike up to 50C. the H115i idle temp is 30-31C

I am using the correct standoffs, the backplace is correctly oriented, the thermal compound is being spread which would indicate that contact with the CPU is being made and i've even tried to using plastic washers to increase pressure on the CPU.

What else can it be, by all accounts the H115i is functioning properly, is it just a garbage product ?

Check the actual output Vcore voltage first. This is so often the problem on Kaby/Coffee Lake CPUs. If you work for Intel, then 4.3GHz is the maximum turbo boost when all 6 cores are active. The single core limit is 4.7 and that is where the boards tend to get into trouble. Even with Multi-Core Enhancement off, Asus boards will try to run 6 cores at 4.7 if set to "all core" on the frequency, rather than "per core". This is part of the problem, but it can be remedied with a few tweaks.

Voltage first, coolant temp second. How much does it go up under a mild sustained load, like Intel XTU? That is a cupcake stress test, so you should be able to run it, even with a problem. That might give us more insight into what's going on.Even in perfect conditions, Prime may be too much for optimized settings. It is too hot to use as a diagnostic tool.

I turned off all automatic settings in the BIOS and set the CPU to a low frequency of 3.7ghz with a voltage of 1.1v

 

The H115i should easily be able to keep the CPU cool while running prime95, however it failed. The temperature was going between 86 - 91C

 

Image linking doesn't seem to work so you can see the results here https://imgur.com/a/UiqQh

 

My Idle temps are 40C with the occasional spike up to 50C. the H115i idle temp is 30-31C

 

I am using the correct standoffs, the backplace is correctly oriented, the thermal compound is being spread which would indicate that contact with the CPU is being made and i've even tried to using plastic washers to increase pressure on the CPU.

 

What else can it be, by all accounts the H115i is functioning properly, is it just a garbage product ?

So that was a 15C coolant temp swing? Over what time period? I guess it doesn't matter. 15C is too much and my guess is you have a partial blockage or some measure of flow restriction. Start a Support Ticket from the menu at the top of the page. Make sure you include the coolant change delta. Incidentally, what does Link report as the max Vcore during that test?