Cooling H100i temps higher than I expected! :-/

Afternoon,

Put me new-ish rig together on weekend, its an i7 4790K, Asus Maximus Gene VII, G-skill 2400mhz Ram, Corsair 650w PSU, Sapphire 290X GPU, Corsair H100i cooler

So all put together and running, first wanted to get the RAM working at its spec'd speed before OC'ing the CPU so leaving all CPU voltages at auto (besides ram) I run IBT and AIDA64 and I was seeing 75-80 degrees on hottest core in IBT and 70 in AIDA and I thought damn that hot considering the H100i is supposed to be good!

I then done a mild overclock to 4.6ghz with a 1.23v core and in IBT was hitting 90-95 degrees and 80-85 in AIDA

I stopped stressing it straight away when I seen those temps, I removed the H100i and then used some IC Diamond TIM and temps have dropped a few degrees but to me they are still very hot!

I installed the Corsair LINK software and can see the pump is running fine etc so not sure what's going on, the radiator barely gets warm when stress testing and I expected it to be much warmer so wondering if I have a defective H100i?

Anyone shed any light on it or things that I can test?

 

Is the cooler seated tight enough? I did my nepton today and at first start up the temp went skyward. . Turns out I hasn't screwed it down tight enough onto the cpu... Rookie error!

Sent from Bittech Android app

 

Yeah done up by hand then nipped up tight with screwdriver, cant go any tighter without anything snapping lol

 

A mate of my put together a similar rig, he found that no matter how many times he reseated H100i it would not improve his temps.

Being the brave sod that he is, he delided and replaced the compound under the IHS. Running much cooler for him. He did loads of research which let him to the conclusion the TIM under the IHS is pants.

 

The latest gen i5/i7 do run hotter than previous generations from my personal experience.

 

Yeah I have heard that a lot about the Ivy's and didn't realise the Haswell's still following the same crappy TIM path

De-lidding just seems a scary route to take just in case it gets damaged but might have a crack at it next year if I feel brave enough!

Maybe its the actual CPU causing the temps then not the H100i

 

I am afraid you've just nailed it.

The sweet spot I found for my little devil is 4.5GHz with 1.13v and the hottest core stays below 68°C in IBT, but it's cooled with custom WC system including EK Supremacy. Just a 0.01v more and all temps go 3°C higher.

1.23v seems a lot for 4.6GHz, especially for a 5GHz capable cpu.

 

This is why my Haswell stays at a lowly 4Ghz

Too hot to mess about trying to match my older SB and Ivy speeds.

 

1.23v on haswell is high. If you have a decent chip sub 1.14 at 4.5ghz is not unheard off. I'd try finding out the minimum volts it will run on.

Auto is just plain bad in most circumstances.

 

Cheers guys, I've not started properly messing with voltages yet as I was shocked with the temps so looks like I need to drop the voltage so I get sub-80 temps and just see what clock speeds I can get from that then...

Anyone been increasing the BCLK on these or just fixing to 100 or 101?

 

After checking another thread I'd like to raise a question:
What software do you use to read the temps?

 

I normally use RealTemp or HW Monitor and I have used CoreTemp as well and they are all pretty much the same, the Asus Ai-Suite always seems to read around 10 degrees less which is a bit concerning!

 

HW Monitor all the way.

 

Last night I had a play with OC settings and seem to have it stable (pending a bit more stress testing) at 4.7ghz (well 4.69) at 1.19v and its just touching 90 degrees on the hottest core now under IBT and 80 degrees on AIDA64.

I tried ROG RealBench and it stayed around 65 degrees then I fired up BF4 with everything on ultra for a 15 min blast and it was around the 55 degree mark but BF4 only used around 25-30% of CPU so is never going to push it!

Getting there, would like to get a stable 4.8ghz with decent temps but will see, apparently this CPU will do 5ghz at 1.33v but I don't think I can cool it enough to stress test it!

 

I've replied to your PM Rob

Sorry but I had a funeral to attend yesterday

BIGTIME!! variety

In all honesty ditch the H100 mate first and foremost, in my experience the CoolIT rebrands truly suck and are prone to pump/flow issues.


EDIT -

That's the variety of back plate I use -





High tension



EDIT 2 -


A quick screen shot of the difference a high tension mount and full loop makes -





My 4690K under full load @ 1.327V



EDIT 3 -


And after 15 minutes full blown stress testing -

 

Is the backplate the right way around ? Going on a picture i found on the Corsair forums the cutouts should line up with the screws...



Maybe even with the screws done up tight the pump still isn't making good contact ? Some people have said they fit washers to the backing plate for a more secure fit.

 

To add onto what others have said:

I own a pretty good 4790k (voltage and thermal wise) and installed a h100i on Friday.

I found the back plate would not fit my mobo at all. Had around 2-3mm of slack on it when tightened right up. I had to double check the box to see if it actually fitted socket 1150 3 times because the standard back plate had such a bad fitting.

In the end I just reused my existing back plate from my ridiculously over sized air cooler and now it has made good contact.


Sounds like you have a hot chip. From the amount of chatter I've seen on this proc there appears to be a lot of them. Have you tried running this with the stock cooler cooler. I'd be willing to bet that this chip fails intel's tests @ intel's stock settings with the stock cooler.


For comparison here are some of my figures:
OC's:
4.4ghz @1.06v
4.5ghz @1.12v
4.6ghz @1.16v
4.7ghz @1.25v
4.8ghz @1.390 (not stable, darn't even try any more volts)

As you can see my hopes of 5ghz were killed. Both my chip and many others refuse to be stable past 4.7ghz no matter how much thermal headroom you have. I might be able to manage it with say 1.5v but I'm not willing to put that through a 22nm processor.


In terms of temps I've tested with IBT, AIDA64 and OCCT. Temps measured with CoreTemp, Corsair Link and HWMonitor.
@4.7ghz @1.25v
73-75c on 100% load
27-31c idles (27 on 0-1%, 31 on 2-3%)
Playing far cry 4 (only game played since installation) - 44-47c

I'm running dual sp120's as exhausts through rad in top of 350d. Both hooked up to fan controller at 50% permanent speed.

Hope this helps.

 

I also had this exact same issue with the H100i. I must have cleaned and reapplied paste 4 or 5 times and the temps were always high. I tried swapping fans and nothing worked.

In the end I took some pliers to the mounting frame around the pump and bent the 4 arms up a bit to increase the gap between the board and the screw heads. When I put it all back together it worked a treat and the temps dropped by 20 degrees.

 

Mounting brackets for the block are pathetic in fairness.

 

The mounting bracket isn't the best design in the world for sure but the slackness of it being free moving through the motherboard holes is by design and not a fault, soon as you install the waterblock the slackness it taken up so its not really an issue, and yes cheers Corky mine is put in right But still I dont think it does up as tight as I would like it to be!

I might have a go at bending the arms up a bit but in reality might just be the chip running hot from what people have said.

Im still playing with overclocks at the moment so will report back on findings with voltages/temps soon