Too hot on water. Why?

[reppel]

Hi

Running a 4930k on a asus rampage iv gene.
At 100mhz*45 with 1.4V
[img=http://s23.postimg.org/pc6fr4oif/temps2.jpg][/img]


Then in idle after a couple minutes



Im running a swiftech h220. cpu fan is the pump, alt cpu fan is the radiator fans.
Weird also that as soon as i start the burn in test, the temps jump like 20 degrees immediately, together with the pump and fans? It was around 70ish and then jumped to 90-95.

Might it be bad contact with the cpu or is this normal, i´m kinda sure i did it right.

And why is my asus ai suite showing different temps`?

 

[MT Alex]

It seems to me that the rad fans should ramp up to a higher rpm, but I also suspect you need to reseat your block. I'm not sure what your ambient temperatures are, but even your idle temps seem high. What was your mounting method?

 

[reppel]

Not sure what you mean by mounting method. I put the cpu, IC Diamond thermal paste, sit the block and screw in the screws. Here´s a couple pics:


I think the fans run at max when burntesting.
22deg ambient btw

 

[MT Alex]

By mounting method, I was most interested in how you applied your thermal paste. I know, from what others have mentioned in another thread, that IC Diamond can be problematic to apply properly.

I'd disconnect your pwm leads and run the pump and fans flat out and see how it works just as a control. I don't, however, see things getting better without a remount.

 

[the54thvoid]

Yeah, I'm at 1.39/1.4v at 4.4Ghz on a 3930k an my temps top out in the high 60's on prime after 30 mins (without aggressive fans speeds - though i'm on a 2x360 rad loop with gpu).

This is a schoolboy error check - was there a plastic film on the base of the cpu cooler? Did you remove it?

If your pump is working and the fans are on, it has to be a cooler mounting problem, surely.

Also, my Asus AI suite records about 40 ish degrees when core temp records those 60's figures. Nothing to worry about - just diff measurements.

 

[reppel]

thx, i guess it is a mounting problem then. But answering the mounting question, imagine a dice, with the number 5. I did like that: a 5mm circle in the middle and 4x 2mm between the middle one and the corners. I read somewhere, but wasn't sure this was the correct method, i was a bit afraid though because a lga2011 processor is quite big.

oh, and no plastic in the middle

 

[jcgeny]

the lga 2011 is very hard to cool , your cpu , like my 3820 , consumes 130w ...
your wblock is not enough , there is a post showing 4 of them and the few of liquid they have [may be someone knows link ?]

you need bigger radiators .
i used a case i made for a sr-2/x to cool my sabertooth with a koolance kit + 2 x hd5970 + gtx570 + apogee for 3820
it runs at 25 - 45 all day

http://postimg.org/gallery/54hlnneu/

in a year or two , i ll update cpu with a Intel Xeon E5-2697 v2 (2.7G,130W,L3:30M,12C,HT)

 

[reppel]

but 90 degrees?

Also, is it supposed to take 1-2 seconds, after i start the burnin test, to go from 35 to 65 degrees and vice versa when i stop the test?

 

[MT Alex]

thx, i guess it is a mounting problem then. But answering the mounting question, imagine a dice, with the number 5. I did like that: a 5mm circle in the middle and 4x 2mm between the middle one and the corners. I read somewhere, but wasn't sure this was the correct method, i was a bit afraid though because a lga2011 processor is quite big.

oh, and no plastic in the middle

Sounds like you just have too much paste, and an h220 will cool a 2011 socket CPU much better than the results you are getting.

 

[reppel]

Sounds like you just have too much paste, and an h220 will cool a 2011 socket CPU much better than the results you are getting.

So, I remounted.
Here's how the block and cpu were after separating them:



I'm guessing it was to few paste? Or like this paste sucks to spread?
Here's how i spreaded the paste now, do you think it is good? (Of course I put way more now)

 

[jcgeny]

90 *c is the temperature where cpu starts to kill itself...
at http://ark.intel.com/products/77780/intel-core-i7-4930k-processor-12m-cache-up-to-3_90-ghz
they say only one temp 66.8

i advice you update your cooling system then check the cpu and mainboard : and may be change them

new gfx are downclocking if temp is too high , soon cpu will do that too

n.b. try to have paste on all the cpu face instead of cross

 

[reppel]

here's a test with the repasting:


There's no way it's the cooling system, unless something's wrong with the pump. I saw like a million (maybe 5) reviews where it had delta 40-50 on overclocked 3930k, like this one http://www.xbitlabs.com/articles/coolers/display/swiftech-h220_5.html#sect0

The weird thing is, as soon as I start the burnin test, it jumps to 60-70 and it stays around that. Then, a couple minutes in, out of nowhere, it jumps to high 80s

 

[jcgeny]

does your pc is looking like their :

• Graphics card: AMD Radeon HD 7770 1 GB GDDR5 128 bit, 1000/4500 MHz with Deep Cool V4000 passive copper heatsink;
• Drive for programs and games: Western Digital VelociRaptor (300GB, SATA-II, 10000 RPM, 16MB cache, NCQ) inside Scythe Quiet Drive 3.5” HDD silencer and cooler
• System case: Antec Twelve Hundred (front panel: three Noiseblocker NB-Multiframe S-Series MF12-S2 fans at 1020 RPM; back panel: two Noiseblocker NB-BlackSilent PRO PL-1 fans at 1020 RPM; top panel: standard 200 mm fan at 400 RPM);

they have plenty of fans and a mid-level vga .
plus you can see the radiator position they use....

clear that hydroxxx-cooler is for others cpu like i7-4770k @ 80 watt / not i7 @ 130-150 watt [not overclocked]

 

[Nordic]

So, I remounted.
Here's how the block and cpu were after separating them:



I'm guessing it was to few paste? Or like this paste sucks to spread?
Here's how i spreaded the paste now, do you think it is good? (Of course I put way more now)

When it comes to thermal paste less is more. You want as thin a layer spread over the whole cpu as possible. There are two methods. One similar to what you are doing is put a rice grain amount of paste. The other is to spread the paste out yourself as evenly as possible.

 

[Kaynar]

Just some input here,

I have a 4930k with an Asus rog mATX mobo, at STOCK clocks and volts my max temps are 67c, while in gaming its around 50+.
My loop is made of two HWLabs SR1 rads (one 280mm, one 240mm) with Noiseblocker 140mm and Corsair SP120 fans all running around 1000rpm. I am cooling the CPU as well as a GTX Titan both with Heatkiller blocks. The D5 Vario pump is running at 3000rpm since 5000rpm makes barely 2c difference at stock cpu settings. My termal paste is latest Coolaboratory (don't remember which version) but before I was using IC Diamond.

Considering the above, I think its normal you get 90c+ at 1.4v with your kit although I am sure you could shave more than 5c if you properly apply thermal paste, and yeah that IC Diamond is just a pain to get right, I always put a pea sized dot in the middle of the CPU but sometimes it turned out my application was bad. My ambient temp is around 20c as well and I have a well-vented CaseLabs case.

EDIT: I think on the last picture where you put the IC Diamond in a cross you just put a bit too much of it....

EDIT2: as for the 66.8c mentioned above, that is a Tcase, meaning the suggested maximum CASE (not CPU) temps (for those that have bad case ventilation) in order to prevent CPU overheating when using Intel's crap stock cooler. Because you might have 20c ambient room but boiling temps inside your case.

 

[reppel]

thx guys, that helps.
i guess it's not so unusual then.
i tried at lower voltages and it got me think that 1.4 warms way too much.
I also thought about bubbles in the loop, it's new, not sure how to get them out though.
Lastly, maybe I have a bad cpu, I mean, bad in a way that if I removed the IHS and reapplied t-paste to it I might get quite many degrees out. Does everyone does this part, everyone that goes to water. I'm just a bit afraid of destroying my cpu doing that.

 

[Law-II]

Hi

i tried at lower voltages and it got me think that 1.4 warms way too much.

Six core CPU @ 1.4v H220 is not enough

I also thought about bubbles in the loop, it's new, not sure how to get them out though.

Edit: My Bad H220 is not a sealed loop [See sneekypeet's post below]
Information about bleading air from the loop can be found here

Lastly, maybe I have a bad cpu, I mean, bad in a way that if I removed the IHS and reapplied t-paste to it I might get quite many degrees out. Does everyone does this part, everyone that goes to water. I'm just a bit afraid of destroying my cpu doing that.

Skt 2011 Ivy-Bridge E is hard soldered and IHS cannot be removed Source [you may be confusing this CPU with skt 1155]

Side note: look here thermal past spread and here for cpu thermal paste application

atb

Law-II

 

[sneekypeet]

H220 is not a sealed loop and will have bubbles in the loop to some extent. The trick is just to make sure it isn't in the head unit. I had an issue getting air out of the pump when I built my ladies rig. Wasn't a big deal, just had to rotate the chassis a bit.

A custom loop isn't going to do much better than what he has, its just more work and the loss of time swapping it all out.

 

[Kaynar]

Yeah the H220 is not sealed, the user can disassemble it to extend it, the rad also has a tiny res. Also the IHS is soldered on that cpu so yeah, you can't remove it. I will also agree that an H220 is not enough for a 1.4v cpu with such high TDP. These "AIO" units usually can cope with 1.3v max.

The thing is, if you are running 4.5ghz at 1.4v then your chip is kind of below average, that's a lot of voltage for 4.5ghz. I guess you have experimented with the overclock to deduce that you NEED that much in order to keep 4.5ghz stable, otherwise you should start to experiment now. Unfortunately I haven't OC'ed my 4930k yet since I wanted to keep it to stock values for some time before OC'ing. I didn't have much time for games anyway...

 

[reppel]

I started pretty much at 1.4v 4.5ghz as base, with the objective of going up to see where it would take me. I can have it stable at lower voltages.
The H220 should be pretty good, i think. I read a ton of reviews before i picked it. Like this one: http://www.xbitlabs.com/articles/coolers/display/swiftech-h220_5.html
If they're able to run it at 1.485 volts / 4.625 GHz at max temps of 73 degrees (ambient 22-25), I was expecting a not much worse result, at least at 1.4v.
Thought there must be something wrong with it then. The paste shouldn't be the problem, i think, it can't be causing such bad values like i did it, i mean, that cross doesn't look too bad

 

[Nordic]

The paste shouldn't be the problem, i think, it can't be causing such bad values like i did it, i mean, that cross doesn't look too bad

It might be a little too much. My experience is that too much paste only hurts cpu temps by maybe 5c, but I also never tested on a overclocked 6core chip. Maybe your fans arn't getting enough airflow. Maybe try reversing it so the h220 pulls cool air from outside into the case. Something easier than that would be to have that cpu area exhaust fan be an intake fan. I bet it is airflow hurting your temps if anything.

I am going to leave this here again just in case you missed it.

Hi
look here thermal past spread and here for cpu thermal paste application

 

[micropage7]

I started pretty much at 1.4v 4.5ghz as base, with the objective of going up to see where it would take me. I can have it stable at lower voltages.
The H220 should be pretty good, i think. I read a ton of reviews before i picked it. Like this one: http://www.xbitlabs.com/articles/coolers/display/swiftech-h220_5.html
If they're able to run it at 1.485 volts / 4.625 GHz at max temps of 73 degrees (ambient 22-25), I was expecting a not much worse result, at least at 1.4v.
Thought there must be something wrong with it then. The paste shouldn't be the problem, i think, it can't be causing such bad values like i did it, i mean, that cross doesn't look too bad

yeah, paste could affect so the little is better
but i just think it could be from your rad, pump or the liquid
have you ever touched your rad, just to make sure your sensor read it right
btw have you tried by using your stock hsf? just to check the temp as comparison

i just think if your cooler in warranty period, better rma it

 

[reppel]

thx for the replies. I put more paste because the first time it was too little. I might have put a bit too much, but i suppose it would spread to the sides. Anyhow, i got the same temps with too much or too few, same behavior.
But i suppose that even with a bit too much or too few paste there shouldn't toooo much temperature rise.
I've tried moving the case around, trying to remove bubbles, i don't notice any weird sounds with the pump.
the fans are maxed out pushing air out, and i have the case opened just in case.
I've got temperature differences of 10-15 degrees between cores, I even suspect something wrong with the reads as I get Core#3 in idle with temps lower than the ambient temperature, at 16-18degrees.
I'm also not sure that the problem is just having one radiator, as I'm just testing it for 2-3 minutes, not giving it time to stabilize at its hottest. Unfortunately i don't have another heatsink to test, and the aio system is completely new, have just packed it out and installed it, twice now.

 

[Kaynar]

Well the reviewer of Xbitlabs is testing a 3960X. might be the same TDP but not the same processor for sure.

Uhhh about what you are saying about those temp readings, that sound really weird

Can you tell us what are you max load temps when your CPU is at stock speed?

 

[reppel]

I've now removed the block again, used the recommended thermal paste that came with swiftech, shaken the block and rad in case of air bubbles, repositioned the block with the tubing upwards, repositioned the radiator with the reservoir opening upwards, same problem overall, same temperatures, although strangely less stable. I was running 4.6ghz at 1.3 which now got unstable.
I don't know whether the problem is with the swiftech thing or the cpu.

Running at stock speed, case open, fans and pump at max speed, prime95 during 5 minutes: